Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Mentoring as work-related support : relationship with employee outcomes Schroeder, Klaus Gerhard 1988

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-UBC_1988_A1 S37.pdf [ 11.49MB ]
Metadata
JSON: 831-1.0098183.json
JSON-LD: 831-1.0098183-ld.json
RDF/XML (Pretty): 831-1.0098183-rdf.xml
RDF/JSON: 831-1.0098183-rdf.json
Turtle: 831-1.0098183-turtle.txt
N-Triples: 831-1.0098183-rdf-ntriples.txt
Original Record: 831-1.0098183-source.json
Full Text
831-1.0098183-fulltext.txt
Citation
831-1.0098183.ris

Full Text

MENTORING AS WORK-RELATED SUPPORT: RELATIONSHIP WITH EMPLOYEE OUTCOMES by KLAUS G. SCHROEDER B.Math., University of Waterloo, 1974 B.A., University of Waterloo, 1977 M.A., The University of British Columbia, 1981 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (Industrial Relations Management Division) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA 5 October 1988 © KLAUS G. SCHROEDER, 1988 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at The University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the Head of my Department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. (Industrial Relations Management Division) The University of British Columbia 2075 Wesbrook Place Vancouver, Canada V6T 1W5 Date: 5 October 1988 ii Abstract This study investigated the relationship between supportive behaviours and employee outcomes. The supportive behaviours were identified in the mentoring literature as being associated with the roles and functions performed by mentors. The term 'supportive' was used in order to recognize that people other than mentors (e.g., co-workers) could provide these behaviours. Questionnaires were used to assess employee outcomes and the level of supportive behaviours received by employees from different members in their organizations. The sample consisted of 624 managerial, technical, supervisory, and professional employees who worked for one of five organizations in British Columbia; 442 employees returned questionnaires. Respondents indicated the extent to which people with whom they had worked had provided them with behaviours associated with the eight supportive functions of Sponsoring, Exposure and Visibility, Teaching the Job, Teaching the Informal System, Protection, Role Modeling, Encouragement, and Personal Counselling. Principal component analysis indicated the presence of one general factor that accounted for over 50% of the variance; separate components for career and psychosocial functions (Kram, 1985) were not found. Principal component analysis indicated that all employee outcomes assessed in the study could be grouped into one of three types of outcomes: Job-Related (job satisfaction, role conflict, role ambiguity, organizational commitment, acceptance by co-workers), Skill Development (job, interpersonal, conceptual), and Promotional (rate of salary increase and promotions, satisfaction with progression). It was hypothesized that the level of supportive behaviours received by employees from as many as three sources would be positively related to all three types of outcomes, but that the relationship would be higher for the Skill Development and Promotional Outcomes than for the Job-Related Outcomes. This hypothesis was only partially supported. Although supportive behaviours were positively and significantly related to all types of outcomes, the relationship between behaviours and the Skill Development Outcomes was significantly higher than the relationships between behaviours and the other two types of outcomes. Failure to find a higher relationship between supportive behaviours and the Promotional Outcomes is discussed in relation to organizational reward systems. iii The level of supportive behaviours received from sources other than the highest source of supportive behaviours did not explain additional variance in employee outcomes over that explained by the level associated with the highest source alone. Failure to find incremental effects due to additional sources was most likely due to the high correlations (.70 to .80 range) among the level of supportive behaviours received from the different sources. These correlations may have been artifactually inflated because of the instructions that were used concerning which sources of supportive functions respondents were to rate on the supportive behaviours (respondents only rated sources on the supportive behaviours if the sources provided three or more functions). Because a number of hazards and disadvantages have been associated with intense mentor-protege relationships, it was hypothesized that the more evenly supportive behaviours are distributed across sources, the higher would be the employee outcomes. Although the way in which given levels of supportive behaviours were distributed across the sources was unrelated to employee outcomes, the hazards associated with given levels of supportive behaviours were negatively and significantly related to employee outcomes (the Job-Related ones, in particular). Methods for reducing the level of hazards are discussed. The scale that was developed to assess supportive behaviours was found to be reliable, content valid, and construct valid. Possible uses of the scale are discussed. iv Table of Contents Abstract ii List of Tables '. viii List of Figures x Acknowledgements xi Chapter 1: Introduction 1 Chapter 2: Mentoring Behaviours and Employee Outcomes 3 Career Functions 4 Sponsoring 4 Exposure and Visibility 5 Teaching the Job 6 Teaching the Informal System 7 Protection by Prevention 8 Protection by Absorption 9 Psychosocial Functions 11 Role Modeling 12 Encouragement 13 Personal Counselling : 14 Friendship 16 What is Mentoring? 17 Are Mentors Necessary for Effectiveness? 20 The Putative Benefits of Mentoring 22 Non-controlled studies 23 Controlled studies 27 Summary of mentoring studies 31 Multiple Sources of Social Support 33 Supportive Behaviours and Supportive Functions 40 The Importance of the Hazards, the Patterning, and the Quality of Supportive Behaviours 41 The hazards and disadvantages of mentor relationships ....41 The patterning of supportive behaviours across multiple sources 43 The quality of supportive behaviours 45 What is Career Success? 46 Career stages 47 Outcome variables 51 V Summary 52 Chapter 3: Research Questions and Hypotheses 53 Chapter 4: Method 72 Pilot Study One 72 Pilot Study Two 73 Pilot Study Three 73 Measures 74 Organizational Commitment 74 Role Conflict and Role Ambiguity ;..74 Job Satisfaction 75 Satisfaction with Progression 76 Skill Development 76 Acceptance by Co-workers 78 Supportive Behaviours and Functions 78 Hazards of Mentor-Protege Relationships 85 Definitions of a Mentor 86 Friendship 86 Characteristics of Relationships 89 Extraversion 89 Self-Esteem 90 Demographic Measures 90 Recruitment of Participating Firms 90 Generation of Employee Sample 91 Questionnaire Distribution and Return 92 Chapter 5: Results 93 Number of Survey Respondents 93 Description of Survey Respondents 95 Employee Outcome and Individual Difference Measures 96 Measures Associated with Sources of Support 99 The Relationship between Supportive Behaviours and Definitions of Mentors: A Validity Check 104 Research Question 1: The Structure of Supportive Behaviours 107 Research Question 2: The Structure of Employee Outcomes 110 Relationship between Employee Outcomes and Supportive Behaviours 116 The Impact of Multiple Sources of Support on Employee Outcomes 122 vi The Impact of Dispersion of Support on Employee Outcomes 127 Concentration of behaviours and functions in the highest source 127 Distribution of behaviours across sources 131 The Relationships among Hazards, Supportive Behaviours, and Employee Outcomes 133 The Relationship between Supportive Behaviours and Friendship 137 Relationship between Characteristics of Sources of Support and Mentor Definitions, Supportive Behaviours, and Employee Outcomes 140 Organizational position 142 Gender differences 145 Organizational tenure 147 Age differences 149 Chapter 6: Discussion and Conclusions 152 Structure of Supportive Behaviours 153 Structure of Employee Outcomes 154 Relationship between Suppportive Behaviours and Employee Outcomes 156 Incremental Effect of Multiple Sources of Behaviours on Employee Outcomes ....158 Effects of Dispersion of Supportive Behaviours on Employee Outcomes 160 Relationship between Hazards and Employee Outcomes 161 Relationships among Friendship, Characteristics of Sources of Supportive Behaviours, and Employee Outcomes 162 Limitations of the Study 163 Sample 163 Causality 164 Retrospective rationality 165 Spurious correlation 167 Quality of supportive behaviours 169 Focus on organizational tenure 172 Implications and Directions for Future Research 173 Theoretical issues 174 Practical applications 176 Conclusion 177 References 178 Appendix A: Items for Measures Developed for this Study 184 Appendix B: Instructions to Judges for Classifying Supportive Behaviours 190 vii Appendix C: Questionnaire Instructions and Item Order 196 Appendix D: Description of Survey Respondents 200 Appendix E: Means and (Standard Deviations) of Supportive Behaviours according to Definitions of Mentors 203 Appendix F: Correlations between Employee Outcomes and Measures for the Three Sources of Support 207 viii List of Tables Table 1: Relationship between Social Support and Mentoring Functions 36 Table 2: Assessment of the Sources of Supportive Functions 84 Table 3: Principal Component Analysis of Friendship Items Followed by Varimax Rotation 87 Table 4: Number of Distributed, Returned, and Excluded Questionnaires 94 Table 5: Descriptive Statistics for All Employee Outcome and Individual Difference Measures 97 Table 6: Descriptive Statistics for Measures Associated with the First Source of Support 100 Table 7: Descriptive Statistics for Measures Associated with the Second Source of Support 101 Table 8: Descriptive Statistics for Measures Associated with the Third Source of Support 102 Table 9: Mean Supportive Behaviour Scale Scores for Respondents with Different Response Patterns for the Mentor Definitions 106 Table 10: Results for Principal Component Analyses of Supportive Behaviours: Eigenvalues and Percentage of Variance Accounted For 109 Table 11: Correlations among Supportive Behaviour Scales - First Source 111 Table 12: Results for Principal Component Analyses of Supportive Behaviours: Component Loadings for Two Forced Components 112 Table 13: Correlations among Employee Outcome Measures 113 Table 14: Results for Principal Components Analysis of Employee Outcome Measures 115 Table 15: Correlations between Employee Outcome Composites and Supportive Behaviours from Different Sources 118 Table 16: Correlations between Employee Outcome Composites and Definitions of Mentors and Supportive Behaviours 121 Table 17: Hierarchical Multiple Regression Results for Employee Outcomes and Supportive Behaviours 124 Table 18: Hierarchical Multiple Regression Results for Employee Outcomes and Supportive Functions 126 Table 19: Hierarchical Multiple Regression Results for Employee Outcomes and Proportion of Supportive Behaviours Received from the Highest Source 129 ix Table 20: Hierarchical Multiple Regression Results for Employee Outcomes and Proportion of Supportive Functions Received from the Highest Source 130 Table 21: Hierarchical Multiple Regression Results for Employee Outcomes and the Distribution of Supportive Behaviours across Sources 132 Table 22: Correlations among Hazards, Supportive Behaviours and Employee Outcome Composites 134 Table 23: Hierarchical Multiple Regression Results for Employee Outcomes and the Hazards Associated with Supportive Behaviours 138 Table 24: Mean Levels of Friendship for Co-workers, Supervisors, and Organizational Members at Higher Levels than Supervisors 141 Table 25: Characteristics of Sources of Supportive Behaviours 143 Table 26: Relationship between Organizational Position of Source and Mentor Definitions, Supportive Behaviours, and Employee Outcomes 144 Table 27: Relationship between Source-Respondent Gender Differences and Mentor Definitions, Supportive Behaviours, and Employee Outcomes 146 Table 28: Relationship between Stage in which Source was Met and Mentor Definitions, Supportive Behaviours, and Employee Outcomes 148 Table 29: Relationship between Source-Respondent Age Discrepancies and Mentor Definitions, Supportive Behaviours, and Employee Outcomes 150 Table 30: Correlations of Extraversion, Self-esteem, and Educational Level with Employee Outcome Composites and Supportive Behaviours 170 X List of Figures Figure 1: Relationships between Supportive Behaviours and Employee Outcomes 57 Figure 2: Relationship between Supportive Behaviours and Hazards 63 Acknowledgements I wish to thank Craig Pinder for acting as my supervisor, ln addition to keeping me on track and motivated over the course of this project, Craig could always be counted on to provide valuable insight when I needed it most. The support and encouragement provided by Craig and the remainder of my committee (Peter Frost, Dan Perlman, and Gordon Walter) contributed to making work on my dissertation a rewarding experience. 1 would also like to thank my 'peer-pal', Laura-Lynne McBain, for so willingly sharing information and ideas as well as Sandra van Duyn for coordinating the work required to ensure a smooth data collection procedure. Finally, my thanks go to Marsha Schroeder for the support that her enthusiasm, interest, and help in my work imply. 1 Chapter 1  Introduction The term 'mentor' has its roots in Greek mythology. Mentor was the friend Odysseus entrusted with the care of his son Telemachus in Odysseus' absence. The relationship that developed between Telemachus and Mentor was a very comprehensive one that affected virtually all aspects of Telemachus' life, in modern management the term mentor usually refers to a person who contributes to the career development of another person (usually called the protege) by acting as a teacher, guide, or coach (Henderson, 1985). As such, mentor relationships in modern organizations often are less intense and comprehensive than the relationship between Telemachus and Mentor. This has led to some disagreement in the literature concerning how comprehensive relationships between individuals have to be before they are termed mentor relationships. Despite this disagreement, in the 1980s and late 1970s much has been written on the topic of mentoring in the work place. This level of activity is probably due to the purported beneficial effects of mentoring on the development and career success of individuals (Hunt & Michael, 1983; Orth & Jacobs, 1971). Unfortunately, relatively little research has empirically addressed the issue of whether mentored employees exhibit different levels of outcomes than do nonmentored employees. Also, little research has been directed at determining the relationship between level of mentoring and the level of outcomes such as employee job and career satisfaction. One purpose of the present research was to redress this relative imbalance in the mentoring literature by investigating the relationship between specific mentoring behaviours (e.g., sponsoring, teaching the job) and positive employee outcomes (e.g., job satisfaction, satisfaction with career progression). A second purpose of this study was to determine whether the relationship between provision of mentoring behaviours and employee outcomes is affected by the number of people who provide these behaviours. Traditionally, studies that have investigated the effectiveness of mentoring have compared individuals who report having had a mentor (the mentored group) with individuals who report not having had a mentor (the nonmentored group). The problem with such an approach is that it ignores the possibility that individuals in the nonmentored group may also have been the recipients of mentoring behaviours, albeit from a variety of sources. Inclusion of such a group of individuals in the nonmentored group may cause studies comparing mentored and nonmentored groups to show fewer significant differences due to mentoring than might 2 be warranted. This would be the case if the subgroup of nonmentored individuals who are the recipients of mentoring behaviours exhibit outcomes similar to the mentored individuals. For this reason, a more appropriate test of the relationship between mentoring behaviours and employee outcomes would involve assessment of the level of mentoring behaviours received from a variety of sources. Such an analysis would address the question of whether the level of mentoring behaviours provided by a group of individuals is more highly related to employee outcomes than is the level of mentoring behaviours provided by a single individual. If the relationship between employee outcomes and mentoring behaviours is found to be higher for behaviours received from a group of individuals than for behaviours received from any one individual, then having 'a mentor' may be less important than having a supportive work group that provides mentoring behaviours in concert. Bales (1958) suggested that the combination of leadership functions that leads to effectiveness can be derived from different sources. This study investigated whether this notion also is applicable for mentoring functions. In order to address the above concerns the variety of mentoring functions and behaviours that have been most commonly identified in the literature to date will be discussed. This will be followed by a review of studies and other works that suggest that individuals can be the recipients of mentor behaviours from a variety of sources in the work place rather than from just a single individual. Next, literature related to the effectiveness of mentoring will be discussed. Finally, by drawing on the literature related to social support, the theoretical framework used in the present study will be developed. This framework is based on considerations concerning the quantity, source, and type of mentoring behaviours received by employees in organizational settings. 3 Chapter 2 Mentoring Behaviours and Employee Outcomes Numerous researchers and writers (e.g., Alleman, 1985; Farren, Cray, & Kaye, 1984; Kram 1985; Missirian, 1980; Reich 1985, 1986; Schein, 1978; Zey, 1984) have delineated the roles that mentors play, the functions that mentors perform, and the behaviours that mentors engage in on behalf of proteges. As noted by Kram (1985), a review of studies on mentoring shows a great deal of overlap in the behaviours discussed and identified. Most of these behaviours can be classified into one of the two broad categories of mentor functions identified by Kram (1985)-career functions and psychosocial functions. The career functions "...enhance learning the ropes and preparing for advancement in an organization" (Kram, 1985, p. 22) whereas the psychosocial functions "...enhance a sense of competence, clarity of identity, and effectiveness in a professional role" (Kram, 1985, p. 22). Because Kram's (1985) typology of mentoring functions appears to be the most comprehensive yet presented, the review of mentoring behaviours, functions, and roles to follow will use this typology for the purpose of outline where possible. Strict adherence to the functions named by Kram (1985) is not possible because some of the behaviours discussed by Kram (1985) under one function are more often discussed by other writers under another name. However, the distinction between career and psychosocial functions appears to be conceptually useful and will be retained in the review to follow. Two points of clarification need to be made before the individual functions are discussed. First, the terms 'function' and 'role' are often used interchangeably in the mentoring literature. For instance, Klopf and Harrison (1981) discussed the roles of sponsor and model, while Kram (1985) discussed the functions of sponsoring and role modeling. Mentoring roles and functions differ, however, from mentoring behaviours. In this study, mentoring behaviours are discussed as behaviours that mentors engage in when they are performing a particular function or playing a particular role. In other words, mentoring behaviours are always associated with a mentoring function or role. The second point of clarification concerns the number of functions provided. Very few mentors will perform all of the mentoring functions in any particular relationship. As noted by Kram (1985), some functions will be more common than others and relationships will differ in the number of functions provided. The point to keep in mind while reading the following is that mentors typically perform a variety of functions but not necessarily all of them. The notion that different people can provide different 4 functions will be discussed at a later point. Career Functions Kram (1985) described the five career functions of sponsoring, exposure and visibility, coaching, protection, and challenging assignments. Although the behaviours discussed under these titles also are discussed by many other writers, these titles will not be completely appropriate for present purposes. The functions to be discussed here are (a) Sponsoring, (b) Exposure and Visibility, (c) Teaching the Job, (d) Teaching the Informal System, (e) Protection by Prevention, and (0 Protection by Absorption. Sponsoring and Exposure and Visibility are taken from Kram's (1985) typology. The terms Teaching the Job and Teaching the Informal System were suggested by Alleman (1985) who distinguished between teaching the job and teaching organizational politics. Both of these functions were discussed by Kram (1985) but under different names. Teaching the Job was discussed under challenging assignments and teaching politics was discussed under coaching. Given that many writers on the topic of coaching include Teaching the Job but not Teaching the Informal System as an aspect of coaching (e.g., The Woodlands Group, 1980), Kram's (1985) conceptualization of coaching will not be used here. Finally, the distinction between Protection by Prevention and Protection by Absorption was suggested by a review of material of various authors (e.g., Alleman, 1985; Zey, 1984). This distinction will be discussed subsequently. In summary, the career functions to be discussed largely are those identified by Kram (1985) but the terminology has been changed somewhat to provide greater conceptual clarity. Sponsoring and Exposure and Visibility are left unchanged, coaching is discussed under Teaching the Informal System, protection is broken down into the two forms of protection, and challenging assignments are discussed under Teaching the Job. Sponsoring. There exists some confusion in the literature concerning the distinction between Sponsoring and providing Exposure and Visibility. This confusion arises because a number of writers discuss the provision of Exposure and Visibility as a Sponsoring function (e.g., Farren et al., 1984; Henderson, 1985; Schein, 1978) while others (Kram, 1985; Missirian, 1980; Zey, 1984) view Exposure and Visibility as distinct from Sponsoring and discuss the two functions separately. For present purposes the two functions will be discussed separately using the distinction made by Kram (1985) as a point of departure. 5 Kram (1985) noted that Sponsoring involves nominating proteges for lateral or vertical moves within an organization. This definition is somewhat restrictive because it does not allow for the possibility that mentors themselves may be responsible for promoting or assigning proteges to different positions. The definition is restrictive also because an organizational move is necessarily involved. Here Sponsoring will be viewed somewhat more broadly as the communication of information concerning the potential and competence of proteges (Kram, 1985; Schein, 1978). As such, Sponsoring may help proteges obtain particular opportunities or positions (Klopf & Harrison, 1981) but need not necessarily do so. The goal of communication of positive information need not be an immediate one but, rather, may be to ensure that proteges are considered when new opportunities arise. One important aspect of Sponsoring is that it involves only mentor activity and not protege activity; mentors act on behalf of proteges. This, as will be discussed shortly, is what distinguishes Sponsoring from the provision of Exposure and Visibility. Kram (1985) noted that Sponsoring is the most frequently observed career function. Reich (1986), in a study of mentoring in both private and public sector organizations, found that over half of the mentors applied pressure on others in order to obtain promotions for their proteges. Missirian (1980) also found that one of the most common mentoring behaviours was promoting proteges steadily and often or suggesting that this be done. When mentors themselves are responsible for promoting or assigning proteges to different positions they still can be viewed as sponsors, as conceptualized above. This is so because promoting or assigning proteges to different positions communicates, at least implicitly, information concerning the potential and competence of proteges. Other forms of Sponsoring include assisting proteges in gaining admission into management or in-house training programs (Zey, 1984), having proteges assigned to task forces, committees (The Woodlands Croup, 1980), or special project teams (Farren et al., 1984). In general, mentors act as sponsors by being advocates (Henderson, 1985) who single out proteges from their peers (Alleman, 1985). In this study Sponsoring will be defined as communicating information to others, either verbally or nonverbally, concerning the potential or competence of an employee. Exposure and Visibility. Although Sponsoring might be viewed as providing proteges with Exposure and Visibility, Sponsoring, as noted above, does not involve any protege activity. The provision of Exposure and Visibility, however, does involve protege activity. 6 Mentors fulfill the Exposure and Visibility function by assigning proteges tasks that will lead proteges to interact with senior managers (Kram, 1985). Unlike the case of Sponsoring where mentors provide information concerning proteges, Exposure and Visibility allows proteges to communicate information concerning their potential and competence personally. While Sponsoring involves having others know about proteges, Exposure and Visibility involves having others know proteges personally. Zey (1984) refers to providing Exposure and Visibility as marketing the protege. Such marketing, while including exposure of proteges to senior management, also may include introducing proteges to different segments of the business community such as salespeople and competitors. Mentors can provide Exposure and Visibility in a variety of other ways. Zey (1984) and The Woodlands Croup (1980) noted that including proteges in meetings attended by senior executives will result in visibility. Other techniques for providing exposure include allowing proteges to make presentations to senior management (Farren et al., 1984), having proteges invited to as many official functions as possible (Zey, 1984), and providing proteges with introductions to others (Alleman, 1985). While not all methods of providing Exposure and Visibility are related to the organizational duties of proteges (e.g., introductions, attending meetings), all involve the physical presence of proteges. Sponsoring, as conceptualized above; does not involve proteges directly. In general, mentors provide Exposure and Visibility when they ensure that key people in the organization and, perhaps outside the organization (Zey, 1984), become acquainted with a particular protege. In this study Exposure and Visibility will be defined as providing an employee with tasks or opportunities that ensure that others will become personally acquainted with that employee. Teaching the Job. The mentor function of Teaching the Job has been discussed under a variety of terms including coaching (Farren et al., 1984; The Woodlands Group, 1980), developing talent (Schein, 1978), and assigning challenging tasks (Alleman, 1985; Kram, 1985). Whatever terminology is used, mentors Teach the Job when proteges develop "...essential technical and managerial skills through work that encourages learning" (Kram, 1985, p. 31). The specific mentor behaviours involved in Teaching the Job are difficult to delineate because they are likely to differ depending on the individuals and organizations involved. Various writers, however, have discussed general ways in which mentors Teach the Job. This can be accomplished by imparting knowledge about the 7 technical side of a position (Zey, 1984), providing guidance in real work (The Woodlands Croup, 1980), and creating hypothetical problems to help proteges learn (Farren et al., 1984). Taken as a group, such behaviours help proteges "...learn how to perform tasks and accomplish work-related goals..." (Alleman, 1985, p. 7). It is important at this point to remind the reader that the provider of only a limited subset of mentoring behaviours is not usually referred to as a mentor. This point needs to be considered especially when one examines the behaviours that are designed to Teach the Job. Numerous writers have pointed out that Teaching the Job can be found in many supervisor-subordinate relationships (e.g., Kram, 1985; Zey, 1984) that would not be referred to as mentoring relationships. Klopf and Harrison (1981), in particular, noted that teachers enable the development of certain competencies but that "...some go beyond this to become mentors" (p. 42). The obverse situation, namely that a relationship might be termed a mentoring one in the absence of behaviours designed to Teach the Job, also must be considered. While it is unlikely that mentors who are also direct supervisors of their proteges would not Teach the Job, this lack of teaching would be a distinct possibility if mentors are employed in different functional areas or at hierarchical levels higher than their proteges' supervisors. Such differences in level or area might preclude or make Teaching the Job difficult. In summary, while Teaching the Job has been found to be one of the most frequently mentioned mentoring behaviours (Missirian, 1980), neither its presence nor its absence will, by itself, determine whether a particular relationship is viewed as a mentoring relationship. In this study Teaching the Job will be defined as providing knowledge or experiences that help an employee learn technical and managerial skills. Teaching the Informal System. Teaching the Informal System includes, among other things, teaching proteges about organizational structure, politics, personalities (Zey, 1984), and culture (Farren et al., 1984). Teaching the Informal System, a topic discussed by Kram (1985) under the heading of coaching, helps proteges to gain the "...knowledge and understanding of how to navigate effectively in the corporate world" (Kram, 1985, p. 28). Zey (1984) noted that technical competence without the concomitant knowledge of the informal system can limit employees' career advancement opportunities. This occurs because, by not understanding structure, politics, and personalities, proteges might be more likely to present themselves in an unfavourable fashion. 8 As with Teaching the Job, it is difficult to delineate the specific behaviours involved in Teaching the Informal System because of individual and organizational differences. In a very broad sense, Teaching the Informal System includes the imparting of knowledge about matters as diverse as unwritten rules and norms, methods of resolving conflict, management style (Farren et al., 1984), how to approach organizational members, how to deal with specific personalities (Zey, 1984), who can be trusted, and about who holds power (Kram, 1985). While almost all writers who delineate mentoring behaviours discuss Teaching the Job, a variety of writers do not distinguish between Teaching the Job and Teaching the Informal System (e.g., Schein, 1978), or do not discuss the latter at all (e.g., Klopf & Harrison, 1981; The Woodlands Croup, 1980). This is unfortunate given the current widespread interest in understanding and managing organizational culture. Also, Kram (1985) noted that Teaching the Job and Teaching the Informal System are directed at different aspects of proteges' careers. Knowledge of the informal system may provide proteges with advancement opportunities but proteges have to have acquired the requisite skills before being able to take advantage of these opportunities. Proteges who are not adequately taught their jobs or the informal system may be hindered from advancing in their organizations. Another reason to view Teaching the Job and Teaching the Informal System as separate aspects of mentoring lies in the possibility that the two functions will be provided at different career stages or by different people in the organization. For instance, mentors who are at least two hierarchical levels removed from their proteges may stress Teaching the Informal System and leave Teaching the Job to their proteges' immediate supervisors. In short, it appears desirable to treat Teaching the Job and Teaching the informal System as two conceptually distinct mentoring functions. In this study Teaching the Informal System will be defined as imparting of knowledge to an employee concerning informal aspects of the organization such as norms, mores, and politics. Protection by Prevention. In providing protection, mentors ensure that proteges do not have undesired contact with other, usually senior, organizational members (Kram, 1985). As such, protection can be viewed as limiting the degree of exposure and visibility until proteges are ready for it. The literature suggests that mentors provide such protection in one of two relatively distinct ways. These will be referred to as Protection by Prevention and Protection by Absorption. Although the distinction between these two 9 types of protection has not been made explicit previously, it is useful to do so. As with the two forms of teaching discussed earlier, it is possible that the two forms of protection are offered at different career stages and/or by different organizational members. Mentors practise Protection by Prevention when they use their organizational influence to ensure that proteges are not assigned to tasks or positions where the proteges are unlikely to succeed (Kram, 1985; Zey, 1984). Protection by Prevention stops proteges from engaging in behaviours that might lead to criticism from superiors, peers, or colleagues (Alleman, 1985) or that have the potential to hinder the proteges' career advancement (Zey, 1984). For example, when mentors intervene to prevent proteges from being transferred (Zey, 1984) or assigned tasks proteges would find difficult to achieve (Kram, 1985), mentors practise Protection by Prevention. As with most mentoring behaviours, the specific forms that Protection by Prevention will take will differ depending on the individuals and organizations involved. In this study Protection by Prevention will be defined as preventing an employee from being exposed to tasks, positions, or situations in which the employee is unlikely to succeed. Protection by Absorption. The key distinction between Protection by Prevention and Protection by Absorption is that, in the former, mentors prevent potentially detrimental behaviours from occurring while, in the latter, mentors ensure that behaviours that have occurred are not associated with their proteges. 1 The goal of Protection by Absorption is to protect proteges from reprimands for behaviours that already have been enacted (Alleman, 1985). Mentors practise this form of protection when they allow proteges to make mistakes without personal risk (Schein, 1978). This is accomplished by not communicating these mistakes to others. Similarly, mentors can absorb the credit or blame for protege actions or nonactions in controversial situations (Kram, 1985); this ensures that proteges are not associated with potentially negative outcomes. Missirian (1980) found that protection, broadly defined, was one of the most often mentioned mentoring behaviours. Others have noted that the provision of protection, while generally a positive characteristic of a relationship, can be harmful to protege career advancement if it is provided excessively (e.g., Kram, 1985; Reich, 1986). However, this 1 Behaviours that have occurred include behaviours of action as well as nonaction. This means that mentors can protect proteges for errors of commission as well as for errors of omission; a mistake can be either type of error. 10 may depend on the type of protection under consideration. Mentors who Protect by Prevention do not allow their proteges to be exposed to situations in which the mentors do not expect their proteges to succeed. This type of protection could be harmful for proteges in a variety of ways. First, the protege may have succeeded in the situation from which the mentor shielded the protege if only exposure had been allowed, ln this case the protege would have forgone a potential learning experience as well as the opportunity to demonstrate competence. Second, even failure in such a situation might be beneficial to the protege. This would occur if what was learned in the situation outweighed, in the long run, the likely aspects of failure such as criticism and possible temporary loss of esteem. As these two examples demonstrate, Protection by Prevention may have negative impact on proteges if the protection results in missed learning opportunities or in an inability to demonstrate competence to others in the organization. On the other hand, the potential for negative protege effects due to Protection by Absorption do not appear to be as great. In the case of Protection by Absorption, protege behaviours have occurred already. This implies that the protege would have been allowed to 'take advantage of learning opportunities by attempting to perform well. It is only in the face of failure or potentially negative after-effects that the mentor protects the protege by attempting to. dissociate the behaviour from the protege in the minds of others. While Protection by Absorption has the negative consequence of not forcing or allowing proteges to take responsibility for their actions, excessive amounts of such protection do not appear to have as many potentially negative consequences for proteges as excessive amounts of Protection by Prevention. This is yet another reason why Protection by Absorption and Protection by Prevention should be viewed as two distinct types of protection. In this study Protection by Absorption will be defined as protecting an employee who has engaged in behaviours or actions that may be detrimental to the employee by taking the blame for or not telling others about these actions or behaviours. This completes discussion of the career functions most commonly identified in the mentoring literature as being related to employee career development. The six functions of Sponsoring, Exposure and Visibility, Teaching the Job, Teaching the Informal System, Protection by Prevention, and Protection by Absorption can help employees learn their role in the organization and prepare them for advancement (Kram, 1985). Also of potential importance to employees, however, are a number of psychosocial functions. 11 These will be discussed next. Psychosocial Functions Numerous writers have noted that one of the factors that distinguishes mentoring relationships from other relationships in the work place is the level of personal and emotional involvement entailed (e.g., Klopf & Harrison, 1981; Kram, 1985; Phillips-Jones, 1982; The Woodlands Croup, 1980). Kram (1985) discussed four psychosocial mentoring functions (role modeling, acceptance and confirmation, counselling, friendship) that, if present in a relationship, "...enhance a sense of competence, clarity of identity, and effectiveness in a professional role" (p. 22). ln contrast to the career functions which aid in career development, psychosocial functions are more personal in nature and are more related to personal development in a professional role. In other words, the career functions are related to factors associated with career success and the psychosocial functions are related to personal effectiveness. Because successful managers may not be the most effective managers (Luthans, 1988), it is important to distinguish between the two types of mentoring functions. While almost all writers discuss one or more of the career functions as being important in mentoring, a variety of writers make little, if any, mention of psychosocial functions (e.g., Farren et al., 1984; Gray, 1986; Missirian, 1980; Schein, 1978). This is more often the case when the primary emphasis concerns the establishment of formal mentoring programs (e.g., Farren et al., 1984; Gray, 1986). However, when psychosocial functions are discussed they usually can be classified into one of Kram's (1985) four types of psychosocial functions mentioned above. As was the case with the career functions, the psychosocial functions discussed here differ somewhat from those identified by Kram (1985). The major difference is that friendship was not viewed as a psychosocial function in this study. The reason for this will be discussed subsequently. The psychosocial functions considered here are (a) Role Modeling, (b) Encouragement, and (c) Personal Counselling. Role Modeling is taken directly from Kram's (1985) typology. The term Encouragement was suggested by Phillips-Jones (1982) and Schockett and Haring-Hidore (1985) and is used here instead of Kram's (1985) 'acceptance and confirmation.' Kram (1985) noted that mentors provide support and encouragement through the acceptance of their proteges. As such, the behaviours discussed by Kram (1985) under the heading of 'acceptance and confirmation' can generally be viewed as behaviours that encourage proteges. 12 The term Personal Counselling is used here instead of the broader term 'counselling' used by Kram (1985). Alleman (1985) developed separate scales for career counselling and general counselling; the latter involves mentor behaviours "...designed to contribute to the general personal growth and development..." (Alleman, 1985, p. 7) of proteges. It is primarily this type of counselling—here referred to as Personal Counselling--that is discussed by Kram (1985) under the heading of 'counselling.' Career counselling is, as the name implies, more concerned with career development and involves the provision of information concerning career paths and appropriate protege-job matches (Zey, 1984). Such information likely is provided through some of the career functions (e.g., Exposure and Visibility, Teaching the informal System). ln summary, the psychosocial functions discussed here are quite similar to those identified by Kram (1985). Role Modeling is left unchanged, acceptance and confirmation is discussed under Encouragement, and the general function of counselling is renamed Personal Counselling. Kram's (1985) remaining psychosocial function (friendship) is discussed separately as a potential aspect of mentor-protege relationships rather than as a mentoring function, per se. Role Modeling. Role Modeling involves protege emulation of mentor attitudes, values, or behaviours (Kram, 1985) and is one of the most frequently mentioned psychosocial functions (Kram, 1985; Missirian, 1980). By acting as role models mentors provide proteges with "...opportunities to learn by osmosis, observation, and association" (Missirian, 1980, p. 88). An important aspect of Role Modeling, and one which distinguishes it from what is commonly referred to as teaching, is that mentors need not be aware that they are providing proteges with models from which to learn. In other words, even though proteges learn by observing mentor behaviours, mentors do not necessarily act specifically for the benefit of proteges in order to teach. Kram (1985) suggested that Role Modeling can be effective because of the respect and admiration that is felt. It might be argued, however, that such emotional attachment may not be necessary, and perhaps even detrimental, for Role Modeling to be effective. When proteges begin to identify too closely with mentors the positive halo produced by this identification may limit the objective selection on the part of the proteges of which values, attitudes, and behaviours to model. Proteges may model too many aspects of their mentors. Zey (1984) noted that proteges can learn from mentors by consciously failing to emulate some of the observed actions. Such learning may be difficult if the emotional attachment or the degree of identification of proteges with mentors is too 13 great. Given these considerations, Role Modeling here will be viewed as learning through observation of mentor behaviours (Schein, 1978). This conceptualization allows for learning to take place through emulation as well as lack of emulation of mentor behaviours. Values and attitudes will not be discussed further because, although a variety of writers discuss emulation of values, attitudes, characteristics, or behaviours (e.g., Klopf & Harrison, 1981; Kram, 1985; Missirian, 1980), most concrete examples of Role Modeling concern mentor behaviours. Most writers who discuss Role Modeling emphasize the importance of models for learning about interpersonal, as opposed to technical, aspects of organizational life. Kram (1985) suggested that proteges may emulate how their mentors manage work groups or relate to significant others in the organization. Zey (1984) discussed proteges who, by observing their mentors, learned how to react to clients and to manage people in meetings. In general, what proteges learn from role models is not so much the content of their jobs but, rather, the processes through which their jobs can be performed effectively. Thus, the knowledge gained through Role Modeling is much the same as that imparted by mentors under the career function of Teaching the Informal System (e.g., methods of resolving conflict, management style). However, the method by which proteges gain this information is different. In Teaching the Informal System mentors consciously impart the knowledge through verbalizations. In Role Modeling the onus is on proteges to notice what is appropriate and effective behaviour; mentors need not be aware that they are serving as models for their proteges. In this study Role Modeling will be defined as being a model or example for an employee thereby allowing the employee to learn by observation. Encouragement. In providing Encouragement mentors help proteges feel confident, competent, and worthwhile in a professional role (Phillips-Jones, 1982). Kram (1985) noted that proteges derive a sense of self from the positive regard that mentor support and encouragement imply. Often the Encouragement function leads to higher levels of trust in the relationship (Alleman, 1985; Kram, 1985; Reich, 1986). While such trust may, in and of itself, boost protege confidence (Zey, 1984), it also allows proteges to take certain risks that might otherwise be avoided (Kram, 1985). Further, with high levels of trust proteges need not overly concern themselves with the conflict that may be created by disagreeing with their mentors. As noted by Kram (1985), relationships characterized by high levels of Encouragement tolerate differences of opinion. In this way learning 14 opportunities also are provided. In general, by providing Encouragement mentors motivate proteges by helping them to see how challenges can be achieved (Henderson, 1985) or by instilling the confidence necessary for achievement (Phillips-jones, 1982; Zey, 1984). As such, Encouragement may increase proteges' expectancies that their efforts will lead to positive outcomes (Vroom, 1964). The provision of Encouragement differs somewhat from the career functions of Sponsoring, Protecting, and the provision of Exposure and Visibility discussed earlier. While the career functions also indicate support for proteges, this support is of a less personal nature than that discussed here under the term Encouragement. The psychosocial function of Encouragement is a relatively private aspect of a mentoring relationship and need not be visible to other organization members. It should be noted that most writers do not explicitly recognize the provision of Encouragement as a mentoring function. However, almost all writers who view emotional attachment as an important, if not essential, aspect of mentoring relationships at some point stress the importance of Encouragement (e.g., Klopf & Harrison, 1981; Kram, 1985). Generally, these writers stress that provision of Encouragement can lead to personal development. Kram (1985), for instance, noted that proteges benefit by deriving a sense of self. Both Zey (1984) and Phillips-Jones (1982) indicated that Encouragement was important for the development of protege self-confidence. However, this psychosocial function also may be beneficial for protege career development because, as noted by Zey (1984), some level of Encouragement and the support it implies is needed in order for proteges to make the adjustment to senior management. This is so because without mentor Encouragement proteges may not develop the necessary confidence to approach situations in which success is necessary for upward mobility. In this study Encouragement will be defined as providing encouragement or support that helps an employee feel competent, confident, or worthwhile. Personal Counselling. Kram (1985) viewed the counselling function as one by which mentors allow proteges to discuss aspects of their lives that may detract from a positive sense of self in the organization. Personal Counselling is directed toward protege personal growth and development (Alleman, 1985) and often involves the discussion or resolution of conflict among concerns about self, career, and family (Kram, 1985; Phillips-Jones, 1982; Zey, 1984). As such, Personal Counselling differs from career counselling in that the latter 15 is directed almost exclusively at providing proteges with information related to career opportunities (Alleman, 1985). Personal Counselling is related more to matters originating outside the organization that may detract from protege career advancement and performance within the organization (Kram, 1985; Zey, 1984). When mentors provide Personal Counselling they act as sounding boards and confidants for protege concerns and ensure that these concerns are not made public. This suggests that a certain amount of trust need be present in a relationship before Personal Counselling will be sought or provided. While the specific content of Personal Counselling will depend on the individuals involved, one of the most frequently cited conflicts is that between family and career commitments (e.g., Kram, 1985; Phillips-Jones, 1982; Zey, 1984). Mentors can be of great benefit to proteges in this area because mentors often have experienced similar conflict at earlier career stages. Thus, mentors may be empathic and understanding and be able to provide proteges with advice. The outcome of Personal Counselling need not necessarily be the resolution of conflict. Kram (1985) mentioned the concern that proteges have over the possibility that values may have to be compromised for the sake of career advancement. This conflict between self and career may be difficult to resolve if such compromise is, indeed, necessary. However, by discussing the concern proteges can come to understand themselves better and, consequently, act more effectively in life (Klopf & Harrison, 1981). Another important consideration concerning Personal Counselling is that mentors may place protege needs above those of the organization (The Woodlands Group, 1980). This occurs when mentors advise organizationally effective proteges to leave the organization in order to enhance sense of self. Phillips-Jones (1982), for instance, discussed the case of a commercial artist who was advised to become self employed in order to balance the desire to play tennis with the need to earn a living. While such advice may not be common, this example underscores the trust and emotional attachment that may develop in mentoring relationships. Although the degree of support provided while proteges adjust to their organizations and careers will differ depending on the individuals involved, it has been suggested that some form of Personal Counselling is an essential aspect of all mentoring relationships (Klopf & Harrison, 1981; Zey, 1984). In this study Personal Counselling will be defined as discussing an employee's concerns involving self, career, or family. 16 Friendship. Friendship, the last psychosocial function discussed by Kram (1985), deserves comment at this point even though it was not viewed as a mentoring function in this study. Friendship is characterized by higher levels of intimacy, mutual liking, and informal social interaction than the other psychosocial functions discussed by Kram (1985). It is important to note that most writers view friendship not as a mentor function or role, but, rather, as a potential outcome of mentor relationships (e.g., Klopf & Harrison, 1981; Missirian, 1980; Reich, 1986). Missirian (1980), for instance, found that friendships tended to develop only after mentoring relationships had been terminated. Kram's (1985) data largely confirm this finding. In discussing the four phases of a mentor relationship (initiation, cultivation, separation, redefinition), friendships were most likely to develop in the redefinition phase. These findings may be due to the fact that many mentor relationships involve proteges' immediate supervisors (Kram, 1985; Roche, 1979). Given that these supervisors are responsible for rewarding and punishing their proteges, it may be difficult for them to enter into a relationship characterized by mutuality and reciprocity until after they no longer have direct authority over their proteges. If the proteges advance to the same hierarchical levels as their mentors and friendships develop, the relationship between the individuals can be viewed as changing from a mentor relationship to a peer relationship (Shapiro, Haseltine, & Rowe, 1978). A peer relationship is characterized by a reciprocity in which individuals help each other achieve success by sharing information, strategies, and advice. As such, a peer relationship differs from a mentor relationship in that individuals enter into the former on a more equal basis than into the latter. The foregoing is not meant to suggest that friendships can develop only after mentor relationships have dissolved. However, the available evidence (Klopf & Harrison, 1981; Kram, 1985; Missirian, 1980) suggests that this is usually the case. This evidence, taken in conjunction with the fact that few writers other than Kram (1985) discuss friendship as a mentor function or role, suggests that friendship should not be viewed as a psychosocial function in this study. The notion of friendship will not be totally ignored, however. The frequency with which it is discussed as a potential outcome of mentor relationships merits its inclusion in a study related to the phenomenon of mentoring. Another reason for including consideration of friendship lies in the possibility that minimal levels of it are necessary before any mentoring functions are provided. While a mentor relationship may not evolve 17 into a comprehensive friendship until the two individuals involved achieve equality in their relationship, it appears unlikely that either individual would take the time and effort required by a mentor relationship without at least a minimal level of liking for the other. This completes discussion of the psychosocial functions most commonly identified in the mentoring literature. The three functions of Role Modeling, Encouragement, and Personal Counselling considered in this study help employees develop personally in a professional role (Kram, 1985). Together with the career functions discussed earlier, the psychosocial functions purportedly lead to the development and career success of individuals (Hunt & Michael, 1983; Orth & Jacobs, 1971). It is important to note that the career and psychosocial functions considered in this study are not the only ones that have been attributed to mentors. Other functions that have been identified include playing the Devil's advocate (Farren et al., 1984), supporting the protege's dream (Burke, 1984; Levinson et al., 1978), and providing job autonomy (Alleman, 1985). Functions such as challenging the views of proteges and stretching proteges to the limits of their abilities also suggest themselves. While these functions also can be expected to lead to the personal and career development of employees, they are not included in the present study because they have been discussed much less frequently in the mentoring literature than have the functions described in detail above. Of concern in the next section are the questions of how many functions need to be provided and whether these functions have to be provided by a single individual before potential beneficial effects can be expected to accrue. What is Mentoring? The conceptualization of mentoring that is referred to most often in the literature as coming closest to the traditional one discussed in The Odyssey is that presented by Levinson, Darrow, Klein, Levinson, and McKee (1978) in their study of adult development. Levinson et al. (1978) noted that while a mentor might serve the functions of teacher, sponsor, host and guide, exemplar, and counsellor, the developmentally most crucial function is to support and facilitate the realization of the protege's dream. A mentor ...fosters the young adult's development by believing in him, sharing the youthful Dream and giving it his blessing, helping to define the newly emerging self in its newly discovered world, and erecting a space in which the young man can work on a reasonably satisfactory life structure that contains the Dream. (Levinson et al., 1978, p. 99) This view of mentoring suggests that comprehensiveness of influence in terms of number 18 of roles served by a particular individual as well as emotional involvement are essential aspects of mentoring relationships. A variety of writers concur that comprehensiveness is necessary before a relationship can be called a mentor-protege relationship. Clawson (1980) argued that when different people play the various roles of teacher, coach, sponsor, confidant, friend, and role model these people cannot be viewed as mentors. For Clawson (1980), a true mentor-protege relationship is one characterized by high levels of mutual personal involvement and a variety of mentoring roles; exactly how many roles are necessary is not discussed. Klopf and Harrison (1981), in discussing the five major mentor roles of counsellor, teacher, advisor, sponsor, and model, argued that a mentoring role is not being enacted when only some of these roles are present. Again, no mention is made of exactly how many of these roles are necessary in order for a relationship to be referred to as a mentor-protege relationship. Vance (1982) referred to the various mentor roles of sponsor, teacher, guide, patron, advocate, benefactor, and advisor and suggested that the mentor relationship is a highly inclusive and influential type of support. However, in defining a mentor as "...someone who acts as a teacher, guide, sponsor, patron, or adviser" (Vance, 1982, p. 8), it is still not clear how many roles need to be fulfilled before a person is referred to as a mentor. Two studies of female managers also stressed the importance of comprehensiveness of influence in mentoring relationships. Missirian (1980) placed the supportive relationships of peer, coach, sponsor, and mentor on a continuum and noted that while all roles could be developmentally significant, mentors had the greatest degree of influence. Mentors were placed at the high end of this continuum because only mentors could assume all of the other roles. Missirian (1980) noted three ways in which mentoring relationships differ from other kinds of supportive relationships; two of these relate to comprehensiveness. First, mentors will have more influence on proteges because of access to various resources obtained by being in a position of power; these resources include expertise, influence, status, money, and information. Second, in mentoring relationships there is a higher level of protege identification with the mentor's personal and professional values and behaviours. Thus, mentoring relationships are viewed as more comprehensive in influence and identification than other types of supportive relationships discussed by Missirian (1980). 19 Phillips-Jones (1982) distinguished between primary and secondary mentors. The major difference between the two is one of comprehensiveness of influence. Primary mentors, being more powerful than secondary mentors, can have much greater career influence because their power can be used to affect decisions relevant to their proteges. Secondary mentors were viewed as offering more specific career help, perhaps only rarely. Not all writers agree that comprehensiveness of influence or functions provided is necessary for a relationship to be called a mentoring relationship. Clawson (1985) noted that the definition of mentor used by Roche (1979) in his widely cited study of high-level executives did not include the developmental and learning aspects of traditional mentoring. The executives were deemed to have had a mentor if they answered 'yes' to the question "At any stage of your career, have you had a relationship with a person who took a personal interest in your career and who guided or sponsored you?" (Roche, 1979, p. 15). The same question was also used by Henderson (1985) in a study of mentoring in public service organizations. This definition is typical of that used in many research studies in that the roles or functions mentioned either are relatively few or do not all have to be present before a mentoring relationship is deemed to exist. This means that researchers studying mentoring in organizations have not required the comprehensiveness of functions suggested by the more classical definition. The reason for this appears to lie in the fact that when the classical Levinsonian definition is used the incidence of mentoring is much lower than that found in studies that define mentoring less restrictively. For instance, a study cited by Fury (1980) that used Levinson's definition found that only one person of the 100 interviewed had a mentor. Although the samples are quite different, this percentage stands in stark contrast to the incidence of mentoring found by Roche (1979)--66 percent. Clawson (1985) also found comprehensiveness of influence to be relatively rare in a study designed to assess the degree to which managers emulated the people who were most influential in their lives. Of over 200 relationships described, Clawson (1985) found only one to be truly comprehensive. Most managers emulated at most five of the fourteen aspects of life Clawson (1985) investigated. Although this study concentrated primarily on role modeling, Clawson (1985) concluded that the "...data highlight a modern definition of mentoring that centers on the work place but excludes the development of personal characteristics..." (p. 37). Earlier, Clawson (1980) noted the distinction between career mentors and life mentors. The results (Clawson, 1985) suggest that life mentors are rarer in organizations than career mentors. 20 Are Mentors Necessary for Effectiveness? Findings such as those cited above have led a number of writers to question whether people who provide a large number of functions are necessary to obtain the positive outcomes that have been associated with mentoring. Kram (1985) noted that one of the major misconceptions concerning mentoring is that "...finding a mentor is the key to individual growth and career advancement" (p. 199). She suggested that a relationship constellation including peers, mentors, supervisors, friends, and family would be more beneficial than a single mentoring relationship because each of the relationships in the constellation might provide one or more of the career and psychosocial functions. Further, in describing the various functions, Kram (1985) repeatedly noted that having only one provider of a particular function could be risky for the protege. Various other writers concur with Kram that multiple supportive relationships can be as beneficial as a single primary mentoring relationship. Darling (1985), who noted that mentors in the traditional sense are extremely rare, suggested the possibility of "...a succession of less intense, less encompassing relationships that, in conjunction with mentoring events, add up to very formative mentoring experiences" (p. 41). This thought is echoed by Halcomb (1980) who found that employees, unable to find one individual who provides a large number of functions, may settle for several short-term mentors and derive some support from each of them. Such short-term mentors appear to be similar to the three types of secondary mentors mentioned by Phillips-Jones (1982). Peer strategizers were viewed as peers, friends, lovers, neighbours, co-workers, and others who could help plan and implement career goals. Unsuspecting-hero role models were people who, without their knowledge, were emulated. Finally, career favour-doers were people who would do favours for individuals but usually just once. Phillips-Jones (1982) suggested that people should build a network of such helpers who can provide mentoring. Shapiro et al. (1978) discussed the four advisory/support roles of peer pals, guides, sponsors, and mentors that can be helpful in achieving career success. Missirian (1980) proposed a similar continuum based on degree of influence: peers, coaches, sponsors, and mentors. Three of these roles-coach, sponsor, and mentor-also were discussed by The Woodlands Group (1980) as management development roles. Implicit in all three of these writings is that different individuals can fulfill each of the roles mentioned. It should be noted that all of these roles are often discussed as particular roles that 21 mentors play (e.g., Kram, 1985; Klopf & Harrison, 1981; Schein, 1978). While mentors may fulfill all roles simultaneously, the notion that different people can provide different functions is being recognized. Bolton (1980) noted explicitly that "...role model, mentor, and sponsor... are not necessarily performed by the same individual and each may be represented by several persons though not necessarily at the same time" (p. 201). Kram (1985) made a similar point: ...it is likely that an individual will have, over the course of an organizational career, several developmental relationships that provide a range of critical career and psychosocial functions at each life/career stage. The wish to find one senior manager who will carry an individual through his or her career, and who will continue to be responsive to individual concerns, is one that is likely to generate considerable disappointment and disillusionment, (p. 623) Other empirical and anecdotal evidence suggests that multiple relationships are relatively common. An executive cited by Henderson (1985) noted: Four individuals have been supportive during the course of my career. Many of my peers consider all four mentors to me; I do not. Two, it is true, were supportive; they provided guidance and counseling. They made available opportunities in terms of assignments which provided high visibility. It should be noted, however, that the conscious effort to advance my career was lacking. Consequently, they were not mentors, ln the other two instances, definite conscious efforts were made to use the authority of their positions to obtain opportunities for me. (p. 858) Evidence of multiple relationships also comes from studies that report on the number of mentors employees indicate having had during their careers. In general, employees who report having had at least one mentor average somewhere between two and three mentors over the course of their careers (Henderson, 1985; Missirian, 1980, Roche, 1979). Unfortunately, these studies do not report the number of mentors present at any particular time. The notion that an individual can be the recipient of mentoring behaviours from different sources leads one to question whether it is mentors, per se, who are needed for career advancement and personal development. The possibility exists that individuals who do not have one comprehensive mentoring relationship might be able to achieve success as readily as those who do as long as the combination of behaviours derived from different people in their relationship constellations is the same as the behaviours found in one comprehensive mentoring relationship. To illustrate this point, we assume that we have two individuals (A and B) who receive identical amounts of Sponsoring, 22 Protection by Absorption, Teaching of the Informal System, and Personal Counselling. We assume further that A's immediate supervisor provides all four of these functions for her; this would be viewed as comprehensive or traditional mentoring. B, on the other hand, is Sponsored by the personnel officer who hired him, is Protected by his immediate supervisor, is Taught the Informal System by one of his co-workers, and receives Personal Counselling from another co-worker. In the traditional sense B would be deemed to not have a mentor because no single provider of mentoring functions provided a high number of functions. Thus, even though B receives the same level of mentoring as A, A is deemed to have a mentor while B is not. The question raised earlier can now be reworded: Is level of mentoring, whatever the source, related to personal and career development or must this level be derived from one source in order to be related to positive outcomes? This question cannot be answered directly because, to date, no research has addressed explicitly the issue of multiple providers of mentoring functions. In summary, because traditional mentors characterized by comprehensiveness of influence are relatively rare in the modern work place (Clawson, 1985; Darling, 1985; Kram, 1985), it is likely that employees will be the recipients of mentoring behaviours from a variety of different sources including co-workers, support staff, supervisors, and, perhaps, even mentors. To date, it has not been shown that employees who are recipients of a large number of mentoring behaviours from one source will exhibit higher levels of personal and career development than will employees who receive these same mentoring behaviours from a variety of sources. One purpose of the present study was to address this point. Before the issues involved in such an investigation are discussed, studies of mentoring are reviewed in order to illustrate the kinds of research done to date. The Putative Benefits of Mentoring Although much has been written concerning the beneficial effects of mentoring for mentors, proteges, and organizations (see Hunt & Michael, 1983 for a review) relatively few studies have demonstrated the effectiveness of mentoring empirically. Most studies on mentoring can be grouped into one of two broad categories-non-controlled or controlled. For present purposes non-controlled studies will be defined as those that describe mentor-protege relationships without comparing proteges to other organizational members who have not had mentors. Such studies (e.g., Bova & Phillips, 1981; Bowen, 1985; Kram, 1985; Reich, 1985, 1986; Vance, 1982; Zey, 1984) either did not sample people 23 without mentors or, if they did, did not discuss such individuals. Controlled studies, on the other hand, are those that compare people who have indicated the presence of a mentor to those people who have not identified a mentor (e.g., Alleman, 1985; Fitt & Newton, 1981; Henderson, 1985; Missirian, 1980; Roche, 1979). Although there are studies in both groups--non-controlled and controlled-that do not discuss outcomes of mentoring relationships, the review of studies to follow will focus on the type of sample employed, the definition of mentoring used, the incidence of mentoring, and the outcomes examined. Non-controlled studies. Perhaps the most widely cited descriptive study is that conducted by Kram (1985). This report is interesting because even though mentoring was not explicitly defined, the incidence of mentoring was not reported, no comparisons were made, and outcomes were only vaguely discussed, the description of mentoring is both rich in detail and thoroughly discussed. Kram's sample consisted of 18 dyadic relationships between junior and senior managers identified in a public utility company with approximately 15,000 employees. In total, 15 junior managers and 16 senior managers were interviewed; three junior managers reported on two separate relationships and two of the senior managers were identified as significant others by each of two junior managers. Kram (1985) did not use a definition of mentoring to identify these dyads because the term 'mentor' was found to have a variety of connotations. Instead, managers were asked to identify more senior managers with whom they had had developmentally enhancing relationships. Interestingly, of the first 15 managers interviewed only three were able to identify such senior managers; this finding again points to the relative unavailability of comprehensive mentoring relationships in organizations. To arrive at her final sample Kram (1985) relied on internal personnel staff to identify individuals who appeared to have mentors or sponsors. Given the lack of a formal definition of mentoring, it is impossible to estimate the incidence of mentoring in the organization studied. Kram's (1985) greatest contribution appears to lie in the rich description of both mentoring functions and the stages or phases of mentoring relationships. In future these descriptions likely will aid other researchers in empirical investigations of the mentoring phenomenon. A study by Bova and Phillips (1981) provided information concerning the incidence of mentoring. Members of professional associations and people enrolled in graduate programs were provided with the following definition of mentors: 24 For the purpose of this study we will define mentors as those who practice most of the following principles: (a) Try to understand, shape and encourage the dreams of their proteges; (b) Often give their blessing on the dreams and goals of their proteges; (c) Provide opportunities for their proteges to observe and participate in their work by inviting their proteges to work with them; (d) Teach their proteges the politics of "getting ahead" in the organization. A mentor is usually a person of high organizational or specific career status who by mutual consent takes an active interest in the career development of another person. (Bova & Phillips, 1981, Appendix I) Respondents, who ranged in age from 19 to 52 years, were asked whether they felt that they currently had or had had a mentor. Of the 87 women and 73 men in the study, 80 women (92 percent) and 67 men (92 percent) answered in the affirmative. This study primarily addressed the issue of when in the life cycle mentors were encountered. Although two questions concerning the effect that mentors had on career advancement were asked, the answers to these questions were not discussed in the study. The primary contribution of this study was the finding that most respondents with mentors, both men and women, were most likely to have first encountered them near the beginning of their organizational tenure. The majority of respondents first encountered their mentors either in early adulthood (66 percent) or during a mid-life transition (31 percent). Bova and Phillips (1981) suggested that first encountering mentors during a mid-life transition was due to career change, first entry, or reentry into the world of work. Two much more comprehensive studies were reported by Reich (1985, 1986). Respondents in the first study (Reich, 1985) were approximately 416 of 520 executives sampled from a Columbia University Executive Program. The average age of the sample, which was over 95 percent male, was 42 years. Respondents included presidents, vice-presidents, general managers, and division managers of major corporations. When asked whether a key individual had influenced their career development, about 90 percent answered in the affirmative. This high incidence of mentoring may, as in the Bova and Phillips (1981) study, be due to the relatively general definition of mentoring used. This study reported primarily on the types of mentor assistance received (e.g., assignment to special projects, political assistance), and perceived benefits to the protege such as chances to develop abilities, to be creative, and to increase self-confidence. Seventy-two percent of the respondents indicated that their mentor had "...contributed substantially to their career development" (p. 45). While this result could be viewed as evidence that mentoring is effective, it should be noted that the definition of mentoring used specifically asked about career development as well. 25 The second study (Reich, 1986) was similar to the first except that only women were sampled. Respondents (37% of the 353 sampled) averaged 41 years of age and included professionals, scholars, and executives. The definition of mentoring used was the same as in Reich's (1985) previous study. The incidence of mentoring among the female respondents was found to be 77 percent. Overall, the results of the second study were quite similar to the first. The comparison between men's and women's mentoring experiences indicated more similarities than differences. For example, both women and men became proteges at similar ages and reported similar benefits and drawbacks of their relationships. Bowen's (1985) study is similar to Reich's (1986) in that the focus was exclusively on female proteges. The definition of mentoring used by Bowen was as follows: Mentoring occurs when a senior person (the mentor) in terms of age and experience undertakes to provide information, advice and emotional support for a junior person (the protege) in a relationship lasting over an extended period of time and marked by substantial emotional commitment by both parties. If opportunity presents itself, the mentor also uses both formal and informal forms of influence to further the career of the protege, (p. 31) This definition differs from that used in most studies in two major ways. First, it is similar to the traditional view of mentoring presented by Levinson et al. (1978) in that it suggests comprehensiveness of influence as well as emotional attachment; most definitions of mentoring used in empirical research do not stress both of these aspects of mentoring. Second, the definition explicitly states that mentors are necessarily older than their proteges. Other studies have found that proteges can be older than their mentors (Phillips-Jones, 1982). It would have been interesting to see the incidence of mentoring that would be found in organizations with the use of such a definition. Unfortunately, Bowen (1985) did not use the definition quoted above nor any other definition to identify mentor-protege relationships. . Rather, the 32 relationships studied were located through word-of-mouth referrals and personal acquaintances. Both mentors and proteges were from a wide variety of fields (e.g., banking, health care, law) and represented a variety of organizational levels from presidents to first-line supervisors. While the major focus of Bowen's (1985) study was to compare proteges with male mentors (n — 18) to those with female mentors (n = 14), the most interesting finding was that gender of mentor was less important in respondents' perceptions of career progress than were the number of mentor functions provided. Agreement with the question "My career is moving along much faster than that of other women of my age and experience" (Bowen, 1985, p. 32) was assessed on a five point scale. Mentor 26 functions that were coded according to Kram's (1985) categories accounted for almost 20 percent of the variance in perceptions of career progress; unfortunately, details of this analysis were not given. This finding is all the more impressive when one considers the homogeneity of the sample; all respondents purportedly were mentors or proteges. This suggests that the relationship between perceptions of career progress and mentoring functions provided would be even higher in a sample including some people with less extensive relationships. The last non-controlled study to be reviewed here is one by Vance (1982), who investigated mentoring in a sample of 71 influential leaders in the field of nursing. Vance (1982) described the sample as an elite group consisting of people who had "made it" in nursing and who had had an impact on the profession. In response to the definition of a mentor as "...someone who serves as a career role model and who actively advises, guides, and promotes another's career and training" (Vance, 1982, p. 10), 83 percent of respondents indicated that they had had one or more mentors. Although this study did not address the beneficial aspects of mentoring empirically, the categorization of the types of help provided by mentors to these influential nurses is of interest because of the large degree of overlap between this categorization and the mentoring functions delineated by Kram (1985). Ranked by frequency of response the major categories of types of help reported were (a) career advice, guidance, and promotion (24%); (b) professional role modeling (20%); (c) intellectual and scholarly stimulation (15%); (d) inspiration and idealism (14%); (e) teaching, advising, and tutoring (13%); (f) emotional support (11%); and (g) other (3%). Unfortunately, these figures do not allow a determination of the comprehensiveness of the types of help received because the percentages given were percentages of the total number of responses; the total number of responses was not reported. Of interest is the finding that similar mentoring functions were identified by relatively diverse samples-influential nurses (Vance, 1982) and managers of a public utility (Kram, 1985), In summary, many non-controlled studies (of which the above are representative) have described the functions and roles that mentors perform. The major limitation of such studies is that individuals who have not had mentors are largely ignored. This leads one to question the validity of the findings because it cannot be concluded that individuals who failed to identify mentors differed in any major way from individuals who did identify mentors (Alleman, Cochran, Doverspike, & Newman, 1984; Merriam, 1983). 27 Studies that use control groups overcome this limitation. Controlled studies. In one of the most widely cited studies on mentoring, Roche (1979) mailed questionnaires to the 3,976 executives listed in the "Who's News" column of The Wall Street Journal in 1977. In response to the question "At any stage of your career, have you had a relationship with a person who took a personal interest in your career and who guided or sponsored you?" (Roche, 1979, p. 15), 63.5 percent of the 1,250 respondents indicated that they had had one or more mentors; one-third of the respondents had had two or more mentors. Based on the response to this question, 'mentored' and 'nonmentored' groups were identified and compared on a variety of measures. Compared to the nonmentored group, those who reported having had a mentor earned higher salaries at younger ages, were more likely to follow a career plan, derived greater pleasure from their work, and expressed greater satisfaction with career progress. Given that this study is usually cited in order to document the effectiveness of mentoring in terms of employee outcomes, these results will be described in somewhat greater detail. It should be noted first that none of the comparisons between the mentored and nonmentored groups was tested for statistical significance. The conclusion that mentored individuals earned a higher salary at a younger age is based on the finding that the mentored group (average age of 47.3 years) earned an average of $118,900 annually while the nonmentored group (average age of 49.2 years) averaged $114,200 annually. In other words, the nonmentored group, while being somewhat older, earned approximately four percent less than did the mentored group. This difference, which does not seem great, must be examined in light of the fact that employees in the mentored group were more likely to have both advanced degrees (50% versus 40% for employees in the nonmentored group) and MBA degrees (25% versus less than 20%). Educational attainment possibly could explain the difference in average salary. The other major aspects of this study (Roche, 1979) cited most frequently are that people in the mentored group (a) were more likely to follow a career plan, (b) derived greater pleasure from their work, and (c) expressed greater satisfaction with their career progress. For the first two of these results no figures were reported. In relation to career progress, half of the mentored group expressed very high satisfaction while only 40 percent of the nonmentored group were satisfied. The Roche (1979) study suggests that people who have had mentors exhibit more positive outcomes than do people who have not had mentors. Some of these differences 28 are relatively small, others are not well documented, and still others may be due largely to differences in education. As mentioned earlier, the relatively small differences may be due to the fact that people in the nonmentored group also had been recipients of behaviours related to the outcome measures concerned. Still, the Roche (1979) study is a good one in that it involved a large sample, included outcome measures related to mentoring effectiveness, and compared mentored and nonmentored groups on these measures. Henderson (1985) was stimulated by Roche's (1979) study of mentoring in the private sector to investigate mentoring in the public sector. Using Roche's (1979) definition of mentoring, Henderson (1985) found that 74 percent of the 822 respondents in a sample of 1600 executives in municipal, state, and federal governments had had mentors. Women and men averaged 2.72 and 2.44 mentors, respectively; respondents in Roche's (1979) study averaged approximately two mentors. One of the major differences between the Roche (1979) and Henderson (1985) studies is that Roche's sample was primarily male while Henderson's was primarily female. Given the many suggestions that male and female mentoring relationships differ significantly (e.g., Bowen, 1985; Haseltine et al., 1980) it is interesting to note that the comparisons between mentored and nonmentored groups in the Roche (1979) and Henderson (1985) studies were quite similar. Compared to nonmentored executives, Henderson (1985) found that mentored executives earned 11 percent more, attained executive positions at younger ages, were more satisfied with their careers, were more likely to have career plans, and had greater feelings of personal success in their careers. Henderson (1985) was even more remiss in reporting numbers than was Roche (1979). Information not provided includes average age and income, gender breakdown of the sample, and mean scale scores. Because no statistical procedures were employed to document the mean differences between people who reported having mentors and those who did not, it is not possible to determine the magnitude of the differences between the two groups. Also, it is not clear whether the higher level of education of the mentored group explains, at least partially, the differences found. This study is important, however, because it, as does Roche's (1979), attests to the beneficial aspects associated with mentoring. Another study that found a relationship between mentoring and salary was conducted by Fitt and Newton (1981). Although the definition of mentoring used was not reported, it was found that 24 of the 30 female managers interviewed had had mentors. This study is df interest because it suggests that, when level of education is controlled, 29 mentored employees still earn more than do nonmentored employees. Fitt and Newton (1981) reported that the 30 women earned between $21,000 and $90,000 annually (average salary was $45,000) and that those with advanced degrees earned only slightly more than the sample average. Given this finding, it appears unlikely that differential education levels can be used to explain why the 24 women who reported having mentors were better paid than the six women who had never had mentors. Still, the relationship between mentoring, salary, and education is in need of further investigation. The final area of work discussed here is that of Alleman (1985) and her associates (Alleman et al., 1984). Alleman (1985) developed the Alleman Leadership Development Questionnaire (ALDQ). The ALDQ can be used to determine "...the frequency or likelihood of specific mentor practices occurring" (Alleman, 1985, p. 3). As such, the ALDQ appears to be the first instrument that was developed for the purpose of measuring mentoring activity on a continuous rather than an all-or-none basis. Although other such instruments have been developed more recently by Douglas and Schoorman (1988), Scandura and Katerberg (1988), and Whitely, Dougherty, and Dreher (1988), the ALDQ is much more comprehensive than these other instruments in terms of the number of mentoring activities assessed. The ALDQ consists of 123 items that were grouped into 12 scales "...based on logical considerations of similar behaviors" (Alleman, 1985, p. 5). There are two forms of the ALDQ, one for use by the member of the dyad of greater rank or expertise (Form A) and one for use by the member with less rank or experience (Form B). For present purposes discussion will be limited to Form B--that completed by potential proteges. Form B asks the respondent to list the individual in relation to whom the respondent will complete the ALDQ. All scale items on the ALDQ ask the respondent to indicate how frequently the person listed by the respondent would engage in the behaviour listed. The validation sample used by Alleman (1985) consisted of 100 individuals; one half of these individuals completed Form A and one half completed Form B. Based on a yes/no response to a question asking whether the 50 individuals who completed Form B considered the person they were describing as their mentor, two groups were formed; there were 29 people in the mentored group and 21 in the nonmentored group. Total and subscale scores of the ALDQ were correlated with mentor group membership. Total scale scores correlated .48 with whether the person was named as a mentor; subscale correlations ranged from .32 to .49. Scale scores also were correlated with the degree to which respondents indicated that the described person had 30 a beneficial influence on their careers. Total scale scores correlated .52 with this index; subscale correlations ranged from .31 to .59. Given the magnitude of these relationships, it appears that Alleman (1985) made a good start in the development of a questionnaire to assess level of mentoring behaviours. There are, however, a number of serious problems with the ALDQ that need to be resolved before the instrument can be judged acceptable. First, it has not been demonstrated that the strategy used to assign items to scales was a defensible one. Although there is overlap between the scale names used by Alleman (1985) and the titles used to categorize mentor behaviours earlier in this chapter, many items placed on a scale by Alleman (1985) would not be exemplars of the categories discussed here. For instance, Alleman (1985) placed an item concerning recommendation of a person for an educational opportunity on a scale called 'teach the job.' In the present scheme, such an item would be considered an example of Sponsoring. A second problem with the ALDQ is that Alleman (1985) did not provide the descriptive statistics needed for an adequate evaluation of the instrument. Because many of the mentoring activities assessed by the ALDQ would be expected to occur very infrequently in many relationships (e.g., sponsoring people for club membership, taking people to professional meetings), scale scores may not be normally distributed. This may affect correlations with other variables of interest. Alleman (1985) also failed to comment on the factor structure of the 12 scales. Because most scales logically can be viewed as career or psychosocial (Kram, 1985) in content, this distinction needs to be addressed empirically. Failure to provide intercorrelations among the scales prohibits interested parties from assessing the structure of the scales. In sum, much more information on the ALDQ is needed. A final problem with the ALDQ is that respondents are asked to indicate how frequently they think they would be treated in a particular fashion by the person they are describing. Thus, respondents are not asked to consider past and present behaviours, but, rather, behaviours that they might expect to occur some time in the future. This view of mentoring is contrary to almost all descriptions of mentoring found in the literature that concentrate on the developmental aspect of the phenomenon. If mentoring is conceptualized as the degree to which a person might engage in particular behaviours (as Alleman clearly does), it is difficult to see how level of mentoring, thus defined, will 31 relate to employee outcomes on such variables as satisfaction with past rates of progress, advancement, and development. Summary of mentoring studies. Studies reviewed above were chosen to be representative of studies that have investigated mentoring in organizations. In the review emphasis was placed on sample used, definition, incidence, and outcomes reported. This was done for a variety of reasons. First, although many researchers have sampled from relatively elite populations (e.g., Missirian, 1980; Roche, 1979), studies such as those by Reich (1985, 1986) clearly indicate that mentors are likely to be found at most organizational levels. Second, incidence of mentoring will be related to the definition used to define one's terms. From the definitions given above it is clear that definitional concensus likely will elude researchers for some time to come. It should be recognized, however, that it is difficult to compare incidence of mentoring across different organizational settings unless there is definitional consistency; this fact made the comparisons between the findings of Roche (1979) and Henderson (1985) of interest. The third important aspect of the studies reviewed is that most studies, if they address the issue at all, do not adequately evaluate the relationship between mentoring and desired employee outcomes. Various outcomes addressed included the following: salary, perceived rate of career progression in comparison to others, contribution of mentors to career development, satisfaction with career and career progress, likelihood of having a career plan, age at which top executive levels are reached, pleasure derived from work, and degree of beneficial career influence. With the exception of Alleman (1985), these outcome measures were not assessed by comparing groups through the use of statistical procedures. This makes it difficult to assess the degree of difference between people who report having had a mentor, and those who do not. Another problem is that most outcome measures are used in only one or two studies. This fact, along with the different definitions of mentoring used in various studies, makes it difficult to discuss the effectiveness of mentoring across settings. An interesting aspect of the studies reviewed is that some studies have assessed the relationship between mentoring and employee development within one organizational setting, while others have assessed this relationship across the course of a career (i.e., potentially, across organizational settings). The distinction between mentoring behaviours 32 received within an organization and behaviours received across organizations deserves further consideration. Most writers suggest that mentors can enhance the career development of individuals (e.g., Hunt & Michael, 1983; Kram, 1985; Orth & Jacobs, 1971). This view is supported by studies that have focused on the extent to which mentors who were encountered at any point in individuals' careers have influenced their career development (e.g., Fitt & Newton, 1981; Henderson, 1985; Reich, 1985, 1986; Roche, 1979). Of interest is that studies that have focused on current mentor-protege relationships also have documented significant relationships between mentoring and protege development (e.g., Alleman, 1985; Bowen, 1985). Strictly speaking, this latter group of studies addressed the extent to which proteges' careers have developed within their current  organizations. Such within-organization development is equivalent to career development only for those individuals who were not previously employed by other organizations. These findings suggest that the relationship between mentoring behaviours and employee development can be studied as a career phenomenon as well as an organizational phenomenon. Because higher levels of control over extraneous factors may be achieved by limiting investigation to shorter periods of time, a focus on organizational experiences may be the preferred choice given the present level of sophistication of studies reported in the mentoring literature. One final aspect of the studies reviewed is of interest. None of the studies explicitly considered multiple sources of mentoring behaviours. By focusing on only one source of mentoring behaviours (i.e., a mentor) the possibility that individuals can be the recipients of mentoring behaviours from a variety of sources is ignored. If the combined level of mentoring behaviours received from a variety of sources is more highly related to outcome variables of interest than is the level of mentoring behaviours received from only one of these sources, then an approach to studying the relationship between mentoring behaviours and outcomes must incorporate the notion of multiple sources. The literature related to social support suggests such an approach; there is much overlap between the conceptual and methodological issues discussed in that area and those that need to be addressed here. 33 Multiple Sources of Social Support Social support has been defined as "...an exchange of resources between at least two individuals perceived by the provider or the recipient to be intended to enhance the well-being of the recipient" (Shumaker & Brownell, 1984, p. 13). If such outcomes as greater career progression and job satisfaction can be viewed as being related to individual well-being, then mentoring behaviours can be viewed as particular forms of social support. Social support is usually conceptualized in terms of either the existence, the structure, or the functional content of relationships (House & Kahn, 1985). When concern is with existence, emphasis is placed on determining whether individuals have access to significant others who are assumed to provide social support. Studies concerned with existence of social support might, for example, compare groups of individuals who have a spouse with groups of individuals who are single. In such a case, the mere presence of a spouse is interpreted as social support even though the quantity or quality of such support is not explicitly assessed. Such studies are conceptually similar to those reviewed earlier that compared mentored and nonmentored individuals; individuals who report having had a mentor are assumed to have been the recipients of mentoring behaviours. Studies concerned with the structure of relationships usually involve some form of network analysis. These studies would relate such network indices as size, density, and homogeneity to measures of well-being. Studies concerning the structure of relationships do not appear to be represented in the mentoring literature to date. Finally, studies concerned with the functional content of relationships investigate the relationship between types of support (e.g., instrumental, emotional) and indices of well-being. Studies of this type are the most common in investigations of social support and are also represented in the mentoring literature. For instance, Alleman's (1985) study that related subscale scores of the ALDQ to degree of beneficial career influence concerned the functional content of relationships. Another example of such a study is the one reported by Bowen (1985) that related perceptions of career progress to the provision of mentor functions coded according to Kram's (1985) distinctions. House and Kahn (1985) noted that assessment of the functional content of social support should address (a) quantity of support, (b) source of support, and (c) type of support. Each of these three factors also needs to be addressed explicitly in order to 34 more fully understand the phenomenon of mentoring. In relation to mentoring behaviours, quantity refers to the degree to which individuals are recipients of such behaviours. This factor has been addressed by a number of studies (e.g., Alleman, 1985; Bowen, 1985) but usually is ignored by the studies that primarily are concerned with the existence of mentoring relationships (e.g., Henderson, 1985; Reich, 1985, 1986; Roche, 1979). The source of mentoring behaviours, which refers to who provides mentoring behaviours, empirically has been the most neglected aspect of mentoring. It is, however, central in any discussion of multiple sources of mentoring behaviours. Although writers have recognized the fact that different individuals can provide mentoring behaviours, this notion has not been incorporated into studies investigating the mentoring phenomenon. Finally, type of mentoring behaviour refers to the mentoring functions discussed under the two broad categories of career functions and psychosocial functions. Mentoring functions have been described by a variety of writers (e.g., Kram, 1985; Missirian, 1980; Reich, 1985, 1986). The framework for the study of social support provided by House and Kahn (1985) is applicable to developing an understanding of how mentoring behaviours can be derived from multiple sources. All three factors-quantity, source, and type-need to be addressed simultaneously. Because this has not been undertaken in the mentoring literature, a brief review of studies relating to social support in work settings will be provided. This review will suggest the issues that need to be addressed in a study of mentoring behaviours associated with multiple sources. Ford (1985), in a study conceptually related to mentoring behaviours, investigated the relationship between structural, informational, and emotional support and employee work outcomes as measured by the Job Description index (JDI) (Smith, Kendall, & Hulin, 1969). The three types of support were assessed using a 27-item inventory developed by Edwards (1980) based on the work of Kanter (1977). The three types of support were conceptualized as follows: Structural support is provided by an influential and powerful official who, through advocacy and manipulation of the organizational system, helps an individual advance within the organization structure. Informational support is provided to an individual through the sharing of information about formal and informal system norms, mores, and procedures, and by communicating resource information and general system knowledge. Emotional support is 35 provided when an individual has someone who shows concern that his/her goals are reached, who is emotionally reinforcing, and who can be trusted and depended upon. (Ford, 1985, p. 6) Table 1 depicts the relationship between social support as conceptualized by Ford (1985) and the mentoring functions discussed in this chapter. It is clear that individuals who give social support would also be providing a number of mentoring functions. Ford did not assess the sources of such support; respondents merely were asked to indicate the "...extent to which they had someone in the work place who provided the type of supportive behavior indicated" (p. 11). Thus, this study investigated the frequency with which employees were provided with the types of support by anyone in the work place. This differs from the methodology used in most mentoring studies that focus on the behaviours provided by one particular individual, but it still does not address multiple sources of behaviours explicitly. Two sets of findings from Ford's (1985) study are of interest. First, the three support scales were highly intercorrelated. Given that structural and informational support seem similar to the career functions, while emotional support seems similar to the psychosocial functions, it may be that the conceptual distinction between career and psychosocial functions will not be demonstrated empirically. Edwards (1980) also reported high intercorrelations among the three scales in a different sample. The second finding of interest is that all three support scales correlated significantly with all of the jDI subscales (satisfaction with work, promotion, pay, supervision, and co-workers). This finding suggests that the provision of mentoring behaviours relates positively to employee outcomes when this provision is assessed globally (i.e., coming from anyone) rather than specifically, as in most mentoring studies. It would be of interest to know the magnitude of the correlations between the JDI scales and the social support received from the one individual who provided the majority or greatest degree of it. This would allow a comparison of the effectiveness of behaviours derived from one source with the effectivenss of behaviours derived from various sources. However, as mentioned above, Ford (1985) did not incorporate source of social support into the research design. A study related to social support that did incorporate source of support was conducted by Abdel-Halim (1982). Of primary interest for present purposes is the relationship between both work-group and leader support and intrinsic job satisfaction as measured by the Job Diagnostic Survey (Hackman & Oldham, 1975). This study was also concerned with the frequency of social support. Work-group support was assessed with Table 1 Relationship between Social S Social Support (Ford, 1985) Structural Support Informational Support Emotional Support and Mentoring Functions Mentoring Functions Sponsoring Exposure and V i s i b i l i t y Teaching the Job Teaching the Informal Syst Encouragement Personal Counselling Note. Protection by Prevention, Protection by Absorption, and Role Modeling were not c l a s s i f i e d . 37 an 8-item scale asking respondents to indicate the extent to which aspects of psychological/emotional as well as instrumental/active support were present in the work place. Leader support was assessed with the Consideration scale of the Leader Behavior Description Questionnaire-Form XII (LBDQ) (Stodgill, 1963). In this sample of managerial personnel, intrinsic job satisfaction was significantly related to both work-group support and leader support; the correlation with work-group support was somewhat higher than the correlation with leader support. Further, the fact that the two support scales were significantly related suggests that the correlation between total support (i.e., work-group and leader support combined) and intrinsic job satisfaction might be even higher than the correlations obtained with the separate support scales. This study demonstrates that support from various sources is related to positive employee outcomes and that total support is probably more highly related to such outcomes than support from individual sources. A methodological note is in order here. Perhaps the relationship between support and satisfaction reported by Abdel-Halim (1982) was deflated because the degree of leader support provided by the work-group and the degree of work-group support provided by the leader were not assessed. For example, items on the Consideration scale of the LBDQ ask about the degree to which the supervisor engages in a variety of behaviours. Bales (1958) suggested that behaviours not provided by one member of a group (i.e., the leader) might be provided by another member; it was suggested that it was the combination of these behaviours that related to group effectiveness. In relation to Abdel-Halim's (1982) study, it is possible that the behaviours not provided by a leader low on consideration would be provided by one or more of the work-group members. Then the degree of leader support reported by Abdel-Halim would be an underestimate of the type of support the leader support measure was designed to assess. In other words, some work-group members may have provided the type of support assessed only in relation to the leader. This could result in lower correlations between satisfaction and support than might be found in a study that measured the amount of each type of support received from each of the sources under consideration. A study that comes closer to this ideal was reported by House (1981). Using a sample of 1,809 hourly employees of a manufacturing plant, House (1981) investigated the relationship between social support received from immediate supervisors, co-workers, spouses, and friends and relatives, on the one hand, and a number of work-related outcomes, on the other. An attempt was made to distinguish between the emotional and 38 instrumental support provided by each of these sources, but this distinction was not retained because of the high correlation between the two types of support. This result supports the findings of Edwards (1980) and Ford (1985) that conceptually distinct types of support may not be found to be distinct empirically. Social support was assessed using a 6-item scale (House, 1981). The amount of social support received was assessed with all six items for immediate supervisors, three of the items for co-workers, and two of the items for both spouses and friends and relatives. An index of total support was derived by summing the support received from each of the four sources. With the exception of the relationship between support from friends and relatives and support from spouse, there was only a moderate relationship among the degrees of support received from the various sources. House (1981) interpreted this to mean that employees can discriminate among people in reporting their supportiveness. ln other words, perceived support from one source does not necessarily imply perceived support from another source. This interpretation requires further investigation, however, because the level of relationship reported among the sources of support is highly related to the percentage of support items the sources had in common. This suggests that some of the correlations among sources of support may have been artifactually low. In order to adequately assess the relationship among sources of support, the same items should be used for all sources. Of primary interest for present purposes are the relationships in House's (1981) research between the amount of support received from the various sources (including total support) and the outcome measures of job satisfaction, occupational self-esteem, job-nonjob conflict, role conflict, responsibility, quality concern, and workload. All correlations relating supervisory support, co-worker support, and total support to the outcome measures were significant and in the predicted direction. Support from spouse and support from friends and relatives were significantly related only to job satisfaction, but these correlations achieved significance only because of the large sample size. These findings led House (1981) to conclude that nonwork sources have little or no effect on work-related outcome measures. However, one particularly puzzling aspect of the correlations reported was that, in absolute magnitude, the correlations of the outcome measures with supervisory support were always higher than the corresponding correlations with total support. This suggests that positive outcomes generally may be due to support from a specific source rather than to support from a variety of sources. This possibility needs further investigation because the findings reported by House (1981) may be due to 39 the possibility that the total support measure was less reliable than the supervisory support measure; recall the relatively low correlations mentioned earlier among the support measures from different sources. Perhaps a more appropriate test of the overriding significance of a single source would involve a multiple regression approach in which the various sources are entered in an hierarchical fashion. One other methodological aspect of House's (1981) study deserves comment. It was noted earlier that this study came closer to a consideration of the amount of each type of support received from each of the sources than many other studies. Unfortunately, the two types of support assessed (instrumental and emotional) were too highly correlated to be discussed separately. But perhaps of more importance is the fact that two of the sources considered by House (co-workers and friends and relatives) themselves can consist of multiple sources. Thus, when employees comment on the degree of support received from these sources it is unclear how many individuals contribute to the reported amount of support. For instance, one employee may have one person in the work group who provides high levels of support, while another employee, who is a member of a cohesive work group, might receive high levels of support from a variety of co-workers. By asking about the level of support received from members of a group in general, as House (1981) did for co-workers and friends and relatives, it may not be possible to distinguish between such employees; both report high levels of support. Such a distinction would be important to make if the pattern of support as well as the level of support could be shown to be related to employee outcomes. This suggests that if one is interested in multiple sources of support or mentoring, consideration must be given to the amount of a certain behaviour people receive from individuals rather than from groups of individuals. Such an approach was used by Norbeck, Lindsey, and Carrieri (1981) in the development of an instrument to measure three types of social support-affect, affirmation, and aid. Respondents, who were nursing students, were asked to list all the poeple who provided personal support or who were currently important to them. The average number of people listed was 13. Respondents were asked to answer each of nine questions for each of the people they had listed. The six items that related to type of support were assessed by asking about the degree to which respondents were recipients of the type of support in question. Scores were obtained for each item by summing the degree of support received from all people listed. Correlations among the support items as well as between the support items and outcomes then were calculated. Unfortunately, the results 40 of this study cannot be interpreted unambiguously because, as noted by House and Kahn (1985), by summing the responses across all of the people listed, the total support received is largely a function of the number of people listed. This method of summing across all of the people listed results in inflated correlations among similarly scored measures; the correlations among the six support items used by Norbeck et al. (1981) ranged from .72 to .98. The magnitude of these correlations led Norbeck et al. (1981) to conclude, perhaps erroneously, that the three types of support under consideration could not be distinguished empirically. Another problem with summing support items across people listed is that such sums are likely to be unrelated to other measures (e.g., outcomes) that are not obtained by summing across people. For instance, if support is largely a function of the number of people listed, one would expect a correlation between support and a variable such as job satisfaction only if those people who are more satisfied also tend to list more people. While this may be the case, it is likely that a larger correlation would obtain if the support measure were strictly a measure of the quantity of support obtained, either independent of the number of people who provide that support or controlled for the number of people who provide that support. The latter approach, controlling for the number of people, probably would be the preferred one if interest lies in understanding the importance of the pattern of behaviours received from a variety of sources. Supportive Behaviours and Supportive Functions Throughout this chapter much reference has been made to mentoring behaviours and mentoring functions. From this point on, mentoring behaviours will be referred to as supportive behaviours and mentoring functions will be referred to as supportive functions. This semantic switch was made for a number of reasons. First, as the various definitions of the term 'mentor' imply, there is not widespread agreement on the exact meaning of the term. This fact, taken in conjunction with the literature that indicates that the behaviours and functions generally associated with mentors also can be derived from other individuals (e.g., co-workers), suggests that a different terminology should be used. By referring to supportive behaviours and supportive functions, these behaviours and functions are not immediately associated with any particular source (i.e., a mentor). It is hoped that the new terms carry less semantic and emotional freight. A final reason for changing the terminology was suggested by the social support literature just reviewed. Social support, as noted earlier, has been defined as "...an 41 exchange of resources between at least two individuals perceived by the provider or the recipient to be intended to enhance the well-being of the recipient" (Shumaker & Brownell, 1984, p. 13). Since mentoring behaviours are often provided to enhance employee well-being in relation to career and personal development (Kram, 1985), mentoring behaviours can be viewed as particular forms of social support. This is why the term 'supportive' was chosen. The way in which social support has been conceptualized in studies conducted in work settings (e.g., Abdel-Halim, 1982; Edwards, 1980; Ford, 1985), also suggests that mentoring behaviours can be viewed as supportive behaviours. It was shown earlier, for instance, that most of the mentoring functions reviewed can easily be associated with structural, informational, or emotional support (Ford, 1985). Given such conceptual overlap between the social support and mentoring literatures, it appears that people who provide mentoring functions and behaviours also are providing social support. In summary, the behaviours and functions usually associated with mentors will be referred to as supportive behaviours and supportive functions. Such terminology avoids the semantic problems associated with the term 'mentor,' recognizes that these behaviours and functions can be provided by a variety of people who are not necessarily mentors, and captures the conceptual overlap in the social support and mentoring literatures. The Importance of the Hazards, the Patterning, and the Quality of Supportive Behaviours Based on a review of studies related to social support it can be concluded that an adequate conceptualization of the mentoring phenomenon should include simultaneous consideration of the quantity, source, and type of supportive behaviours to which an individual is exposed. The mentoring literature, to date, has not considered these three factors simultaneously. Going beyond the social support literature, three additional factors suggest themselves as being important in understanding the mentoring phenomenon. These are: (a) the hazards associated with mentor-protege relationships, (b) the patterning of supportive behaviours, and (c) the quality of supportive behaviours. Each of these three factors will now be discussed. The hazards and disadvantages of mentor relationships. Henderson (1985) noted that mentors, proteges, and other organizational members often view mentor-protege relationships as less than ideal. Although the hazards or disadvantages of such relationships are generally thought to be outweighed by the benefits that accrue, the potential negative aspects of mentoring have been discussed by a variety of writers. For 42 example, Reich (1985) found that 43 percent of the mostly male sample that reported having had a mentor noted few or no disadvantages due to their mentoring relationships. This implies that over half of the respondents did find some significant drawbacks to their relationships. The most common disadvantage (as reported by over 30 percent of the respondents) was the feeling on the part of proteges that others identified them too closely with their mentors. Further, 25 percent felt that it was disadvantageous that their peers considered them to be their 'mentor's person.' Other disadvantages cited by the proteges in Reich's (1985) study included being kept from better jobs (12 percent), being given too much protection (9 percent), and being shielded from the results of their mistakes (6 percent). Similar findings were reported by Reich (1986) for a study with exclusively female respondents, although the incidence of few or no reported disadvantages was slightly higher--50 percent. The two studies by Reich (1985, 1986) are noteworthy because most sources merely describe hazards or disadvantages without providing figures on incidence. One of the most frequently mentioned hazards (at least in mixed-gender pairings) is that mentors and proteges can become sexually involved or that others in the organization will suspect sexual involvement (e.g., Berry, 1983; Fitt & Newton, 1981; Halatin & Knotts, 1982; Henderson, 1985; Kram, 1985; Zey, 1984). The situation experienced by Mary Cunningham at the Bendix Corporation illustrates this problem nicely. In relation to the problem of potential sexual involvement, Bovven (1985) noted that over 20 percent of individuals in cross-gender mentoring relationships indicated that jealousy of spouse was or had been a problem. Spousal jealousy is not the only kind of jealousy reported to arise due to presence of mentoring relationships. Various writers have noted that other employees, particularly co-workers, may be jealous of the recognition, assistance, or rewards that proteges receive (e.g., Bowen, 1985; Halatin & Knotts, 1982). Such jealousy may lead to less than ideal peer or co-worker relations and may, as noted by Halatin and Knotts (1982), result in retaliation of various kinds. Bowen (1985) found that co-worker envy of the perceived preferential treatment afforded proteges was mentioned by over 30 percent of the mentor-protege pairs interviewed. Another problem frequently cited concerns proteges becoming overly dependent on their mentors (e.g., Berry, 1983; Halatin & Knotts, 1982; Henderson, 1985). Such dependency may lead to lower levels of confidence and self-esteem (Berry, 1983). 43 Henderson (1985) noted that the organization also can suffer from relationships characterized by high degrees of dependence. This may occur if mentors attempt to mold their proteges in their own images. Related to the notion of dependency, proteges may be negatively affected should their mentors either fall out of favour with others (Berry, 1983; Halatin & Knotts, 1982) or leave the organization altogether. Clearly, the problems and hazards associated with mentor relationships are less than desirable because of their potential negative effects. One way that individuals might avoid some of the hazards associated with mentoring is to reduce the number of supportive behaviours with which they are provided. Such an approach presents a dilemma, however. If higher levels of supportive behaviours are associated with more positive employee outcomes (as is usually assumed), then a reduction in supportive behaviours, while reducing the hazards, would tend to have deleterious effects on these employee outcomes. The problem that presents itself is how to reduce the hazards associated with mentor-protege relationships while, at the same time, retaining the level of supportive behaviours provided. The answer to this problem lies in consideration of the patterning of supportive behaviours to which an individual is exposed. The patterning of supportive behaviours across multiple sources. Instead of reducing the level of supportive behaviours, it may be possible to avoid some of the hazards or disadvantages associated with mentor-protege relationships by ensuring that one becomes the recipient of supportive behaviours from a variety of sources rather than from just one. This does not mean that the number of supportive behaviours need to be increased. Rather, for a given level of behaviours, the hazards may be reduced to the extent that the individual receives these behaviours from multiple sources. This would prevent an employee from becoming identified too closely with a single person; peers would be less likely to consider the employee to be the 'mentor's person.' There would also be less risk of becoming sexually involved or overly dependent on a single individual. The employee would be less likely to experience negative effects if one person fell out of favour with others or left the organization. In relation to the hazard of being kept from better jobs, this would be less likely for an individual who is occasionally Sponsored by a variety of people than for an individual who is Sponsored repeatedly by a single person. Different people who Sponsor an individual may have more credibility in the eyes of other organizational members than would a single individual. In summary, it is possible that individuals might be able to avoid some of the hazards associated with mentor relationships by relying on a variety of people rather than just one or two for 44 the provision of supportive behaviours. This suggests that individuals who are the recipients of a given level of supportive behaviours from a variety of sources may be more fortunate or 'safer' than individuals who are the recipients of this same level of supportive behaviours from only one source. Few writers in the area of social support have addressed this notion. Reis (1984), in a review paper, argued forcibly that well-being is more related to the social support derived from close confidants or friends than to the social support derived from diffuse contacts such as co-workers and neighbours. Such an argument makes good sense because the close contacts likely would provide more support than would any one of the more diffuse contacts. However, Reis' (1984) argument does not address adequately the distinction between single and multiple sources of support (or supportive behaviours), ln order to compare the relative effectiveness of single and multiple sources of support, care must be taken to account for absolute levels of support received. For each level of support, it would then be possible to compare individuals who received the majority' of their support from a single source to individuals who received the support from multiple sources. Such a procedure does not appear to have been employed to date. It is important to consider that the social support and mentoring literatures may lead to different conclusions concerning the benefits of the way in which support is distributed across sources. In general, the social support literature suggests that an individual can never be the recipient of too much social support from anyone. As such, any additional support received from any source should provide additional benefit to the individual. This suggests that support received from one individual should be as beneficial as support received from a variety of individuals when level of support is held constant. When the focus is on mentor-protege relationships, such a conclusion may not be warranted. If, as suggested, the hazards of mentor relationships might be avoided by ensuring that supportive behaviours are received from multiple sources, then consideration of the source of any additional behaviours is important in determining the potential benefit to the individual. If the behaviours are received from the person already providing the largest number of behaviours, the hazards which are assumed to be related to less positive outcomes are more likely to arise. On the other hand, if the additional behaviours reside in a number of sources, the hazards associated with mentor-protege relationships would be less likely to arise. Thus, behaviours received from multiple sources are more likely to result in positive outcomes than are behaviours received from a single 45 source when level of behaviours is held constant. This is not what the social support literature would suggest. The quality of supportive behaviours. Throughout this paper the term quantity has been used to refer to the degree to which individuals or groups are perceived to provide a certain type of supportive behaviour. Implicit. in the social support literature is the idea that as the quantity of social support increases, individuals can be expected to exhibit more positive outcomes (i.e., much of the social support literature posits a linear relationship between quantity of support and well-being). In other words, the more social support, the better. On the other hand, the review of material related to the hazards of mentor relationships suggests that more supportive behaviours may not necessarily be better. For this reason a distinction must be made between the quantity and quality of supportive behaviours received. The problem of how to conceptualize the quality of supportive behaviours arises immediately. Although criteria such as timeliness and appropriateness have been mentioned (Phillips-Jones, 1982), one is left with the questions of timely in relation to what, and appropriate for what? Given that supportive behaviours have been purported to be related to enhanced career development, quality, in this study, was conceptualized in terms of the effect that the provision of supportive behaviours has on an employee's career development. Thus, if a certain quantity of supportive behaviour, be it high or low, has a harmful effect on an employee's career development, then that quantity will be of low quality. If, on the other hand, a certain quantity of supportive behaviour has a beneficial effect on an employee's career development, then that quantity will be of high quality. To the extent that low and high quantities of supportive behaviours are perceived as ineffective and effective for career development, respectively, quantity and quality will be related to outcomes similarly. However, to the extent that high quantities are perceived as ineffective, possibly because of the disadvantages they bring with them, quality may be a better indicator of employee outcomes than quantity of supportive behaviours. This also would be the case if a low quantity of supportive behaviours of high quality resulted in high levels of career development. The problem of determining the effectiveness of mentoring deserves further comment. Although supportive behaviours are assumed to result in the career success of individuals, the concept of career success has not been adequately developed in the mentoring literature to date. For instance, while some of the outcome measures that have 46 been used in previous studies concern employees' current positions (e.g., salary, pleasure derived from work), others are more concerned with the rate of events over a period of time (e.g., rate of career progression). Given that mentoring is assumed to be related to employee development, it is surprising that few, if any, researchers distinguish between these two types of outcomes. The concept of career success is obviously important for the adequate evaluation of the effectiveness of supportive behaviours. For this reason, it is important to delineate what is meant by career success. This will be discussed next. What is Career Success? Two relatively recent meta-analytic studies on the relationship between academic and occupational achievement (Cohen, 1984; Samson, Graue, Weinstein, & Walberg, 1984) provide clear evidence that career success can be defined in a variety of ways. Cohen (1984) reviewed 108 studies that reported one or more correlations between college grades and some index of career success. The most common measure of success was job performance ratings (usually supervisory as opposed to self and peer). Other measures, in order of frequency of use, were income, attainment of a graduate degree, promotions, satisfaction with career success, and eminence. These findings are similar to those of Samson et al. (1984). Of the 209 correlations between academic and occupational achievement, over one-half used job performance ratings as the outcome measure. The next most frequently used outcomes were, in order, salary, career position, job satisfaction, pupil achievement, job position or title, and salary gain. Three aspects of the measures most commonly used to assess career success are of interest. First, most of the measures involve assessment of employees' current career or job situations and do not incorporate the notion of progression or development. It can be argued that the rate at which certain aspects of employees' careers transpire is important in the consideration of the effectiveness of a developmental phenomenon such as mentoring. Second, most of the measures used do not involve the perceptions of the individuals involved in the studies. This is a major shortcoming because externally visible measures of attainment (e.g., income, promotions) may not be the ones used by individuals to judge their own success. For instance, Schein (1978) suggested that people who hold different career anchors (e.g., managerial competence, security and stability, 47 autonomy and independence) may have different criteria for success by which they will evaluate themselves. Thus, there likely are individual differences concerning what constitutes career success. These may be reflected more in measures that relate to employee perceptions. The third aspect of interest concerning the measures used to assess career success is that they do not incorporate notions of career stage. This is an important consideration because what constitutes career success at one stage in a person's career may not be indicative of career success at a different stage. in summary, an adequate conceptualization of career success should incorporate rates of outcomes, present levels of outcomes, individual perceptions of outcomes, and outcomes as they relate to career stage. Most of the outcomes that were considered in this study were suggested by the literature related to concerns expressed at different career stages. Career stages. A variety of authors have suggested that the concerns expressed by individuals in relation to their jobs will differ depending on the career or job stages at which these individuals find themselves (Could & Hawkins, 1978; Katz, 1980; Mount, 1984; Slocum & Cron, 1985; Stumpf & Rabinowitz, 1981). Three separate stages have been of major concern in the literature: (a) establishment, (b) advancement, and (c) maintenance (Adler & Aranya, 1984; Mount, 1984; Stumpf & Rabinowitz, 1981). The establishment stage has been defined in terms of age, years in a profession, and years of organizational tenure. Individuals under the age of 30 are generally assumed to be in the establishment stage (Adler & Aranya, 1984; Slocum & Cron, 1985). When professional or organizational tenure is used to define stages, people with up to two years tenure are deemed to be in the establishment stage (Could & Hawkins, 1978; Mount, 1984; Stumpf & Rabinowitz, 1981). Despite the discrepancy in how long the . establishment stage may last depending on whether one defines it on the basis of age or tenure, most authors agree on the concerns expressed by individuals during this stage. The primary concern appears to be the development of competence (Gould & Hawkins, 1978; Katz, 1980; Mount, 1984; Slocum & Cron, 1985; Stumpf & Rabinowitz, 1981). In general, individuals concern themselves with developing and learning the technical skills required to perform their jobs. During the establishment stage individuals are also concerned with peer acceptance (Gould & Hawkins, 1978; Katz, 1980; Mount, 1984; Stumpf & Rabinowitz, 1981) and with learning about informal organizational requirements 48 (Katz, 1980). These concerns suggest that individuals in the establishment stage may view particular work-related outcomes as indices of career success. Individuals may consider themselves successful if they are satisfied with their work and if they are accepted by their peers and co-workers. These outcomes have been viewed as either desirable outcomes or as indices of career success by a variety of writers (e.g., Harpaz, 1985; Mossholder, Bedeian, Touliatos, & Barkman, 1985; Tsui & Gutek, 1984). Further, given that the development of competence to do one's job is important at this stage, people may view overcoming role ambiguity as an aspect of success (Feldman, 1976; Katz, 1980). The importance of these outcomes to individuals during the establishment stage suggests that individuals who receive a variety of supportive behaviours will exhibit more positive outcomes than will individuals who do not receive these behaviours. Related to satisfaction with the work itself may be such functions as Teaching the Job, Protection by Prevention, Protection by Absorption, and, perhaps, Role Modeling. These functions should provide employees with an atmosphere that would allow the development of competence. Other functions might be related to acceptance by co-workers. Such acceptance may be higher for individuals who are provided with the functions of Encouragement and Teaching the Informal System by their co-workers. Finally, role ambiguity likely would be reduced by the degree to which organizational members Teach the Job, Teach the Informal System, and provide Personal Counselling. In summary, during the establishment stage major employee concerns may be with developing competence, gaining acceptance, and learning about informal organizational requirements. Employees who overcome these concerns may be more satisfied with their work, be more accepted by their co-workers, and experience lower levels of role ambiguity. Following the establishment stage, individuals enter the advancement stage, individuals between the ages of 31 and 44 (Slocum & Cron, 1985) or 45 (Adler & Aranya, 1984) are generally assumed to be in the advancement stage. When professional or organizational tenure is used to define stages, people with between two and ten years tenure are deemed to be in the advancement stage (Gould & Hawkins, 1978; Mount, 1984; Stumpf & Rabinowitz, 1981). Having developed competence and knowledge of the informal system in the establishment stage, some of the major concerns of individuals become achievement and accomplishment (Gould & Hawkins, 1978; Katz, 1980; 49 Mount, 1984; Stumpf & Rabinowitz, 1981). Such achievement and accomplishment is rewarded organizationally by promotions to positions with higher authority and responsibility. Could and Hawkins (1978) argued that satisfaction with promotions becomes more important than satisfaction with the work itself and acceptance by co-workers. At the advancement stage many of the most frequently identified desirable work outcomes and criteria of career success likely become important. These include satisfaction with promotions and promotional opportunities (Harpaz, 1985; Mossholder et al., 1985), promotion rate (Tsui & Gutek, 1984), salary (Harpaz, 1985; Mossholder et al., 1985), salary increases (Tsui & Gutek, 1984), and the extent to which role conflicts between career, self, and family have been resolved (Feldman, 1976). Thus, by the end of the advancement stage, individuals may view their career success largely in terms of the organizational position they have achieved. This might be reflected by satisfaction with absolute levels of salary and number of promotions, as well as by rates of promotion and salary increase. As in the establishment stage, individuals who are the recipients of supportive behaviours in the advancement stage may exhibit more positive outcomes than might individuals who do not receive these behaviours. Such functions as Exposure and Visibility and Sponsoring may help individuals achieve positions with greater authority, responsibility, and salary. Personal Counselling and Encouragement may help individuals resolve role conflicts. In summary, during the advancement stage, employees have different concerns than those experienced in the establishment stage. Because of this, individuals might be expected to shift their definitions of career success. In particular, success in the advancement stage might be defined in terms of organizational position and the concomitant benefits such as salary. The final stage to be discussed is the maintenance stage. Individuals beyond the age of 45 or with ten or more years of tenure are generally assumed to be in the maintenance stage (Adler & Aranya, 1984; Gould & Hawkins, 1978; Mount, 1984; Slocum & Cron, 1985; Stumpf & Rabinowitz, 1981). Having advanced to higher levels within the organization, individuals become less concerned with competition and more concerned with developing peer and professional relationships and with helping other people develop (Could & Hawkins, 1978; Katz, 1980; Mount, 1984; Stumpf & Rabinowitz, 1981). Thus, as was the case in the establishment stage, acceptance by co-workers becomes an important 50 outcome. Without such acceptance, it seems unlikely that developmental efforts would be welcomed. Another outcome that appears to be relevant for individuals in the maintenance stage is organizational commitment. It is difficult to imagine that people without such commitment would engage in many helping and developmental activities on the behalf of others. As noted by Mowday, Steers, and Porter (1979), commitment "...involves an active relationship with the organization such that individuals are willing to give something of themselves in order to contribute to the organization's well being" (p. 226). Clearly, taking an interest in the development of other organizational members reflects such an organizational relationship. Also, Rhodes' (1983) review of age differences in work-related outcomes suggested that organizational commitment tended to increase with age and organizational tenure. It appears, then, that organizational commitment is an important outcome for individuals in the maintenance stage. While commitment can be expected to be present at all career stages, it should be at its highest level during the maintenance stage. Most of the available literature suggests that individuals are less likely to be recipients of supportive behaviours during the maintenance stage than in the establishment and advancement stages. Individuals are most likely to be recipients of supportive behaviours at early ages, near the beginning of their careers, or near the beginning of their tenure with organizations (Bova & Phillips, 1981; Henderson, 1985; Kram, 1983; Levinson et al., 1978; Missirian, 1980; Roche, 1979). Henderson (1985) and Roche (1979) found that most mentor-protege relationships began during the first five years of proteges' careers. Similarly, Bova and Phillips (1981) found that the majority of such relationships began in early adulthood or after entry into the work force. Levinson et al. (1978) suggested that men over 40 years of age rarely have mentors. Other studies (Kram, 1983; Missirian, 1980) suggest that the typical mentor-protege relationship lasts approximately ten years. Considering the time of initiation and duration of mentor-protege relationships, it is likely that individuals will receive the majority of supportive behaviours in the establishment and advancement stages. Indeed, many writers (e.g., Hall, 1976; Reich, 1985, 1986) have noted that by the time individuals are over 40 years of age or in late career, they may become mentors to more junior people. Given the above considerations concerning the likely timing of supportive behaviours, the indices of career success appropriate to the maintenance stage (e.g., organizational commitment and acceptance by co-workers) may not be as highly related to supportive behaviours received 51 as the indices of career success appropriate to the other two stages. In summary, in addition to security and stability, during the maintenance stage employee concerns appear to involve mostly social needs. Because of the shift in concerns from the advancement stage, notions of what constitutes career success also may change. However, indices of career success in the maintenance stage should be less likely to be highly related to the level of supportive behaviours received at an earlier career stage. Outcome variables. The above review of the major concerns expressed by individuals at different career stages suggested a variety of measures of career success appropriate to these stages. It is important to remember, however, that the different indices may be appropriate for some people at one stage and for different people at another stage. Situational and individual differences may be important in how success is defined at various times in one's life. For this reason, outcome variables not suggested by the above review also were included for study. Based mostly on the career stage literature, the following outcomes suggested themselves as being important in a study designed to assess the effectiveness of supportive behaviours: a) satisfaction with progression, b) rate of promotions, c) rate of salary increase, d) skill development, e) job satisfaction, f) role conflict, g) role ambiguity, h) acceptance by co-workers, and i) organizational commitment. As noted previously, outcomes related to the present situation in which employees find themselves have received vastly greater amounts of attention in previous studies than have outcomes related to the more developmental aspects of employees' careers. This shortcoming is overcome in the outcomes considered here. Specifically, the first four outcomes, which will be referred to as career-related outcomes, assess development in various aspects of employees' careers; all involve change over time. The remaining five outcomes, which will be referred to as job-related outcomes, pertain more to the current 52 situation in which employees find themselves. This completes discussion of the outcomes considered to be indicators of career success in this study. With few exceptions (e.g., likelihood of having a career plan), the outcomes assessed in mentoring studies to date are represented by the outcomes listed above. The current study should represent an improvement over these previous studies because outcomes were assessed more comprehensively. Summary In recent years much has been written suggesting that employees who enter into mentor-protege relationships with more senior or experienced organizational members can expect enhanced levels of career and personal development compared to employees who do not enter into such relationships. Unfortunately, few empirical studies have addressed this issue. The studies that have been done suffer from one or more serious shortcomings. First, different studies have used different definitions of the term 'mentor.' This makes it difficult to compare the incidence of mentoring across studies. Second, the employee outcomes that are purportedly related to mentoring have been poorly conceptualized and operationalized. Third, mentoring is usually viewed as an all-or-none phenomenon. Such a view does not entertain the possibility that individuals who report not having had a mentor might also have been the recipients of behaviours associated with mentors, albeit from a variety of organizational members (e.g., co-workers as well as supervisors). The purpose of the present study was to redress some of these shortcomings. Rather than relying on a single definition of what a mentor is, this study assessed the degree to which employees were the recipients of supportive behaviours. These behaviours were identified through a review of the literature pertaining to the functions and roles served by mentors. This study departed somewhat from traditional conceptualizations of mentoring in that it entertained the possibility that it is the sum total of supportive behaviours received from any source that is important for career development; previous work on mentoring suggests that the supportive behaviours may have to be derived from one source (the mentor) in order to be effective. 53 Chapter 3  Research Questions and Hypotheses This chapter considers a number of research questions and hypotheses suggested by the review of the mentoring and work-related social support literatures. In general, the research questions are of a more exploratory nature than are the hypotheses. Research Question 1: What is the structure of the supportive functions? Based on a literature review concerning mentor roles, functions, and behaviours, nine supportive functions were identified. Although all of these functions were discussed, at least tangentially, by Kram (1985), the way in which the functions are conceptualized in this study was dictated by the similarities and overlap found in the work of numerous writers, Kram included. Thus, although there are similarities between the functions delineated by Kram (1985) and those discussed here, there is not a one-to-one correspondence between the two sets of functions. Despite this lack of correspondence, all of the functions under consideration in this study can be viewed logically as one of the two types of functions (career and psychosocial) discussed by Kram (1985). In particular, the functions of Sponsoring, Exposure and Visibility, Teaching the Job, Teaching the Informal System, and the two types of Protection can be viewed as career functions. The psychosocial functions consist of Role Modeling, Encouragement, and Personal Counselling. While the distinction between career and psychosocial functions is conceptually useful, little work has addressed the issue of whether this distinction also can be made empirically. Schockett and Haring-Hidore (1985) found that the career and psychosocial functions loaded on separate oblique factors. Unfortunately, these authors did not report factor loadings or the correlation between the two factors. Also, because their analysis was based on written descriptions of the functions, their study cannot be viewed as an adequate evaluation of the distinction between career and psychosocial functions. Empirical validation for the distinction between career and psychosocial functions would be demonstrated if employees who receive relatively high (or low) levels of behaviours associated with one function do not necessarily receive high (or low) levels of behaviours associated with the other functions. Such a pattern of behaviours will obtain if the providers of the various functions are selective in relation to which functions they provide employees. While this may be the case at any specific point in time, it is iikely 54 that most functions will be provided (or not provided) when an extended period of time is considered. For instance, Kram (1983) noted that, during the cultivation phase of developmental (i.e., mentoring) relationships, the number of career and psychosocial functions increased to a maximum. Such a pattern makes intuitive sense in that it is difficult to imagine that the provider of Sponsoring and Exposure and Visibility (i.e., career functions) would not also provide an employee with a certain amount of Encouragement and Personal Counselling (i.e., psychosocial functions) in helping a person attain the opportunities at which the career functions are directed. At the other extreme, it seems unlikely that an individual would provide an employee with Protection and engage in Teaching the Informal System in the absence of functions that suggest some degree of interpersonal closeness (i.e., Encouragement and Personal Counselling). The foregoing suggests that the career and psychosocial functions, while conceptually distinct, may not be found to be so empirically. Research Question 2: What is the structure of the employee outcomes? Employee outcomes that have been discussed in relation to mentoring were previously described as either career-related outcomes or job-related outcomes, included in the former group are rate of salary increase, rate of promotions, skill development, and satisfaction with progression. The job-related outcomes to be considered are job satisfaction, role ambiguity, role conflict, acceptance by co-workers, and organizational commitment. There is a temporal distinction between the two types of outcomes. The career-related outcomes all involve change over time and are developmental in nature. The job-related outcomes, in contrast, pertain more to employees' current situations. Although there may be variance in how quickly the latter group of outcomes might be expected to change given a change in an employee's position, they are, none-the-less, more related to a specific position than are the career-related outcomes. Thus, for example, an employee who has recently been moved into a rather disliked and/or difficult job might exhibit lower levels of job satisfaction and higher levels of role ambiguity and conflict than would have been the case immediately prior to the move. However, depending on the type of move (e.g., a lateral one), changes in the career-related outcomes of rate of promotions, rate of salary increase, and skill development would remain rather stable because these outcomes reflect experiences the employee has been exposed to in a variety of organizational positions, rather than just the current one. Given that mentoring is largely a developmental phenomenon, the career-related outcomes, which are also 55 concerned with development, should be more highly related to the supportive behaviours to which an employee has been exposed than should the job-related outcomes. However, before differential relationships between the two types of outcomes and supportive behaviours can be assessed, the empirical distinction between the two types of outcomes must be present. Such a distinction would be found if the outcomes within each of the two categories are more highly related to each other than to the outcomes in the other category. This pattern of relationships would obtain if individuals with high (or low) levels of outcomes within one category do not necessarily exhibit high (or low) levels of outcomes within the other category. The conceptual distinction between the two types of outcomes would not receive empirical support, however, if high (or low) levels of career-related outcomes go hand-in-hand with high (or low) levels of job-related outcomes. Both of the above possibilities must be entertained. On the positive side, employees who exhibit high levels of skill development might be promoted into higher paying, more satisfying jobs in which they are accepted by co-workers, experience low levels of role ambiguity and role conflict, and develop high levels of organizational commitment. A more negative scenario is one in which employees have little opportunity for advancement, hold jobs that are inherently dissatisfying because they do not provide avenues for skill development and change, and experience conflict with their co-workers and various organizational subgroups. In both of these cases, the conceptual distinction between career-related and job-related outcomes would not likely be obtained empirically because the levels of outcomes in one category would be highly related to the levels of outcomes in the other category. Typically, organizational life is not as extreme and onesided as that depicted in either of the above two scenarios. For example, many present-day British Coumbia employees are faced with situations in which opportunities for career advancement have been limited for quite some time. Because of a recession that began around 1981, many organizations reduced staffing levels. This practice, which has the overall effect of limiting promotional opportunities for the remaining employees over the recessionary period, suggests that many employees would not be satisfied with the rate at which their careers have advanced in recent years. But this does not necessarily imply that these same employees would be dissatisfied with work and organizational life, in general. Outcomes concerning one's job and organizational life, while, perhaps, related to career development 56 factors, are related also, at least tangentially, to such factors as adequate senior/co-worker relationships and the availability of social and recreational activities (Louis, Posner, & Powell, 1983). Further, equity theory (Adams, 1963; Cosier & Dalton, 1983) would suggest that being plateaued in one's position should not necessarily be associated with low levels of such outcomes as job satisfaction and organizational commitment. This is because dissatisfaction should not result in situations in which employees do not perceive themselves to be less well off than other employees in their reference group. A final reason why the career-related outcomes might not be highly related to job-related outcomes lies in individual differences in which organizational factors and opportunities are valued. For some people advancement and the development of competence (i.e., career-related outcomes) are important, while, for others, stability and/or autonomy (i.e., job-related outcomes) are important (Schein, 1978). Thus, differences in orientation concerning which organizational factors and opportunities are important may lead employees to seek out organizational situations and positions, aspects of which may be highly related to the levels of some outcomes, but not necessarily to others. If this is the case, the conceptual distinction between career-related outcomes and job-related outcomes should be empirically documented. Hypothesis 1: The level of supportive behaviours received by employees will be positively related to employee outcomes. The precise nature of this hypothesis depends upon the dimensionality of both the supportive behaviours and the employee outcomes. If the distinctions between the two types of functions (Research Question 1) and the two types of employee outcomes (Research Question 2) receive empirical support, then it would be possible to evaluate four sets of relationships. Figure 1 illustrates the possible combinations. Speculation concerning the nature of the relationship within each of the four cells is based primarily on time considerations. Regardless of the type of behaviours (i.e., career or psychosocial), the relationship between supportive behaviours and the career-related outcomes should be higher than the relationship between supportive behaviours and the job-related outcomes. The job-related outcomes all concern employees' current jobs or situations. The supportive behaviours, on the other hand, make reference to a certain span of time and, potentially, could have been received long before employees obtained their present Figure 1 Relationships between Supportive Behaviours and Employee Outcomes Behaviours Associated with Career Functions Psychosocial Functions Career-Related Outcomes Job-Related Outcomes 58 positions. If none of the supportive behaviours were received while employees were in their present positions, it might be argued that the supportive behaviours and the job-related outcomes should not be related at all. It must be remembered, however, that employees' present positions are, at least in part, a function of past positions. The argument then can be made that the supportive behaviours received by employees, no matter when, should be related to job-related outcomes associated with employees' current jobs. Further, this relationship should be somewhat higher for the job-related outcomes that are less likely to change immediately when employees are assigned new jobs (organizational commitment, acceptance by co-workers) than for the potentially more volatile outcomes of job satisfaction, role conflict, and role ambiguity. However, the relationship between job-related outcomes, of whatever sort, and supportive behaviours should be lower than the relationship between career-related outcomes and the supportive behaviours. This is so because the career-related outcomes, unlike the job-related outcomes, are relevant to the same period of time as the supportive behaviours. In summary, based on temporal considerations, it is expected that the relationships between supportive behaviours and career-related outcomes (the first row of Figure 1) will be higher than the relationships between supportive behaviours and job-related outcomes (the second row of Figure 1). It is also possible that the relationships between supportive behaviours and the employee outcomes will be different for behaviours associated with career functions than for behaviours associated with psychosocial functions. Writers who have commented on the stages or phases of mentor-protege relationships (e.g., Kram, 1985; Missirian, 1980) have noted that the career functions generally emerge earlier than do the psychosocial functions. If the psychosocial functions follow the career functions, then it is more likely that the psychosocial functions will have been provided while employees held their current positions. Since the job-related outcomes pertain primarily to employees' current positions or situations, it follows that the job-related outcomes should be more highly related to supportive behaviours associated with psychosocial functions than to supportive behaviours associated with career functions. The situation may be reversed for the career-related outcomes. Most writers in the mentoring area focus more on career functions than on psychosocial functions and, at the same time, emphasize outcomes related mostly to salary and position (here referred to as career-related). Thus, without invoking temporal considerations, most writers appear 59 to hold the view that the behaviours associated with the career functions, rather than the psychosocial functions, should exhibit the higher relationship with career-related outcomes. Such a pattern of relationships makes intuitive sense when one considers that the psychosocial functions, which are usually associated with higher levels of personal interaction than are the career functions, do not necessarily concern themselves with matters related to employees' positions in the organization. For instance, Personal Counselling and Encouragement may be directed primarily at personal and family matters rather than at strictly career-related matters. The career functions, on the other hand, are usually performed in relation to other organizational members who may hold the power to influence employees' advancement in the organization (e.g., the functions of Sponsoring and Exposure and Visibility). As such, the supportive behaviours associated with the career functions should be more highly related to such outcomes as rate of promotions and salary increase (i.e., career-related outcomes) than should the more personally based behaviours associated with the psychosocial functions. The foregoing suggests that the magnitude of the relationship between supportive behaviours and employee outcomes will depend on both the type of supportive behaviours (i.e., career or psychosocial) and the type of employee outcome (i.e., job-related or career-related). Based on temporal considerations, it is expected that supportive behaviours, of whatever type, will be more highly related to career-related outcomes than to job-related outcomes. Within the two types of outcomes, the magnitude of this relationship should depend on the type of supportive behaviours under consideration, in particular, behaviours associated with the psychosocial functions should be more highly related to job-related outcomes than should behaviours associated with the career functions. This argument was based on the assumption that psychosocial functions would more likely be provided while employees held their current positions. On the other hand, the career-related outcomes should be more highly related to behaviours associated with the career functions than to behaviours associated with the psychosocial functions. This pattern of relationships is suggested primarily by the heavy emphasis placed on career-related outcomes and career functions in the vast majority of the literature related to the effectiveness of having a mentor. Having discussed the structure of supportive functions, the structure of employee outcomes, and the relationship between behaviours associated with the functions and employee outcomes, attention now will be focused on a number of additional hypotheses suggested by the literature reviewed earlier. These hypotheses refer to supportive 60 behaviours and employee outcomes in a general sense because the structure of these behaviours and outcomes has not been assessed. It is helpful to explain the terminology used in each of next three hypotheses. These hypotheses are based on the assumption that information is available concerning the level of supportive behaviours received by an individual from each of several different people. These people are referred to as sources of supportive behaviours. For the sake of explanation, it will be assumed that an individual received 50, 30, and 20 units of a certain supportive behaviour from sources A, B, and C, respectively. The highest source of supportive behaviours is the source who provides a higher level of supportive behaviours than does any other source; in this case, A, with 50 units, is the highest source of supportive behaviours. The total level of supportive behaviours is the level of supportive behaviours received from all sources combined and, in this case, is 100 (50 + 30 + 20). Finally, proportion from highest source is the proportion of total level received from the highest source. In this case, proportion from highest source is .50 (50/100). Hypothesis 2: Level of supportive behaviours received from the highest source will be less positively related to employee outcomes than will the total level of supportive behaviours received from all sources combined. A variety of writers (e.g., Darling, 1985; Halcomb, 1980) have noted that mentors, in the traditional sense, are extremely rare in organizational settings. Perhaps because of this, it has often been suggested that employees might be able to experience some of the purported benefits of a single primary mentoring relationship by relying, instead, on a variety of sources of support (Darling, 1985; Halcomb, 1980; Kram, 1985; Missirian, 1980; Phillips-Jones, 1982). Darling (1985), in particular, suggested that a succession of relationships that are less encompassing than a mentoring relationship might be very formative. These relationships might involve peers, guides, or secondary mentors (Phillip-Jones, 1982). Such individuals can contribute to an employee's career success by providing one or more of the career and psychosocial functions (Kram, 1985). While individuals who provide an employee with a high number of these functions (i.e., potential mentors) may have more impact on that employee's career development than a single individual who provides only one or two of these functions, it appears that the functions provided by all sources combined will have the greatest impact on career development. This is why the supportive behaviours (which are assumed to be associated 61 with the functions) received from all sources are expected to be more highly related to employee outcomes than the supportive behaviours received from a single source. The pattern of correlations among sources of support found by House (1981) and by Abdel-Halim (1982) also suggests that the present hypothesis will be confirmed. In both of these studies, the support derived from different sources (e.g., co-workers and supervisors) was found to be positively related to a variety of employee outcomes. While it was not clear that the level of support received from all sources combined was more highly related to the outcomes than was the level of support received from each source separately, the pattern of correlations suggests that this might have been the case. The notion that different individuals each can have an incremental effect on an employee's outcomes also makes sense on an intuitive level. Any one potential source of supportive behaviours may not have the resources (e.g., time, skill) or be in an organizational position to provide some of the supportive behaviours. For instance, an employee's co-worker may not have the power to make Sponsoring and Exposure and Visibility effective for that employee. This suggests that the employee might look to a supervisor or someone higher in the organization for behaviours associated with Sponsoring and Exposure and Visibility. But such a person may not be familiar enough with the employee's job to provide the behaviours associated with Teaching the Job; for this, the employee might look to the co-worker. While this scenario involves only three functions, it can be used to illustrate why the present hypothesis should be confirmed. It will be assumed that the supportive behaviours associated with all three functions (Sponsoring, Exposure and Visibility, Teaching the Job) have incremental effects on employee outcomes. This means that an employee who receives two functions should be less well off than an employee who receives all three functions. By concentrating on only one source of supportive behaviours (i.e., the supervisor who provides two functions), the employee's outcomes may be underestimated because knowledge of the additional supportive behaviours (i.e., those received from the co-worker) would be missing. The employee's outcomes could be estimated accurately only if the supportive behaviours received from all sources are considered simultaneously. This is why Hypothesis 2 states that the level of supportive behaviours received from the highest source should be less positively related to employee outcomes than should the level of supportive behaviours received from all sources combined. 62 Hypothesis 3: For a given level of supportive behaviours, the proportion received from the highest source will relate negatively to employee outcomes. This hypothesis is suggested by the literature concerning the hazards associated with mentor-protege relationships. !f the supportive behaviours discussed in this study are related to the phenomenon of mentoring, then the highest source of supportive behaviours would have a higher probability of being a mentor than would any of the other sources. As such, the hazards associated with mentor-protege relationships should be highest in the highest source of supportive behaviours. These hazards include being identified too closely with this source by other organizational members, envy on the part of co-workers, and being kept from better jobs (Bowen, 1985; Reich, 1985, 1986). The validity of Hypothesis 3 depends, to a large extent, on the nature of the relationship between the level of hazards and the level of supportive behaviours. Stated in a somewhat different way, the hypothesis suggests that as the percentage of a given level of supportive behaviours received from one source increases, employee outcomes will be less favourable. Implicit in this hypothesis are the assumptions (a) that the level of hazards is negatively related to the employee outcomes, and (b) that the level of hazards for a given level of supportive behaviours can be mitigated by having a variety of sources who can provide the given level of supportive behaviours. The first assumption follows directly from the kinds of hazards that have been discussed in the literature. For instance, it is easy to see how a hazard like being kept from a better job would be reflected in outcomes pertaining to career advancement. The second assumption is somewhat more problematic. The hazards have been identified as aspects of mentor-protege relationships. What is not made clear in the literature is whether the same hazards, albeit at lower levels, also can be found in what might be viewed as less intense relationships with people at work. The precise relationship between supportive behaviours and hazards will be discussed in more detail in relation to Hypothesis 5a. It appears that the present hypothesis will more likely be supported if the relationship between supportive behaviours and hazards is a curvilinear one rather than a linear one;, an example of a curvilinear relationship is depicted in Figure 2. !f some hazards also are found in less intense relationships (i.e., suggesting a linear relationship), then the proportion of supportive behaviours received from the highest Figure 2 Relationship between Supportive Behaviours and Hazards Supportive Behaviours Hazards 64 source might not be related to lower levels of employee outcomes. This is because a reduction in the proportion of supportive behaviours received from the highest source would merely distribute the hazards more uniformly; they would not be reduced by spreading them across sources. O n the other hand, if the relationship between level of hazards and level of supportive behaviours is a curvilinear one, then the present hypothesis is more likely to be supported. In such a relationship, a change in hazards would be greater for unit changes in supportive behaviours at higher levels of supportive behaviours than at lower levels; by decreasing the proportion of supportive behaviours received from the highest source, the level of hazards also would be reduced. Because of the first assumption--that hazards are negatively related to employee outcomes-such a reduction in hazards should result in more positive employee outcomes. The following hypothesis considers the patterning of supportive behaviours across sources in more detail. Hypothesis 3 was concerned, primarily, with the relationship between employee outcomes and the extent to which a given level of supportive behaviours was concentrated in a single source (i.e., the person most likely to be viewed as a mentor). The reasoning behind Hypothesis 3' can be extended logically to multiple sources of supportive behaviours. Hypothesis 4: For a giveh level of supportive behaviours, the more uniformly these behaviours are distributed across different sources, the more positive will be the employee outcomes. This hypothesis supplements Hypothesis 3 in that it entertains the possibility that, just as concentration of supportive behaviours in one source might be detrimental, concentration of supportive behaviours in a few sources also might be detrimental. While this hypothesis largely is speculative in nature, it is, again, based primarily on the literature concerning the hazards associated with mentor-protege relationships. A uniform distribution of behaviours across a variety of people decreases the likelihood of excessive dependence on any one or two of these people. If one of the people leaves the organization or falls out of favour, the potential negative consequences for a remaining employee would not be as great for an employee who has other sources of supportive behaviours as they would be for an employee who had only the one source. Also, a number of people who a!! provide the same supportive behaviours occasionally (e.g., sponsorship for positions) may have more combined credibility in the eyes of others than 65 would one or two people who provide the same behaviours continuously. Another reason why an even distribution of a given level of supportive behaviours across sources might result in higher levels of employee outcomes than concentration of the behaviours in one or two sources, involves consideration of the timing of these behaviours. It may be the case, for instance, that a number of people, each of whom are in a position to provide a few supportive behaviours at different points in an employee's career, can result in higher levels of outcomes than can one or two individuals who provide the same level of supportive behaviours at one point in time. This makes intuitive sense when it is considered that, as an employee changes positions in an organization, different individuals become important to that employee in terms of the supportive behaviours they can provide (e.g., Teaching the Job). Rather than having high levels of supportive behaviours in just one of these positions, it may be more beneficial to be the recipient of moderate levels of supportive behaviours in each of the positions. The present hypothesis supplements Hypothesis 3 in that consideration is given to the complete distribution of supportive behaviours rather than to just the proportion received from the highest source. An example will serve to illustrate this point. Consider three individuals (A, B, and C) each of whom receives a total of 100 units of supportive behaviours from four people. Assume that from each of the four people (a) A receives 50, 20, 20, and 10 units; (b) B receives 40, 40, 10, and 10 units; and (c) C receives 40, 20, 20, and 20 units. Of interest are the comparisons between the patterns of supportive behaviours received by A and B as well as those received by B and C. For example, although A receives a higher proportion of supportive behaviours from the highest source than does B, the dispersion of supportive behaviours received by both A and B is equal. Before completing discussion of the hypothetical example, dispersion will be discussed in more detail. Dispersion can be viewed as the variance of supportive behaviours across the number of sources of these behaviours. It can be computed by summing the four squared differences between observed level and expected level of supportive behaviours assuming an even distribution and, then, dividing this sum by one fewer than the number of sources. For example, for each source the expected level assuming an even distribution is total ievel (= 100 for each of A, B, and C) divided by the number of sources providing the behaviours (= 4 in each case). Thus, expected level for any 66 particular source is 100/4 = 25. For A, the sum of the four squared differences between observed and expected levels for the four sources is: (50 - 25)2 + (20 - 25)2 + (20 - 25)2 + (10 - 25)2 = 900. This sum is then divided by 3 (one fewer than the number of sources) to yield the dispersion (i.e., variance) of supportive behaviours across sources. For hypothetical person A, dispersion is 900/3 = 300. The dispersion scores for B and C are 300 and 100, respectively. The three dispersion scores show that the lower the dispersion of supportive behaviours across sources, the more uniformly the behaviours are distributed; C, who receives the same level of supportive behaviours from three of the four sources (i.e., a relatively uniform distribution) has the lowest dispersion score. To summarize the hypothetical example, A, B, and C, each of whom receives 100 units of supportive behaviours, receive 50%, 40%, and 40% of their behaviours from the highest source and have dispersion scores of 300, 300, and 100, respectively. It can now be shown how Hypotheses 2, 3, and 4 lead to different predictions concerning the level of employee outcomes for A, B, and C. Hypothesis 2 stated that the total level of supportive behaviours received would be more highly related to employee outcomes than would the level of behaviours received from the highest source. This hypothesis would predict equal levels of employee outcomes for A, B, and C because they each receive 100 units of supportive behaviours. Hypotheses 3 and 4, however, would predict a different pattern of outcomes. Hypothesis 3 stated that, for a given level of supportive behaviours (100 in this case), the proportion received from the highest source would be negatively related to employee outcomes. This hypothesis would predict equal outcomes for B and C (they receive the same proportion of behaviours from the highest source), but lower outcomes for A because A receives a higher proportion of behaviours from the highest source (.50) than do B and C (.40 in both cases). Hypothesis 4, which takes the complete distribution of supportive behaviours into account, would predict yet another pattern of outcomes. Hypothesis 4 stated that, for a given level of supportive behaviours (100 in this case), the more uniformly these behaviours are distributed, the higher will be the employee outcomes. This hypothesis would predict equal outcomes for A and B (dispersion of 300 in each case), but higher outcomes for C (lower dispersion of 100 means more uniformity). These predictions stand 67 in contrast to those made on the basis of Hypothesis 2 (A, B, and C all equal) and Hypothesis 3 (B and C equal, A lower). It should be noted that Hypotheses 2, 3, and 4 are not competing hypotheses; all three are based on different considerations. Hypothesis 2 concerns only the total level of supportive behaviours and does not consider the distribution of this level across sources. Hypothesis 3 controls for total level and only considers the proportion of behaviours received from the highest source. Hypothesis 4 supplements Hypothesis 3 in that, while it also controls for total level of behaviours, it considers the distribution of behaviours across all sources (i.e., not just the proportion from the highest source). Hypotheses 3 and 4 largely were based on the assumptions (a) that the hazards associated with mentor-protege relationships are positively related to the level of supportive behaviours received and (b) that these hazards are negatively related to the level of employee outcomes. These assumptions now will be stated in terms of formal hypotheses. Hypothesis 5a: The level of hazards associated with a source of supportive behaviours will be positively related to the level of supportive behaviours received from that source. Hypothesis 5b: The level of hazards associated with a source of supportive behaviours will be negatively related to employee outcomes. The reasoning behind Hypothesis 5a was discussed earlier and will not be repeated here. If drawbacks and disadvantages of mentor-protege relationships are as common as reported by Reich (1985, 1986), then the recipients of high levels of supportive behaviours should experience some of the same disadvantages. The major point of interest in Hypothesis 5a is the exact nature of the relationship between hazards and supportive behaviours. It was suggested previously that the hypotheses concerning the distribution of supportive behaviours would more likely be supported if the relationship between hazards and supportive behaviours is found to be curvilinear rather than linear. Although many different types of curvilinear relationships could be posited, the most likely one would involve a higher rate of increase in hazards compared to the rate of increase in the level of supportive behaviours (see Figure 2). In other words, the level 68 of hazards would increase relatively slowly at low levels of behaviours and increase relatively rapidly at high levels of behaviours. In a linear relationship, on the other hand, the level of hazards would increase at the same rate at all levels of behaviours. Unfortunately, the studies concerning the hazards of mentor-protege relationships do not shed light on the type of relationship that might be expected. The problem with these studies is that they do not report on the disadvantages and hazards of work relationships that are not mentor-protege relationships. If some of these work relationships involve the same hazards as mentor-protege relationships, then a linear relationship between hazards and supportive behaviours is more likely. However, if the hazards are virtually absent in relationships that are not characterized by high levels of supportive behaviours, then the relationship between supportive behaviours and hazards is more likely to be curvilinear. In order to address this issue adequately, work relationships that are characterized by both low and high levels of supportive behaviours need to be considered simultaneously (i.e., a controlled study). Hypothesis 5b, that hazards will be negatively related to employee outcomes, was suggested directly by the types of hazards and disadvantages that have been associated most commonly with mentor-protege relationships. Although the effects of these disadvantages on employee outcomes rarely have been delineated in the literature, it is easy to speculate. Being identified too closely with one's mentor may be associated with lower levels of co-worker acceptance, role conflict, and job satisfaction. These outcomes might be affected most severely if the nature of the work performed requires interaction with others who might be envious of the attention afforded a protege. Lower levels of organizational commitment might also be expected if a protege identifies more with the mentor than with the organization. All of the outcomes discussed to this point are job-related outcomes; career-related outcomes also may be affected by the disadvantages of mentor-protege relationships. Skill development may be hindered by being shielded from situations in which mistakes are likely to occur; learning opportunities may be lost. Finally, being kept from better jobs, a disadvantage cited by 12% of the proteges in Reich's (1985) study, likely would affect rates of promotion and salary increase. Given these potential negative effects on employee outcomes of the hazards and disadvantages associated with mentor-protege relationships, the hypothesis (5b) that hazards will be negatively related to employee 69 outcomes suggests itself. Hypothesis 6: Level of supportive behaviours received from a person will be positively related to the level of friendship with that person. Writers have viewed friendship both as a mentoring function (e.g., Kram, 1985) and as a potential outcome of mentor-protege relationships (e.g., Klopf & Harrison, 1981; Missirian, 1980). in this study friendship was treated as a potential aspect of a supportive relationship. It was suggested previously that minimal levels of friendship may be necessary in a relationship before any of the supportive functions are provided. Because some of the supportive functions (e.g., Encouragement and Personal Counselling) imply relatively high levels of trust and interpersonal closeness, friendship can be expected to be related to the level of supportive behaviours an individual receives from another. This relationship is expected also because many of the supportive behaviours may require considerable time investments on the part of the provider. Because time appears to be a highly valued organizational resource, it is unlikely to be provided willingly to others who are not valued personally. Viewed as an outcome of mentor-protege relationships, friendship may be most likely in relationships in which the individuals involved have achieved or are in a position of equality in terms of organizational status. (Missirian, 1980). This suggests that the level of friendship between co-workers and peers will be higher than the level of friendship between supervisors and subordinates. As mentioned earlier, supervisors may be hesitant or unable to enter into relationships that are characterized by mutuality and reciprocity (i.e., friendship) with people over whom they hold authority. Given that supervisors are responsible for administering rewards and punishments to their subordinates, a high level of emotional involvement may not be viewed as desirable. It appears, however, that many supervisors would be in a better position to provide a variety of supportive behaviours (e.g., Sponsoring, Exposure and Visibility) than would people of similar organizational position (e.g., peers and co-workers). In general, though, it seems likely that the level of supportive behaviours provided an employee by another individual, regardless of whether they are supervisors or co-workers, should be related to how much this individual likes the employee. Research Question 3: How do the characteristics of supportive relationships relate to employee outcomes, level of supportive behaviours, and the definitions of mentors? 70 Many studies concerning mentor-protege relationships have described these relationships in terms of what might be viewed as individual differences between proteges and mentors. For instance, many writers have reported that mentors usually occupy organizational positions at hierarchical levels higher than their proteges (e.g., Henderson, 1985; Kram, 1985; Roche, 1979). Age of mentors relative to proteges also has been considered a factor worthy of mention (e.g., Henderson, 1985; Kram, 1983) and some writers view mentors as necessarily being older than their proteges (e.g., Bowen, 1985). Numerous studies also have addressed the issue of when, in the course of a career, proteges can expect to meet their mentors (e.g., Bova & Phillips, 1981; Henderson, 1985; Roche, 1979). One final factor that has received widespread attention is gender differences/similarities in mentor-protege relationships (Clawson & Kram, 1984; Hunt & Michael, 1983; Noe, 1988). While the factors mentioned above (organizational position, age, gender, career stage) have not been the only ones that have been used to describe mentor-protege relationships, they have been mentioned with surprising frequency. In relation to sources of supportive behaviours, this leads to the question of whether individual differences on these factors between employees and their sources of support actually make a difference in regard to employee outcomes and the level of supportive behaviours that employees can expect to receive. In relation to differences in organizational position, co-workers may be less likely than supervisors to be identified as sources of supportive behaviours. But this does not necessarily imply that employees who identify co-workers as sources of support will be the recipients of lower levels of supportive behaviours and exhibit lower levels of employee outcomes than will employees who identify supervisors as sources of support. In other words, the fact that mentors are usually supervisors or other high-ranking organizational members (Henderson, 1985; Roche, 1979), does not mean that co-workers cannot be of benefit to employees. Similarly, although mentors usually are older than their proteges, younger sources of supportive behaviours also can be expected to be of benefit to employees. When employees first encounter their sources of supportive behaviours also may be unrelated to employee outcomes and the level of supportive behaviours received. Studies on mentor-protege relationships indicate that mentors usually are met during the early stages of one's career or organizational tenure. Such studies imply that if one is to 71 benefit from having a mentor, the mentor should be met early on. But what these studies fail to take into consideration is the possibility that most people (not just mentors) will be met during the early career stages or in the first few years of organizational tenure. If the probability of being provided with supportive behaviours by a person is unrelated to when that person is first encountered, then employees who first encounter their sources of supportive behaviours at different career stages may not exhibit different levels of employee outcomes. Such differences would be expected only if individuals who are met at different career stages provide differing levels of supportive behaviours. The mentoring literature suggests that people who are first encountered at early career stages provide the highest levels of behaviours; this has not been empirically demonstrated. Numerous writers have described the problems that can arise in cross-gender mentor-protege relationships (e.g., Bowen, 1985; Clawson & Kram, 1984). Many of these problems involve co-worker perceptions that proteges are being afforded special treatment and consideration because of sexual attraction with their mentors. Such perceptions may cause organizational members external to a cross-gender mentor-protege relationship to envy the protege and/or to denigrate both members of the relationship. It is important to consider, however, that organizational members may perceive similar levels of favouritism in same-gender mentor-protege relationships. While such favouritism in same-gender relationships may not be attributed to sexual attraction, the effects on organizational members, proteges, and mentors may be the same for same-gender as for cross-gender relationships; the hazards and disadvantages of mentor-protege relationships reviewed earlier may apply equally to both types of relationships. For this reason, employees involved in same-gender and cross-gender supportive relationships may not differ much from each other in terms of the level of supportive behaviours received and the level of employee outcomes exhibited. This notion is addressed empirically in this study. In summary, studies of mentor-protege relationships have found that most mentors supervised their proteges, were older than their proteges, were male, and were encountered during the proteges' early career stages. The purpose of the present research question is to evaluate whether employees who have different kinds of supportive relationships differ in terms of level of supportive behaviours received, level of employee outcomes exhibited, and likelihood of viewing their sources of support as mentors. 72 Chapter 4  Method A questionnaire was the chosen method of data collection. The questionnaire used had three parts. The first part consisted entirely of items related to most of the employee outcomes. The second part contained items related to the assessment of supportive behaviours and functions as well as items related to the relationships employees have with some of the people who provide these behaviours and functions. Finally, the third part of the questionnaire addressed primarily individual difference and demographic variables. The content of all three parts of the questionnaire was based, in part, on the results of three pilot studies. Pilot Study One The questionnaire used in this pilot study was the original version of the second part of the final questionnaire (i.e., the part related to supportive functions and behaviours). The sample included people who were working full-time and who were enrolled in evening courses offered by the Faculty of Commerce at the University of British Columbia. Three separate classes were approached. Two of these consisted of first-and third-year students enrolled in a three-year Sales and Marketing Executives Diploma Program. The other class consisted of full- and part-time first-year MBA students. Each class was provided with a short (about 5 minute) description of the nature of the study. Students who were employed full-time were asked to take a copy of the questionnaire used and to return it at the next week's class. Of the 105 questionnaires distributed, 23 (21.9%) were returned. Response rates were 22.7% (10 of 44) and 34.2% (13 of 38) for the first- and third-year diploma program courses, respectively. None of the 23 MBA students returned a questionnaire. Of the 23 returned questionnaires, 5 could not be used because most questions were not answered. Of particular interest and concern was the relatively low response rate. While factors such as lack of motivation, incentive, or time suggested themselves as plausible explanations, the much higher response rate in the second pilot study (described next) pointed to other factors. The primary reasons for the low response rate appear to have been due to (a) questionnaire design and (b) the composition of the pilot study sample. The issue of questionnaire design is dealt with in the discussion of measures used in this study. The possibility that the composition of the sample was primarily responsible 73 deserves further comment here. Discussion with one of the instructors of the Sales and Marketing Executives Diploma Program indicated that most people enrolled in the program were either low-level managerial staff or aspirants to managerial-level positions. Most worked for small businesses. Further, many of these people travelled much of the time, moving from client to client. Taken as a group, the characteristics of the people enrolled in this diploma program suggest that these people were employed at lower levels than people usually included in studies on mentoring. People in the present sample may not have been recipients of many supportive behaviours. This may explain the low response rate. Pilot Study Two The questionnaire used in this pilot study contained the original versions of the first and third parts of the final questionnaire. The major reason for conducting a pilot study using this questionnaire was to determine the reliability of most of the employee outcome measures used. The sample consisted of second-year students enrolled in one class in the same diploma program as participants in the first pilot study. The class was provided with a short (about 5 minute) description of the nature of the study. Students who were employed full-time were asked to take a copy of the questionnaire used and to return it at the next week's class. Of the 29 questionnaires distributed, 18 (62.1%) were returned. Pilot Study Three The goal of this study was to identify a subset of items that could be used to assess friendship. The results of the study are described in the discussion of measures used in the present study. The sample consisted of 37 undergraduate students enrolled in a second-year course in organizational behaviour. The class was provided with a short (about 5 minute) description of the nature of the study. Students were asked to complete a 20-item questionnaire concerned with the types of personal relationships that develop between 74 people who work closely together. Measures Organizational Commitment. Organizational commitment was measured using the 9-item short form of the Organizational Commitment Questionnaire (OCQ) developed by Mowday et al. (1979). Organizational commitment was defined as "...the relative strength of an individual's identification with and involvement in a particular organization" (Mowday et al., 1979). The OCQ items are scored on a 7-point Likert scale ranging from 1 = strongly disagree to 7 = strongly agree. All items are positively worded. Thus, a high score indicates high organizational commitment. The OCQ was found to have adequate internal consistency (alpha coefficients ranging from .84 to .90) in such diverse samples as university employees, hospital employees, and scientists and engineers (Mowday et al., 1979). Convergent validity was evidenced by relatively high negative correlations between OCQ scores and a measure assessing intent to leave an organization; employees with high OCQ scores were less intent on leaving their present organizations (Mowday et al., 1979). Discriminant validity, while not as conclusive as convergent validity, was suggested by the relatively low percentage of variance shared by the OCQ and other measures of job attitudes such as job involvement, career satisfaction, and job satisfaction (Mowday et al., 1979). In the second pilot study, the OCQ was found to have an internal consistency of .93. Role Conflict and Role Ambiguity. Role conflict and role ambiguity were assessed using the 14 items (8 for role conflict) proposed by Rizzo, House, and Lirtzman (1970). Although Rizzo et al. (1970) did not provide specific definitions of the two constructs, role conflict occurs when demands are placed upon an employee that are incongruent or incompatible with the role that one is expected to perform. Role ambiguity, on the other hand, reflects uncertainty about how to perform one's job or a lack of clarity concerning organizational policies. The role conflict and role ambiguity items are scored on a 7-point Likert scale ranging from 1 = very false to 7 = very true. The six role ambiguity items are reflected (i.e., 7 = 1, 6 = 2, etc.) before summation so that a high score indicates high role ambiguity. Two wording changes were made to the items proposed by Rizzo et al. (1970). The item that originally read "clear, planned goals and objectives for my 75 job" (Rizzo et al., 1970, p. 156), was changed to 'I have clear, planned goals and objectives for my job' in order to give the statement a verb. The other change was the substitution of the phrase 'human resources' for 'manpower' in one of the items. This change was made in order to make the item less gender-specific. The role conflict and role ambiguity scales, which were derived factor-analytically, were found to have adequate internal consistency (ranging from .78 to .82) in two samples of salaried managerial and technical employees (Rizzo et al., 1970). Convergent and discriminant validity were demonstrated by Rizzo et al. (1970) by correlating role conflict and role ambiguity with measures as diverse as satisfaction with advancement opportunity and job security, various aspects of leadership, and organization-level variables such as formalization and benefits. Rizzo et al. (1970) concluded that the pattern of correlations obtained is what would be predicted from various theoretical viewpoints as well as from previous research. In the second pilot study, role conflict and role ambiguity were found to have internal consistencies of .79 and .72, respectively. lob Satisfaction. Job satisfaction was measured using the 18-item index proposed by Brayfield and Rothe (1951). No explicit definition of job satisfaction was provided by the authors, but their questionnaire asked people to describe how they felt about their present jobs. For this reason, one might assume that the Brayfield and Rothe (1951) measure of job satisfaction provides an index of the positive affective orientation employees have toward their jobs (Price & Mueller, 1986). The job satisfaction items are scored on a 5-point Likert scale ranging from 1 = strongly disagree to 5 = strongly agree. One-half of the 18 items are reflected before summation so that a high score indicates high job satisfaction. Both Brayfield and Rothe (1951) and Cook, Hepworth, Wall, and Warr (1981) provided evidence concerning the reliability and validity of the Brayfield and Rothe index of job satisfaction. Typical estimates of internal consistency were in the .90 range in samples as diverse as adult night-school students and nursing, clerical, and support staff in a hospital. Validity was evidenced by a correlation of .92 with a previously developed index of job satisfaction (Brayfield & Rothe, 1951), as well as by moderate negative correlations with quitting (r = -.21) and thinking of quitting (r = -.54) (Cook et al., 1981). In the second pilot study, the Brayfield and Rothe (1951) index of job satisfaction was found to have an internal consistency of .94. 76 Satisfaction with Progression. Satisfaction with progression was assessed using a 5-item scale developed for the present study. Satisfaction with progression was conceptualized primarily in terms of employee satisfaction with changes in salary, positions, and professional development over the course of organizational tenure. This measure was included because most attitudinal measures used in organizational research assess employees' feelings concerning their present situations without regard to employees' feelings concerning how they came to be in their present situations. By assessing attitudes toward the rate of change of salient aspects of one's tenure with an organization, this limitation of typical attitudinal measures is overcome. The satisfaction with progression items, which are scored on a 5-point Likert scale ranging from 1 = very unsatisfied to 5 = very satisfied, are listed in Appendix A.2 In the second pilot study, the satisfaction with progression scale was found to have an internal consistency of .84. Skill Development. Three different kinds of skill development - job, interpersonal, and conceptual - were assessed. These three skills, which were referred to as technical, human, and conceptual by Katz (1955) and Reber and Van Gilder (1982), were identified by Katz (1955) as being important for successful administration, job skill refers to "...a person's proficiency in, and understanding of, the specific techniques, processes, methods, and procedures required in carrying out a particular job" (Reber & Van Gilder, 1982, p. 12). In other words, job skills are those skills required to perform one's job on a day-to-day basis. Interpersonal skill is "...the ability to interact effectively with people" (Reber & Van Gilder, 1982, p. 12). Finally, conceptual skill is the ability to "...visualize the organization as an integrated whole, recognizing how a change in any one part affects all the other parts" (Reber & Van Gilder, 1982, p. 12). The present study assessed the degree to which respondents developed job, interpersonal, and conceptual skills. In order to assess job skill development a list of 25 separate types of skills identified by Pinder (1982) were used. The degree to which each of the 25 skills has been developed is scored on a 5-point Likert scale ranging from 0 = not at all to 4 = a great deal. The skills are listed in Appendix A. The internal consistency of the job skill development scale was found to be .96 by Pinder (1982) and .92 in the second pilot study. The average inter-item correlations found in these two studies (.48 and .31, respectively) indicate that the skills identified 2 All items developed for the present study are listed in Appendix A. 77 formed highly homogeneous clusters in the two studies; attempts by Pinder (1982) to form subscales met with limited success. Although validity data on the job skill development scale are, as yet, unavailable, the skills listed appear to represent a relatively exhaustive list of the types of skills that employees require in their day-to-day work. Of the 356 respondents in Pinder's (1982) study, only 13 (3.7%) indicated that they had developed some skill other than the skills listed. Interpersonal skill development was assessed using a 5-item scale developed for the present study. Reber and Van Gilder (1982) suggested that the five factors of empathy, self-awareness, acceptance of individual differences, perceptual awareness, and an employee orientation are necessary for the development of interpersonal skill. Accordingly, one item was written to assess each of these five factors. Each of the five interpersonal skill items is scored on a 7-point Likert scale ranging from 1 = very false to 7 = very true. The items are listed in Appendix A. These are the same items that were used in the second pilot study, with one exception. The question which now reads 'I believe that it is important to understand the specific concerns of other employees', replaced the pilot study item 'I believe that it is important to get to know other employees individually.' Both of these items relate to the factor of acceptance of individual differences in that they assess the desirability of knowing other employees individually (Reber & Van Gilder, 1982). The pilot study item was replaced because it was responsible for significantly reducing the internal consistency of the interpersonal skill development scale in the pilot study to .65. It appears that the 'get to know' phrase in the pilot study item was too social in nature. The final skill, conceptual, was assessed using the 5-item scale developed by Pinder (1982). All items refer to the ability to understand how different parts of an organization relate to one another. This is the primary aspect of conceptual skill development discussed by both Katz (1955) and Reber and Van Gilder (1982). Each of the five conceptual skill development items is scored on a 7-point Likert scale ranging from 1 = very false to 7 = very true. The items are listed in Appendix A. The internal consistency of the scale was .86 in Pinder's (1982) study and .82 in the second pilot study. 78 Acceptance by Co-Workers. Acceptance by co-workers was assessed using a 5-item scale developed for the present study. Acceptance by co-workers was conceptualized primarily in terms of the degree to which the people an employee works with accept that employee's opinions, beliefs, and judgment. The item content was suggested by a scale used by Pinder (1982) designed to assess the degree to which an employee was accepted in a new location following a job transfer. The acceptance by co-workers items, which are scored on a 7-point Likert scale ranging from 1 = very false to 7 = very true, are listed in Appendix A. Two of the items are reflected before summation so that a high score indicates high acceptance by co-workers. This scale was not included in the pilot study, and, as such, its reliability and validity have not been demonstrated. Supportive Behaviours and Functions. Based on the literature describing mentor roles, functions, and behaviours, six items were written for each of the nine supportive functions of Sponsoring, Exposure and Visibility, Teaching the ]ob, Teaching the Informal System, Protection by Prevention, Protection by Absorption, Role Modeling, Encouragement, and Personal Counselling (see Appendix A for the items). In order to determine whether the items were representative of the categories for which they were written, five judges (both graduate and undergraduate students who were unfamiliar with the mentoring literature) were asked to indicate with which of the nine functions they believed each item to be associated. The materials contained in Appendix B were used for this purpose. The 54 items were written onto individual index cards that were presented to the judges in random order. Three of the five judges correctly sorted all 54 items into the appropriate supportive functions. The other two judges each placed one item into an inappropriate function. In both cases, an item that was intended to be representative of Protection by Absorption was sorted into the Protection by Prevention category. Overall, these results indicate that there is high agreement concerning the representativeness of the items for the supportive functions for which they were written. For this reason, all 54 items were included in the first pilot study. For each of the nine supportive functions, participants in the first pilot study were asked to give the initials of up to five people who had provided them with the particular supportive function. Of all the people listed in response to the nine supportive functions, participants were asked to single out for futher consideration those five people 79 who had been most important to them as far as their careers were concerned. If fewer than five people were identified in relation to the supportive functions, all people identified were singled out for further consideration. For each of the people singled out using the method specified above, participants in this pilot study were asked to indicate, for each of the 54 activities (behaviours) described in the items, (a) the extent to which each person engaged in that activity on their behalf, and (b) the effect they felt this level of activity to have had on their career development. Extent was scored on a 5-point Likert scale ranging from 0 = never to 4 = frequently. For effect on career development, participants were asked to use the letters H (harmful effect), N (no effect), or B (beneficial effect). H, N, and B were subsequently scored 0, 1, and 2, respectively. Based on the results of the pilot study, a number of changes were made in the procedure used to assess the supportive functions and behaviours. These changes can be summarized as follows: (a) five items instead of six were used to assess the behaviours associated with each supportive function, (b) the two functions of Protection by Absorption and Protection by Prevention were collapsed into one function (i.e., Protection), (c) the 'effect' of each behaviour was no longer assessed, (d) participants were asked to single out a maximum of three instead of five providers of each supportive function, (e) participants were asked to rate a maximum of three people on the supportive behaviours, and (f) a slightly different format was used to elicit the initials of people who provided supportive functions. Each of these changes will now be discussed. Although only 18 participants in the pilot study returned usable questionnaires, these 18 participants rated 82 people on most of the 54 supportive behaviours. The sample size used in the following analyses is a maximum of 70. This sample will be referred to as the 'total sample'; people with missing data on any of the 54 behaviours were excluded. Coefficient alpha was computed for each of the nine supportive behaviour scales using all six items. These alphas ranged from .85 to .93. While all of the estimates of internal consistency were acceptably high, it is possible that they were inflated because some of the people identified by participants provided virtually no supportive behaviours. This would have had the effect of increasing the average inter-item correlation and, thereby, coefficient alpha. For this reason, a subsample of the people who were rated by 80 the 18 participants was chosen for further analysis. This subsample consisted of the 44 people who scored 51 or above on the extent to which they provided the participants with the 54 supportive behaviours. This cutoff score was chosen because it provided an adequately sized subsample of people who provided some supportive behaviours. The alpha coefficients for each of the nine supportive behaviours scales were recalculated for this subsample. As expected, these alphas were somewhat lower (range of .79 to .92) than the coefficients obtained using the total sample. However, they remained acceptably high. Due to considerations of questionnaire length, an attempt was made to reduce the number of items, while still retaining adequate internal consistency. The first step in this data reduction procedure was to examine the correlations among the nine supportive behaviour scales for both the total sample and the subsample. These correlations indicated that the correlations in the total sample were, without exception, higher than the correlations in the subsample; this is to be expected on the basis of the argument concerning the inflated alpha coefficients in the total sample. Of interest were the high correlations (.89 and .88) between Protection by Absorption and Protection by Prevention in the two samples. Given the magnitude of these correlations, it was decided to assess Protection using one 5-item scale. The first step in deriving the 5-item Protection scale was to treat the original 12 Protection items as one scale. Examination of the item-total correlations in both the total sample and subsample indicated that six of the seven lowest item-total correlations were common to both samples. These six items were removed and coefficient alpha for the remaining six items was recalculated for both samples. The final 5-item scale was derived by dropping the item that reduced the internal consistency the least; this item was common to both the total sample and the subsample. For the remaining seven scales two criteria were used to reduce the number of items from six to five. First, if an item on a particular scale in either sample correlated more highly with another scale total than with its own, that item was removed. Based on this criterion, one item was removed from each of the following scales: Role Modeling, Exposure and Visibility, and Encouragement. In order to remove items from the remaining four scales, the item that would reduce the alpha coefficient the least in both samples was identified. For three of the scales the items identified were the same for both samples; these items were removed. For the final scale, Sponsoring, different items 81 were identified in the two samples. This problem was resolved by removing the item that resulted in the largest drop in internal consistency in either sample. The alpha coefficients for the eight 5-item scales ranged from .86 to .94 in the total sample and from .79 to .91 in the subsample. Although these coefficients were generally somewhat lower than the coefficients for the 6-item scales, all scales met the criterion alpha of .70 suggested by Nunnally (1967). As a final check on the adequacy of the 5-item scales, all items were correlated with the eight total scale scores. None of the items correlated more highly with another scale total than with its own. The items retained for the 5-item scales are given in Appendix A. The 5-item scales capture largely the same content as the 6-item scales; the lowest correlation between a 6-item scale and its corresponding 5-item scale in either sample was .92. As mentioned previously, participants were asked to indicate the extent to which the people they identified provided them with the 54 supportive behaviours, as well as the effect that this stated level had on their career development. In order to determine the relationship between extent and effect, the correlations between extent and effect were calculated for the total sample and for the subsample for each of the eight supportive behaviour scales using the 5-item scales. These correlations, for the total sample and for the subsample, respectively, were as follows: Sponsoring (.90, .80); Exposure and Visibility (.90, .85); Teaching the ]ob (.90, .81); Teaching the Informal System (.83, .74); Protection (.75, .68); Role Modeling (.77, .45); Encouragement (.85, .63); and Personal Counselling (.89, .78). Given the magnitude of these correlations, as well as considerations of questionnaire length, it was decided to assess extent only. It appears that the degree to which supportive behaviours are provided is highly related to the perceived effect on career development. The final major changes in questionnaire design suggested by the pilot study concerned the number of people to be rated and the manner by which their initials were elicited. In the first pilot study, participants were asked to provide the initials of up to five people in response to each of six statements which identified one or more supportive functions. These six statements referred to all 10 of the supportive functions reviewed earlier (Friendship included). The six statements concerned the following functions: (a) Sponsoring and Exposure and Visibility, (b) Role Modeling, (c) Teaching the 82 Job and Teaching the Informal System, (d) Personal Counselling and Friendship, (e) Protection by Prevention and Protection by Absorption, and (f) Encouragement. In the final questionnaire each supportive function was assessed by a single statement. This represents an improvement over the pilot study because, in the pilot study, the statements referring to two supportive functions made it impossible to discern whether an individual identified by a participant provided one or both of the functions. Also, friendship was dropped from the statements because none of the 40 supportive behaviours (54 in the pilot study) make reference to friendship. Thus, the final questionnaire asked participants to give the initials of people in their organization who had provided them with the following eight functions: Sponsoring, Exposure and Visibility, Teaching the Job, Teaching the Informal System, Protection, Role Modeling, Encouragement, and Personal Counselling. On the basis of the pilot study results, the potential number of people identified for each function was reduced from five to three. This was done because the 18 participants who returned usable questionnaires listed an average of about three people for each of the six statements used to elicit initials. It is possible that an inability to identify people for the various functions caused many participants to not complete the questionnaire. In order to encourage potential future participants to complete the questionnaire even if they are unable to identify many providers of the eight functions, a statement was added to the instructions indicating that the number of people listed would depend largely on organizational tenure and positions held. This statement was meant to reduce any expectations that participants might hold that listing only a few people was, somehow, inadequate. The pilot study also suggested that respondents should be asked to rate people on the supportive behaviours only if these people provided more than two supportive functions. For each of the 70 people in the total sample, a count was made of the number of times they were listed in response to the six statements concerning supportive functions. A global supportive behaviour index, with a potential range of 0 to 160, also was derived by summing across the eight 5-item supportive behaviour scales. The correlation between this index and the number of times a person was listed was .60. Further, people who were listed only once or twice had significantly lower global supportive behaviour scores than people who were listed three or more times, f(68) = 3.94, p < .001. The mean global supportive behaviour score for those people who were listed once or twice (52.58) indicates that these people provided virtually no supportive 83 behaviours. This finding may also be responsible for the low response rate in the pilot study. Participants may well have become disillusioned with completing a questionnaire that asked them to respond in virtually the same manner to each of 54 statements for someone who provided them with few, or no, supportive behaviours. For this reason it was decided to revise the questionnaire in relation to how many and which people would be rated on the 40 supportive behaviours. The following procedure was used. The questions in Table 2 were used to elicit the initials of people who had provided employees with the eight supportive functions (see Appendix C for the instructions used). Table 2 contains information concerning the number of functions provided by each of three hypothetical individuals (A, B, and C); A provided six functions, B provided four functions, and C provided only two functions. The pattern of the initials of people listed by respondents in response to the eight functions was used to determine which of these people subsequently would be rated on the supportive behaviours (see Appendix C for the instructions used). All respondents were asked to rate the person who provided the highest number of functions (maximum of eight) on each of the 40 supportive behaviours. In the hypothetical example in Table 2, this would be A because A provided more functions than did either B or C. Asking respondents to rate the person listed most often in relation to the eight functions ensured that at least one person was rated on the supportive behaviours by all respondents. For instance, respondents who listed only one person's initials once in response to the functions still rated that person on the supportive behaviours. A second person was rated on the supportive behaviours only if that person provided a respondent with three of more functions. If more than two people provided three or more functions, respondents rated the person who provided the most functions. In the hypothetical example, B would have been the second person rated on the supportive behaviours. Finally, a third person was rated on the supportive behaviours only if that person provided a respondent with three or more functions. In the hypothetical example, C would not have been rated because C provided only two functions. The number of people rated on the supportive behaviours was limited to three on the basis of the pilot study results. Only two of the 18 participants in the pilot study identified more than three people more than three times in relation to the six statements used to elicit initials of providers of supportive functions. This result, in conjunction with the finding Table 2 Assessment of the Sources of Supportive Functions Since j o i n i n g your present organization have any people you have worked with: Place the i n i t i a l s here 1) t o l d others that you are competent? a) A b) B c) C 2) helped you lea r n things just by watching them? a)_A_ b) B c) 3) taught you about informal aspects of the organization such as norms or p o l i t i c s ? a) A b) C c) 4) discussed personal matters with you concerning y o u r s e l f , your career, or your family? a) A b) B c) 5) helped you lea r n job-related s k i l l s ? a) A b) c) 6) made sure that other organizational members got to know you? a) b) c) 7) provided you with encouragement or support? a) A b) B c) 8) stopped you from ge t t i n g into s i t u a t i o n s where you might not look your best? a) b) c) Note. Questions 1 to 8 r e f e r to the functions of Sponsoring, Role Modeling, Teaching the Informal System, Personal Counselling, Teaching the Job, Exposure and V i s i b i l i t y , Encouragement, and Protection, r e s p e c t i v e l y . A, B, and C r e f e r to hypothetical providers of functions. 85 that people who provided less than three functions provided relatively few supportive behaviours, provides the rationale for limiting the number of people to be rated on the supportive behaviours to three. In summary, the supportive functions and behaviours were assessed by asking participants to provide the initials of up to three people for each of the eight supportive functions. The person identified most often was rated on the extent to which he or she had provided the participant with each of 40 supportive behaviours. The two people identified next most often also were rated on the 40 behaviours, but only if they were identified at least three times in relation to the supportive functions. This means that there were three identifiable subgroups of respondents: (a) those who rated only one person on the supportive behaviours, (b) those who rated two people, and (c) those who rated three people. The 40 behaviours, which are scored on a 5-point Likert scale ranging from 0 = never to 4 = frequently, consist of five items for each of the eight supportive functions of Sponsoring, Exposure and Visibility, Teaching the Job, Teaching the Informal System, Protection, Role Modeling, Encouragement, and Personal Counselling. Supportive behaviour scale scores for each person rated are derived by summing across the five items on each scale. Finally, a global supportive behaviour score for each person rated is obtained by summing the eight scale scores. Hazards of Mentor-Protege Relationships. The hazards associated with mentor-protege relationships were assessed using a 4-item scale developed for the present study. This scale assesses the degree to which a relationship between two people is characterized by the disadvantages most commonly associated with mentor-protege relationships. These disadvantages include co-worker envy, being identified too closely with one person, being denied desirable opportunities, and being kept from assuming responsibility. The items used to assess hazards are listed in Appendix A. Participants were asked to rate each of the people singled out in response to the number of times they were identified in relation to the eight supportive functions (ie., the same people that were rated on the 40 supportive behaviours). The items, which are scored on a 5-point Likert scale ranging from 1 = no, definitely not to 5 = yes, definitely, are summed to yield a total hazards score; a high score is indicative of more hazards. 86 In the pilot study a 9-point Likert scale with the same labels as above was used to assess the hazards associated with mentor-protege relationships. Internal consistencies of the scale were .65 and .64 in the total sample and in the subsample, respectively. In both of these samples, deletion of the item relating to co-worker envy had the effect of increasing the estimates of internal consistency to over .70. It was decided to retain this item, however, because of the frequency with which jealousy and envy are discussed in the literature related to the hazards of mentor-protege relationships. Definitions of a Mentor. For each of the people rated on the 40 supportive behaviours, participants were asked to indicate whether or not these people acted as mentors. Because of the lack of definitional consensus, two definitions of mentors used in previous research were included. The first asked participants, using a Yes or No response format, 'Does (did) this person take a personal interest in your career and guide or sponsor you?' This item was adapted from the definition used by Henderson (1985) and Roche (1979). The second statement asked 'Does (did) this person serve as a career role model and actively advise, guide, and promote your career and training?' This item, which was adapted from the definition used by Vance (1982), also used a Yes or No response format. Friendship. Although a number of questionnaires have been developed to assess friendship (e.g., La Gaipa, 1977; Rubin, 1970), it was felt that none of these questionnaires was adequate for the assessment of friendship in a work-related context. For this reason 20 items were written or adapted from existing measures that would be applicable to a situation in which people worked closely with others. Students in the pilot study on friendship were asked to indicate how much each of the items applied to the relationship they had with the student with whom they worked most closely during the preceding school term. Items were scored on a scale ranging from 1 = very little to 5 = a great deal. Responses to the questionnaire were submitted to a principal components analysis and the items that loaded most highly on the first component (see Table 3) provided the initial subset of items considered for use in the present study. Based on considerations of questionnaire length, the two items in this subset with the lowest item-total correlations were removed. The remaining five items had an internal consistency of .95 in the pilot study. Table 3 P r i n c i p a l Component Analysis of Friendship Items Followed  by Varimax Rotation Component Items I II III IV How much time d i d you spend pursuing non-school rela t e d i n t e r e s t s with t h i s person? 25 07 21 76 To what extent d i d you and t h i s person value each others' opinions? 71* 43 10 -11 To what extent was t h i s person considerate of your feelings? 79* 40 -10 01 School-related matters aside, to what extent did t h i s person t r y to take advantage of you or use you? (reverse scored) 28 67* -24 -46 To what extent d i d t h i s person permit differences of opinion to come between the two of you? (reverse scored) 47 36 -10 -62 To what extent were you and t h i s person supportive of each other i n r e l a t i o n to personal matters? 43 38 53* 22 To what extent d i d t h i s person acknowledge your r i g h t to your convictions even i f he or she disagreed with you? 39 60* 19 -40 How often d i d t h i s person show praise and appreciation f o r your non-school rela t e d accomplishments? 12 79* 21 06 To what extent d i d t h i s person give advice honestly when asked f o r i t ? 30 75* 16 00 Your school r e l a t i o n s h i p aside, how close, personally, d i d you f e e l to t h i s person as a friend? 68* 46 38 10 How much d i d you support and encourage each other when one of you f e l t unhappy? 76* 26 41 -06 Table 3 continued P r i n c i p a l Component Analysis of Friendship Items Followed  by Varimax Rotation Component Items I II III IV To what extent was th i s person more interested i n you as a person than i n what you could do f o r him or her? 36 63* 05 11 To what extent d i d you exchange stimulating non-school related ideas with t h i s person? 22 46 25 63* To what extent could you t a l k to t h i s person about personal matters? 44 17 71* 02 To what extent did you and t h i s person seek each others' advice on personal matters? 17 11 80* 27 To what extent did th i s person do things f o r you w i l l i n g l y , and not expect anything i n return? 41 66* 00 16 How often d i d you confide very personal information i n each other? 05 -04 91* 11 How often d i d you show that you l i k e d each other as individuals? 83* 11 35 11 How personally close d i d you f e e l to t h i s person most of the time? 79* 31 33 20 How personally s a t i s f y i n g was your r e l a t i o n s h i p with t h i s person? 74* 50 21 09 Note. * indicates the highest loading for each item (decimal points omitted). 89 The five items were reworded slightly for the final questionnaire (see Appendix A), but the same response scale was used. Interpretation of the content of these items suggests that they all pertain to how much individuals mutually like each other. As such, friendship was conceptualized in this study as mutual liking. Characteristics of Relationships. Four questions were used to solicit information that could be used to describe the relationship between respondents and the people who were rated on the 40 supportive behaviours. These questions (see Appendix A) asked about the organizational relationship that respondents had with the people rated (e.g., supervisors, co-workers), the age and gender of the people rated, and the age at which respondents first met these people. The latter item was used to determine at what stage of organizational tenure respondents first met their providers of supportive behaviours. One other item concerning the people rated was included: 'Overall, what effect has this person had on your career development?' This item was scored on a 5-point Likert scale that ranged from 1 = very harmful to 5 = very beneficial. Extraversion. Extraversion was measured using the 12-item short form of the Extraversion scale from the Eysenck Personality Questionnaire (E-EPQ) developed by Eysenck, Eysenck, and Barrett (1985). Although Extraversion was not explicitly defined by Eysenck et al. (1985), it appears to be related to the extent to which individuals are social, outgoing, and talkative (Eysenck & Eysenck, 1975). The E-EPQ was included in the present study primarily as a measure of sociability; most items refer to whether or not an individual interacts with people in a variety of situations. The E-EPQ items are scored using a Yes/No response format; a high score is indicative of Extraversion. The short form of the E-EPQ was found to have an internal consistency of .88 in a sample of 902 students, teachers, and what were referred to as other willing subjects (Eysenck et al., 1985). Validity data on the short form of the E-EPQ was iimited to demonstration that the Extraversion scale showed low correlations with the other three factors assessed by the EPQ (Eysenck et al., 1985). However, in a study using the longer, 23-item Extraversion scale, Extraversion was found to correlate positively with measures of impulsiveness and venturesomeness (Corulla, 1987); these correlations were as expected. In the second pilot study, the long form of the E-EPQ was used. This allowed for the computation of internal consistency for both forms of the scale. The alpha 90 coefficients were .80 and .82 for the long and short forms, respectively. The correlation between the two forms was .91. For these reasons it was decided to use the short form of the E-EPQ in the final questionnaire. Self-Esteem. Self-esteem was measured using the 10-item Rosenberg Self-Esteem Scale (RSE) developed by Rosenberg (1965). Although a precise definition of self-esteem was not offered by Rosenberg (1965), Robinson and Shaver (1973) noted that the RSE measures self-acceptance and liking, or approving of the self. Demo (1985) viewed the RSE as a measure of global positive or negative self-assessment. The RSE items are scored on a 4-point Likert scale ranging from 1 = strongly disagree to 4 = strongly agree. Responses to negatively worded items are reflected before summation so that a high score indicates high self-esteem. Robinson and Shaver (1973) and Demo (1985) provided extensive evidence concerning the reliability and validity of the RSE. Typical estimates of internal consistency were in the .90 range and correlations between the RSE and other measures of self-esteem were acceptably high. In the pilot study the RSE was found to have an internal consistency of .89. Demographic Measures. A final group of items were used to provide, information concerning respondents' age, gender, education, company name, length of organizational tenure, number of positions held, number of promotions, current salary, salary upon joining the organization, and type of position held. Recruitment of Participating Firms Sample questionnaires along with cover letters explaining the nature of the survey were distributed to 60 member organizations of the Business Council of British Columbia. The cover letter explained that the survey was to be limited to managerial, technical, professional, and supervisory personnel with 15 years or fewer of organizational tenure. Potential participants were informed that there would be no charge for participation and that feedback concerning the results of the study would be provided to the organization and to participating employees who desired it. Five organizations, one from each of the energy, food wholesaling, petroleum, communications, and mining industries, agreed to participate. The author discussed the sampling procedures to be used for generating the employee sample with a senior personnel department representative from each organization. 91 Generation of the Employee Sample The manner by which the employee sample was generated differed somewhat, depending on the organization, because of relatively unique situations. However, two criteria were applied for each organization. The first of these was that employees had to be in managerial, technical, professional, or supervisory positions. The meaning of these terms was discussed with each of the personnel representatives in order to arrive at a consensus concerning the kinds of employees that would be included in the study. The major reason for specifying this criterion was to exclude employees such as line workers, clerical staff, and field workers; most research on mentors has not sampled such employees. The second criterion was that employees should have had no more than 15 years of organizational tenure. The reason for specifying this criterion was because the mentoring literature suggested that employees would most likely be the recipients of supportive behaviours in early career stages or in the early years of organizational tenure. For the organization in the energy industry, employees who met the first criterion were defined on the basis of job classification levels. A list was generated of 1185 British Columbia employees who were above a certain job level and who met the tenure requirement. It was agreed to sample 500 of these employees. Because only 11.6% of the generated list of employees were women, all women (n = 137) were included in the sample. The 363 men included in the sample were randomly selected from the 1048 men on the generated list. The food wholesaling organization decided to restrict its participation to employees who worked in one region of British Columbia. The names and addresses of the 23 employees who met the two criteria were provided to the author. This group consisted of five women and 18 men. The sample for the petroleum company was derived by choosing all exempt employees in British Columbia who met the two criteria and who were not field workers. The sample consisted of 61 employees, 14 of whom were women. The sample for the organization in the communications industry was selected by a senior personnel department representative. Because this organization had recently conducted a variety of managerial surveys, there was some concern that yet another survey would be met with relatively little enthusiasm. For this reason, the personnel representative telephoned employees and asked them if they would be willing to respond 92 to the present survey. Due to the large amount of time involved in recruiting employees in this manner, the sample for this organization was limited to 22. This group consisted of 10 men and 12 women. The organization in the mining industry decided to limit its participation to employees who worked at one British Columbia mining site. All 18 employees who met the two criteria were included in the sample; 17 of these employees were men. In summary, the sample consisted of 624 employees (169 women, 455 men) who worked for one of five British Columbia organizations. At the author's request, each organization provided information concerning the age, gender, and organizational tenure of all employees included in the sample. This information was requested (a) to ensure that the criterion of 15 or fewer years of organizational tenure was met, and (b) so that the demographic characteristics of respondents to the survey could be compared to the characteristics of the total sample. Questionnaire Distribution and Return Each of the 624 employees in the sample was mailed a package containing a questionnaire, a cover letter from the author and thesis advisor, a cover letter from a senior personnel representative in the employee's organization, and a stamped envelope in which to return the completed questionnaire directly to the thesis advisor. The first cover letter described the survey, ensured employees of confidentiality and anonymity, and indicated that employees would be provided with feedback. The major purpose of the personnel representatives' cover letters was to convey to employees that their company supported the project. Two weeks after the questionnaires were mailed, a follow-up letter was sent to all 624 employees in the sample. This letter thanked people who had already returned their questionnaires, and asked others to do so within the next few days. All questionnaires were distributed within a four week period, but were returned over an, approximately, eight week period. Employees who requested feedback were mailed a two page summary of the survey. 93 Chapter 5  Results In this chapter the research questions and hypotheses discussed earlier are evaluated. Because many measures were developed for use in the present study, the psychometric properties of all measures used in this study are described before the research questions and hypotheses are evaluated. These psychometric properties include scale means, standard deviations, reliabilities, and, where applicable, validities. First, the sample of respondents is described. Number of Survey Respondents Table 4 contains information about the number of respondents to the survey for each organization. The overall response rate for returned questionnaires was 70.8% (442 of 624); this rate ranged from 54.1% in the organization in the petroleum industry to 77.3% in the organization in the communications industry. Of the 442 returned questionnaires, 45 (10.2%) could not be used in analyses. Of these 45 questionnaires, 21 were excluded either because of a large number of missing responses or because respondents failed to follow instructions. The most common problem with these questionnaires was that respondents failed to complete the part of the questionnaire that asked about supportive functions and behaviours. The other 24 questionnaires were excluded because respondents indicated that they had been employed by their current organization for more than 16 years. One of the major criteria used for sample generation was that employees should have had 15 years or fewer of organizational tenure. It was felt that respondents with more than 16 years of tenure should be deleted from further consideration because they would not be representative of the population of interest in the study. The 19 respondents who indicated that they had been working for their present organization for 16 years were included in the sample of respondents used for analyses. Although strict application of the 15-year criterion would argue for their removal, it was felt that the relatively large reduction in sample size would not be justified by excluding respondents so close to the specified cutoff. In summary, the response rate for returned questionnaires was 70.8%. Of returned questionnaires, 10.2% could not be used because of missing responses or because respondents had been employed by their present organizations for more than 16 years. Table 4 Number of D i s t r i b u t e d , Returned, and Excluded Questionnaires Excluded: Industry D i s t r i b u t e d Returned Missing Tenure Retained Energy 500 363 16 22 325 Food Wholesaling 23 16 1 0 15 Petroleum 61 33 3 1 29 Communications 22 17 1 0 16 Mining 18 11 0 0 11 Not S p e c i f i e d 0 2 _ 0 _1 1 Totals 624 442 21 24 397 95 The response rate for usable questionnaires was 63.6% (397 of 624); this rate ranged from 47.5% for the petroleum organization to 72.7% for the communications organization. Description of Survey Respondents In this section, survey respondents are described in terms of age, gender, organizational tenure, education, and job category. For the first three of these variables it is possible to compare the sample of respondents (n = 397) to the sample of potential respondents (n = 624). Respondents to the study averaged about 38 years of age and had been employed by their current organizations for an average of slightly over 10 years (see Appendix D). Of the 397 respondents, 111 (28.0%) were women. The figures in Appendix D show that the demographic characteristics of the potential sample were quite similar. The potential sample of respondents consisted of (a) the 397 respondents used in analyses, (b) the 45 respondents whose questionnaires were excluded, as well as (c) the 182 employees who did not return questionnaires. The figures in Appendix D indicate that the sample of respondents (n = 397) is highly representative of the sample of employees to whom questionnaires were distributed. This implies that respondents to the survey did not differ appreciably from non-respondents in terms of age, gender, and organizational tenure. The figures in Appendix D also show that the majority of sample respondents had some form of post-secondary school diploma or degree (76.8%). This result is not surprising given the selection criterion that employees should be in managerial, technical, professional, or supervisory positions. Respondents were asked to describe themselves in terms of one or more of the following job categories: senior executive, middle manager, first line supervisor, professional, technical, or 'other.' Of those respondents who indicated only one of these categories (n = 324), over 50% described their positions as professional or technical-only three respondents provided no information regarding job category. Of the remaining 70 respondents, most described their positions as technical or professional in some combination with one of the other job categories. The questionnaire did not ask respondents for their specific job titles because of potential problems concerning anonymity for employees from firms with small samples. However, some job titles were made available by the personnel representatives of the participating firms. Included in the 96 sample were accountants, lawyers, engineers, personnel managers, word processing supervisors, production supervisors, systems analysts, and marketing analysts. Given such a mix of positions, it is clear why most respondents described their jobs as being professional or technical in nature. Employee Outcome and Individual Difference Measures Table 5 lists all of the employee outcome and individual difference measures of concern in this study. With the exception of salary and number of promotions, all measures were composed of multiple items. In order to retain as many respondents for analysis as possible, missing responses were substituted with individual level means for most measures as long as a respondent answered at least 80% of the items for a given measure. For example, if a respondent answered eight of the nine Organizational Commitment items, the mean score of the respondent on those eight items (rounded to the nearest integer) was substituted for the one missing item. No means were substituted for the one respondent who missed two of the Organizational Commitment items because seven items is less than 80% of the total number of items. The means, standard deviations, and alpha coefficients presented in Table 5 are based on responses following mean substitution because these statistics would be relatively unaffected by the small number of substitutions that were made. Of primary interpretative interest, at present, are the alpha coefficients. These coefficients are all high enough to suggest that the outcome and individual difference measures are sufficiently internally consistent (Nunnally, 1967) for the purposes of this study. The measures of rate of promotions and salary increase deserve further comment. Much literature suggests that people who have had mentors should have higher rates of salary increases and promotions than should people who have not had mentors. For this reason, attempts were made to derive measures related to salary and promotions that would take time into account and, further, that would be relatively unrelated to organizational tenure. A low relationship with organizational tenure was sought because such a relationship would preclude the possibility that supportive behaviours would be highly related to rates of promotion and salary increase merely because employees who had been with their organizations longer had experienced more supportive behaviours than had employees who had more recently joined their organizations. Table 5 D e s c r i p t i v e S t a t i s t i c s f o r A l l Employee Outcome and I n d i v i d u a l D i f f e r e n c e Measures Sea le Number of Number of P o t e n t i a l Range Number of M i s s i n g S t a n d a r d C o e f f i c i e n t Respondents Items of S c o r e s Data S u b s t i t u t i o n s Mean Dev i a t ion AIpha O r g a n i z a t i o n a l Commitment 396 9 to 63 38.18 11.73 .92 Role A m b i g u i t y Role C o n f l i c t Job S a t i s f a c t i o n S a t i s f a c t i o n w i t h P r o g r e s s i o n 397 397 397 395 6 8 18 6 to 42 8 to 56 18 to 90 5 to 25 12 3 18.97 30.67 6.47 9.46 67.87 11.82 16.02 4.63 .82 . 79 .91 .81 Co-worker Acceptance 397 5 to 35 28.66 4.42 .83 Job S k i l l Development 383 25 0 to 100 55.52 17.49 .93 I n t e r p e r s o n a l S k i l l Development 397 5 to 35 28.85 3.98 77 Co n c e p t u a l S k i l l Development 396 5 to 35 22.61 6. 10 .82 Average Annual S a l a r y I n c r e a s e 371 NA 2363.45 987.90 NA Average Number of P r o m o t i o n s / Y r . 393 NA . 25 . 23 NA E x t r a v e r s ion 392 12 0 to 12 37 7.11 3 . 53 .86 S e l f - e s t e e m 397 10 10 to 40 35. 75 4 . 46 . 85 98 In relation to salary, three measures concerning rate were derived, but only one of these (Average Annual Salary Increase) was retained for further analyses. The other two measures, both based on salary projections using increases in the Consumer Price index (Statistics Canada, 1988), could not be used because they were too highly related to organizational tenure. The first of these two measures was the ratio of current salary to initial salary prorated into present-day dollars using the increase in the Consumer Price Index (CPl) over the course of organizational tenure. This prorated figure is referred to as the adjusted initial salary. It was computed by multiplying initial salaries (i.e., salaries when employees first joined their present organizations) by the number of times the CPl of February 1988 exceeded the CPl of the February in the year in which employees first joined their organizations. The February figures were chosen because the present study was conducted in February. The ratio of current salary to adjusted initial salary is a measure of the extent to which employees' salaries increased relative to increases in the CPl. This ratio is not used in subsequent analyses because it correlated .29 (p < .001) with years of organizational tenure. This correlation, while taking time into account, is higher than the correlation of .25 (p < .001) between organizational tenure and current salary. It is obvious that the goal of deriving an index of salary increase that would be independent of organizational tenure was not accomplished using the ratio of current salary to adjusted initial salary. The second measure using CPl data was merely the difference between adjusted initial salary and current salary. This measure, which also takes time into account, was not retained for further analyses because it also was related significantly to organizational tenure (r = .16, p < .001). Both of the CPI-based salary measures were highly related to organizational tenure because the adjusted initial salaries of employees with longer tenure were much higher relative to their current salaries than were the adjusted initial salaries of employees who had more recently joined their organizations. In other words, it appears that use of the CPl over-adjusts the salaries of employees who have been employed by their organizations for a relatively long period of time (e.g., between 10 and 15 years). The measure of rate of salary increase used in subsequent analyses is the Average Annual Salary Increase. This average is calculated by subtracting initial annual salary from current annual salary and, then, dividing this difference by the number of years of organizational tenure. The mean and standard deviation for this average are provided in Table 5. The correlation between Average Annual Salary Increase and organizational tenure 99 was .08 (p > .05). The magnitude of this correlation indicates that Average Annual Salary Increase is relatively independent of organizational tenure. The rate of promotions was also calculated in a straightforward manner. The absolute number of promotions was divided by the number of years of organizational tenure to yield the Average Number of Promotions per Year. This index, while significantly related to organizational tenure (r = -.16, p < .001), is, as might be expected, less highly related to organizational tenure than the absolute number of promotions (r = .42, p < .001). The mean and standard deviation for the Average Number of Promotions per Year are given in Table 5. This concludes discussion of the psychometric properties of the employee outcome and individual difference measures used in the present study. In the next section, the properties of the measures associated with the three sources of support are discussed. Measures Associated with Sources of Support Respondents to the survey were asked to indicate the initials of up to three people who had provided them with each of the eight supportive functions of Sponsoring, Exposure and Visibility, Teaching the Job, Teaching the Informal System, Protection, Role Modeling, Encouragement, and Personal Counselling. Depending on how many people provided a respondent with three or more of the functions, respondents rated between one and three people on the measures listed in Table 6, Table 7, and Table 8. Table 6 provides the descriptive statistics for the measures associated with the person who provided respondents with the highest number of supportive functions. All respondents rated at least one person; the number of respondents listed in Table 6 differs from 397 because of missing data. Of the 397 respondents who rated at least one person, 106 rated only one source of support, 129 rated two people, and 162 rated three people. The information in Table 7 concerns the person who provided the second highest number of supportive functions as long as at least three functions were provided. The maximum number of respondents for Table 7 is 291 because only 291 respondents listed the initials of two or more people at least three times in relation to the eight supportive functions. The number of respondents for any particular measure differs from T a b l e 6 D e s c r i p t i v e S t a t i s t i c s f o r Measures A s s o c i a t e d with the F i r s t Source of Support Measure Number of Number of P o t e n t i a l Range Number of M i s s i n g S t a n d a r d C o e f f i c i e n t Respondents Items of Scor e s Data S u b s t i t u t i o n s Mean Dev i a t ion Alpha Sponsor ing Exposure and V i s i b i l i t y T e a c h i n g the Job Teach i n g the Informal System P r o t e c t ion Role Modeling Encouragement P e r s o n a l C o u n s e l l i n g S u p p o r t i v e B e h a v i o u r s F r iendsh ip Hazards Number of F u n c t i o n s P r o v i d e d 390 395 392 392 388 393 394 394 383 394 396 395 40 0 to 20 0 to 20 0 to 20 0 to 20 0 to 20 0 to 20 0 to 20 0 to 20 0 to 160 5 to 25 4 to 20 1 to 8 10 3 4 2 6 2 1 1 29 3 12.07 8.79 12.57 9 . 33 6 . 20 12.19 13.82 12.02 15.52 6 . 80 5. 38 4.86 5.34 4.78 5.27 4.71 5 . 1 6 4.34 5.26 87.32 31.37 4.65 2.94 1 .85 .89 .86 .85 .87 .85 .92 .87 .91 .96 .89 .59 NA O O T a b l e 7 D e s c r i p t i v e S t a t i s t i c s f o r Measures A s s o c i a t e d with the Second Source of Support Measure Number of Number of P o t e n t i a l Range Number of M i s s i n g S t a n d a r d C o e f f i c i e n t Respondents Items of Scores Data S u b s t i t u t i o n s Mean Dev i at i o n Alpha Sponsor ing Exposure and V i s i b i l i t y T e a c h i n g the Job Teach i n g the Informal System 283 288 286 287 0 to 20 0 to 20 0 to 20 0 to 20 10 10.72 7.89 11.21 8.62 5.02 4.93 4.70 4.85 .89 .84 .82 .84 P r o t e c t ion 284 0 to 20 5.82 4.54 .86 Role Modeling 287 0 to 20 11.71 4.85 .89 Encouragement 287 0 to 20 12.86 4.59 .87 P e r s o n a l C o u n s e l l i n g S u p p o r t i v e B e h a v i o u r s Fr iendsh ip Hazards 288 278 291 291 40 0 to 20 0 to 160 5 to 25 4 to 20 2 30 10.76 5.40 79.94 29.82 14.50 6.40 4.91 2.71 .92 .96 .90 .64 Number of F u n c t i o n s P r o v i d e d 289 3 to 8 4 . 37 1 . 35 NA T a b l e 8 D e s c r i p t i v e S t a t i s t i c s f o r Measures A s s o c i a t e d with the T h i r d Source of Support Measure Number of Number of P o t e n t i a l Range Number of M i s s i n g Standard C o e f f i c i e n t Respondents Items of Scores Data S u b s t i t u t i o n s Mean Dev i a t ion A Ipha Sponsor ing Exposure and V i s i b i l i t y T e a c h i n g the Job Teach i n g the Informal System P r o t e c t ion Role Modeling Encouragement P e r s o n a l C o u n s e l l i n g S u p p o r t i v e B e h a v i o u r s Fr iendsh i p Hazards Number of F u n c t i o n s P r o v i d e d 157 159 158 160 156 159 159 160 154 162 162 161 5 5 5 5 40 0 to 20 0 to 20 0 to 20 0 to 20 0 to 20 0 to 20 0 to 20 0 to 20 0 to 160 5 to 25 4 to 20 3 to 8 3 0 0 20 10.43 7.43 10.92 8 .43 5.22 10 .87 12.72 14 . 36 6 .07 3.58 5.08 4.91 4.54 4.87 4. 45 4.55 4. 30 10.18 5.37 76.83 27.36 5. 19 2.44 .93 .90 .86 78 .83 .85 .86 .84 .91 .95 .91 .60 NA 103 291 because of missing responses to items. Table 8 concerns the responses associated with the person who provided the third highest number of supportive functions as long as at least three functions were provided. The maximum number of respondents for a particular measure is 162 because 162 respondents listed the initials of three or more people at least three times in relation to the supportive functions. As before, any discrepancy between the number of respondents for a given measure and 162 is due to missing item responses. In all three tables presently under discussion, the number of respondents associated with the Number of Functions Provided (the last line in each table) is less than the maximum number of respondents. This discrepancy arises because two respondents did not associate the initials they had listed in relation to the eight supportive functions with the other measures described in the tables. One of these respondents rated three sources of support on all measures, while the other respondent rated only two sources of support. Because these two respondents completed the remaining aspects of their questionnaires well, it was decided to retain them in the sample. As with the employee outcomes and individual difference measures, individual level means were substituted for missing responses to items if at least 80% of the items for a particular measure were completed. The only exception made to this general rule was for the measure of Hazards. Because this measure consisted of only four items, the mean of the three completed items was substituted for the fourth missing item for two respondents, in general, very few mean substitutions were required. The alpha coefficients for the measures associated with the three sources of support indicate that, with the possible exception of the Hazards measures, internal consistency is sufficiently high for the purposes of the present study (see Tables 6, 7, and 8). Of particular interest is that the magnitude of the alpha coefficients is virtually the same for all three sources of support for any given measure. Although the Hazards measure has the lowest internal consistency of all measures associated with the sources of support, the fact that this measure has only four items must be taken into account. This concludes discussion of the psychometric properties of the measures associated with the three sources of support. Before the research questions and hypotheses are evaluated, analyses related to the construct validity of the supportive 104 behaviour scales are presented. The Relationship between Supportive Behaviours and Definitions of Mentors: A Validity  Check The supportive functions considered in this study have been identified by numerous writers as functions that mentors perform on behalf of proteges. While any particular mentor may not perform all of the functions, it is implicit in the literature that mentors will perform more of the functions than will people who are not mentors. Many writers have noted that a person who does not perform a variety of roles or functions (how many is not clear) cannot be viewed as a mentor (e.g., Clawson, 1980; Klopf & Harrison, 1981; Missirian, 1980; Philiips-jones, 1982). It has not often been demonstrated empirically, however, that mentors perform more of the functions than do people who are not mentors (see Alleman, 1985 for an exception). This is because many studies of mentor-protege relationships merely describe the functions performed by mentors; no attention is paid to the functions provided by people who are not mentors. This, as mentioned previously, is one of the problems with non-controlled studies. The responses to. this study can be used to compare directly the level of supportive behaviours provided by people who are identified as mentors with the level of supportive behaviours provided by people who are not identified as mentors. Because most of the definitions of mentors that have been proposed in previous work make explicit mention of one or more of the supportive functions identified in the mentoring literature (e.g., the definitions used by Roche, 1979 and by Vance, 1982), the level of supportive behaviours associated with these functions should be related to these definitions. Failure to distinguish between mentors and people who are not mentors would seriously question both the content and construct validity of much of the work related to the phenomenon of mentor-protege relationships (as well as the reasoning in this study). Two definitions that previously were used in research concerning mentors were adapted for use in this study. Definition 1: 'Does (did) this person take a personal interest in your career and guide or sponsor you?' (adapted from Roche, 1979). Definition 2: 'Does (did) this person serve as a career role model and actively advise, guide, and promote your career and training?' (adapted from Vance, 1982). The first definition is relatively broad and general, while the second is somewhat more specific and 105 exclusionary; note, for instance, the use of the word 'and' in the second definition. initial assessment of the relationship between level of supportive behaviours received and responses to the two definitions of mentors involved separate treatment of the two definitions. For each definition and for each source of supportive behaviours, a Hotelling's T2 analysis was used to compare respondents who answered 'yes' to the definition to respondents who answered 'no' to the definition. The eight supportive behaviour scales were used in these multivariate analyses. Because all six of the analyses were significant at the .001 level, univariate analyses were conducted. For each definition, f-tests were performed to compare respondents who answered 'yes' to the definition to respondents who answered 'no' to the definition; the eight supportive behaviour scales as well as the sum of the eight scales (i.e., TSBS) served as the dependent variables. These analyses were performed for each of the three sources of supportive behaviours, separately. The means and standard deviations of the supportive behaviour scales for people who were or were not viewed as mentors according to the two definitions of mentors are given in Appendix E. Of the 54 mean differences in Appendix E, all were significant at the .05 level (one-tailed) and 50 were significant at the .001 level. In all cases people who were viewed as mentors provided significantly higher levels of supportive behaviours than did people who were not viewed as mentors. The means for these two groups of people usually differed by between one-half and one standard deviation. The relationship between supportive behaviours and the definitions of mentors becomes somewhat more complicated to assess when the two definitions are considered in combination. This is so because some respondents answered 'yes' in response to one definition and 'no' in response to the other definition. The response patterns to the two definitions are given in Table 9. While it was expected that some respondents would answer 'yes' to the broader definition (i.e., Definition 1) and 'no' to the more specific definition (i.e., Definition 2), the reverse pattern was not expected. The reason for the obtained response pattern may be due to the term 'sponsor' that was used in the broad definition, but not used in the specific definition. Based on the obtained response pattern, four groups of respondents were formed for each of the three sources of supportive behaviours: (a) respondents who answered 'yes' to both definitions (i.e., the source was viewed as a mentor according to both definitions), (b) respondents who answered 'yes' to the broad definition and 'no' to the Table 9 Response Patterns f o r the Mentor De f i n i t i o n s D e f i n i t i o n 1: Yes Yes No No F i r s t Source D e f i n i t i o n 2: Yes No Yes No Number of respondents 198 57 19 90 Mean for TSBS-1 101.39 78.35 87.11 59.98 F(3, 360) = 54.11, p_ < .001. Multiple comparisons: Yes-Yes > No-Yes, Yes-No > No-No. D e f i n i t i o n 1: Yes Yes No No Second Source D e f i n i t i o n 2: Yes No Yes No Number of respondents 111 47 18 88 Mean for TSBS-2 98.49 77.04 77.06 58 .74 F(3, 260) = 42.11, p < .001. Multiple comparisons: Yes-Yes > No-Yes, Yes-No > No-No. D e f i n i t i o n 1: Yes Yes No No Third Source D e f i n i t i o n 2: Yes No Yes No Number of respondents 56 32 5 55 Mean for TSBS-3 93.96 72 .19 67.20 64.18 F(3, 144) = 15.04, p < .001. Multiple comparisons: Yes-Yes > No-Yes, Yes-No, No-No. Note. D e f i n i t i o n 1 = Does (did) t h i s person take a personal i n t e r e s t i n your career and guide or sponsor you? D e f i n i t i o n 2 = Does (did) t h i s person serve as a career r o l e model and a c t i v e l y advise, guide, and promote your career and training? TSBS = Total Supportive Behaviour Score. 107 specific one, (c) respondents who answered 'no' to the broad definition and 'yes' to the specific one, and (d) respondents who answered 'no' to both definitions. These four groups were compared on the total supportive behaviour scale scores (i.e., TSBS-1, TSBS-2, and TSBS-3 for the first, second, and third sources, respectively). Student Newman-Keuls multiple comparisons indicated that respondents who answered 'yes' to both definitions had significantly higher mean supportive behaviour scale scores than did the other three groups of respondents (see Table 9 for the group means). This was found for all three sources of supportive behaviours. For the first and second sources of supportive behaviours, respondents who answered 'no' to both definitions had significantly lower mean supportive behaviour scale scores than did the other three groups of respondents (i.e., the 'no-no' group of respondents scored significantly lower than did the 'yes-no,' the 'no-yes,' and the 'yes-yes,' groups of respondents. These pair-wise comparisons were the only ones that were significant at the .05 level. For all three sources of supportive behaviours, respondents who answered 'yes' to only one definition had mean supportive behaviour scale scores that were lower than those of respondents who answered 'yes' to both definitions (significant at .05 for all sources) and higher than those of respondents who answered 'no' to both definitions (significant at the .05 level for the first two sources). This pattern of means suggests that the supportive behaviour scale score differentiates among respondents who definitely view a source of behaviours as a mentor (the 'yes-yes' group), respondents who are not sure whether the person is a mentor (the 'yes-no' and 'no-yes' groups), and respondents who definitely do not view the person as a mentor (the 'no-no' group). The foregoing results indicate that the level of supportive behaviours received by an individual was highly related to whether the source of the behaviours was viewed as a mentor. In general, the definitions of mentors accounted for approximately 20% of the variance in the level of supportive behaviours; the correlations between the definitions (coded as no = 0 and yes = 1) and the supportive behaviours ranged from .39 (p < .001) to .52 across the three sources. Research Question 1: The Structure of Supportive Behaviours Kram (1985) suggested that the functions mentors perform on behalf of other employees can logically be grouped into the two broad categories of career functions and psychosocial functions. Although the supportive functions under consideration in this study differ somewhat from the functions identified by Kram (1985), it is possible to 108 investigate the relative independence of the career and psychosocial functions using the method of principal components (Harman, 1976). Empirical support for Kram's (1985) distinction would be obtained if the supportive behaviour scales for Sponsoring, Exposure and Visibility, Teaching the )ob, Teaching the Informal System, and Protection loaded on one component, while the scales for Role Modeling, Encouragement, and Personal Counselling loaded on another component. For all analyses described in this section, principal components analysis was used to explore the structure of the eight supportive behaviour scales. The Kaiser-Guttman rule was used to determine the number of components (Harman, 1976). This rule stipulates that the number of components equals the number of eigenvalues greater than unity. Evaluation of the structure of supportive behaviours is complicated somewhat because of the multiple sources of support (up to three) and because different respondents rated differing numbers of people on the supportive behaviour scales. For the sake of completeness, eight different component analyses were performed. Four of these analyses concern the first source of supportive behaviours, three concern the second source, and one concerns the third source. Because all respondents rated at least one person on the supportive behaviours, it was possible to evaluate the structure of supportive behaviours for the first source for the entire sample. Separate analyses for the first source were then performed (a) for respondents who rated only one source, (b) for respondents who rated two sources, and (c) for respondents who rated three sources. For the second source of supportive behaviours, analyses were performed (a) for respondents who rated at least two sources, (b) for respondents who rated two sources, and (c) for respondents who rated three sources. Finally, for the third source of supportive behaviours, the only analysis that could be performed was for respondents who rated three sources. Table 10 indicates the number of respondents involved in each analysis, the values of the first and second eigenvalues, and the percentage of variance accounted for by each of the first two components. Additional eigenvalues are not reported because they were less than unity in all analyses. The major conclusion to be drawn from the data presented in Table 10 is that the eight supportive behaviour scales loaded on one general component that accounted for over 50% of the variance. Only one of the Table 10 Results f or P r i n c i p a l Component Analyses of Supportive Behaviours:  Eigenvalues and Percentage of Variance Accounted For F i r s t Second F i r s t source of supportive behaviours Eigenvalue Eigenvalue Total Sample 5.04 (63.0%) 0.79 ( 9.9%) (n = 383) Respondents with 1 source 4.95 (61.9%) 0.86 (10.8%) (n = 103) Respondents with 2 sources 4.74 (59.3%) 1.04 (13.0%) (n = 125) Respondents with 3 sources 4.87 (60.8%) 0.90 (11.2%) (n = 155) Second source of supportive behaviours Respondents with at least 2 sources 4.78 (59.7%) 0.81 (10.1%) (n = 278) Respondents with 2 sources 4.50 (56.2%) 0.87 (10.8%) (n = 122) Respondents with 3 sources 4.93 (61.7%) 0.84 (10.5%) (n = 156) Third source of supportive behaviours Respondents with 3 sources 4.23 (52.9%) 0.99 (12.4%) (n = 154) 110 analyses suggested the presence of a second component. The correlation matrix for the eight supportive behaviour scales for the first source of supportive behaviours is given in Table 11; the correlation matrices for the different subgroups of respondents identified in Table 10 were similar. Despite the overwhelming evidence for a single component, it was decided to force a two-component solution for all of the subgroups listed in Table 10. Because of the large number of complex loadings using an orthogonal Varimax rotation, all two-component solutions were followed by an oblique Oblimin rotation. The component pattern loadings for the eight solutions and the correlations among the two oblique components are presented in Table 12. The most striking aspect of the loadings across the different subgroups is their inconsistency. For instance, no single supportive behaviour scale had a high loading on the second component in all solutions. At times, the second component was defined primarily in terms of Protection and Teaching the Informal System, while in other solutions it was defined primarily in terms of Personal Counselling and Encouragement. Also of note in these solutions is the relatively high number of complex loadings obtained despite using an oblique rotation. Given the high percentage of variance accounted for by the first component and the high correlations among the two forced components in all solutions, the major conclusion of one general supportive behaviour component is reinforced. Two components corresponding to career functions and psychosocial functions were not obtained from the responses of repondents to the present study. For this reason, the Total Supportive Behaviour Score (TSBS) is used in most subsequent analyses. The TSBS for a particular source of support is the sum of all eight supportive behaviour scales. The scores for the first, second, and third sources of supportive behaviours are referred to as TSBS-1, TSBS-2, and TSBS-3, respectively. Research Question 2: The Structure of Employee Outcomes The employee outcome measures assessed in this study previously were rationally grouped into the two categories of job-related outcomes and career-related outcomes. In order to evaluate whether this distinction could be made empirically, a principal components analysis was performed on the correlation matrix for the 11 employee outcomes presented in Table 13. The Kaiser-Guttman rule suggested that three Table 11 Correlations among Supportive Behaviour Scales - F i r s t Source Scale ; SP EX TJ TI PR RM EN Sponsoring (SP) -Exposure and V i s i b i l i t y (EX) 72 Teaching the Job (TJ) 59 65 Teaching the Informal System (TI) 57 66 44 Protection (PR) 52 59 45 63 Role Modeling (RM) 61 63 72 52 45 Encouragement (EN) 76 66 67 54 46 61 Personal Counselling (PC) 55 51 49 52 38 50 67 T a b l e 12 R e s u l t s f o r P r i n c i p a l Component A n a l y s e s of S u p p o r t i v e B e h a v i o u r s :  Component Loadings f o r Two F o r c e d Components S u p p o r t i v e Behaviour S c a l e s F i r s t source of s u p p o r t i v e b e h a v i o u r s Component SP EX TJ TI PR RM EN PC r T o t a l Sample 1 69 52 91 12 -07 83 89 76 (n = 383) 2 23 45 - 10 81 94 00 00 00 62 Respondents with 1 source 1 68 45 -09 79 17 02 66 97 (n = 103) 2 -22 -51 -95 -10 -69 -85 -24 19 -59 Respondents with 2 s o u r c e s 1 75 53 92 13 -09 78 88 78 (n = 125) 2 19 45 - 12 80 94 05 -20 -03 49 Respondents with 3 s o u r c e s 1 62 42 87 -03 -02 83 86 76 (n = 155) 2 30 56 -06 91 89 -01 06 -05 59 Second source of s u p p o r t i v e b e h a v i o u r s Respondents with at l e a s t 2 sources 1 47 73 65 81 87 77 13 -07 (n = 278) 2 -50 - 16 -22 05 13 -03 -85 -91 -59 Respondents with 2 s o u r c e s . 1 73 40 73 10 -08 65 94 81 (n = 122) 2 16 50 14 76 95 12 - 10 - 1 1 56 Respondents with 3 s o u r c e s 1 49 77 75 84 74 91 15 -08 (n = 156) 2 49 13 09 -07 05 -10 84 94 59 T h i r d source of s u p p o r t i v e b e h a v i o u r s Respondents with 3 s o u r c e s 1 82 88 87 27 40 58 60 - 13 (n = 154) 2 10 -06 - 18 64 47 12 35 94 45 Note. SP = S p o n s o r i n g , EX = Exposure and V i s i b i l i t y , TJ = Teaching the Job, TI = T e a c h i n g the Informal System, PR = P r o t e c t i o n , RM = Role Modeling, EN = Encouragement, PC = P e r s o n a l C o u n s e l l i n g , r = c o r r e l a t i o n between the two components. T a b l e 13 C o r r e l a t i o n s among Employee Outcome Measures Sea l e OC RA RC JS SP CA JO IS . CS SI O r g a n i z a t i o n a l Commitment (OC) Rol e A m b i g u i t y (RA) -46 Rol e C o n f l i c t (RC) -25 37 --Job S a t i s f a c t i o n (JS) 59 -45 -31 --" S a t i s f a c t i o n w i t h P r o g r e s s i o n (SP) 50 -31 -25 50 --Co-worker Acceptance (CA) 25 -44 -27 39 25 Job S k i l l Development (JO) 44 -26 01 33 36 25 I n t e r p e r s o n a l S k i l l Development (IS) 17 - 13 09 09 04 22 16 Co n c e p t u a l S k i l l Development (CS) 41 -42 -12 32 30 27 45 29 Average Annual S a l a r y I n c r e a s e (SI) 12 -09 -04 14 27 10 19 08 22 Average Number of P r o m o t i o n s / Y r . (PR) 16 - 12 -09 15 33 17 17 06 1 1 31 Note. C o r r e l a t i o n s g r e a t e r than .16 i n a b s o l u t e magnitude are s i g n i f i c a n t at the .001 l e v e l , o n e - t a i l e d . C o r r e l a t i o n s between .13 and .16 are s i g n i f i c a n t at the .01 l e v e l and c o r r e l a t i o n s between .09 and .12 are s i g n i f i c a n t at the .05 l e v e l . n = 355 f o r a l l c o r r e l a t i o n s ( d ecimal p o i n t s o m i t t e d ) . 114 components should be retained for rotation. These three components accounted for 33.4%, 12.1%, and 11.1% of the variance, respectively. The components were rotated both orthogonally (Varimax) and obliquely (Oblimin) and the resulting component pattern matrices are presented in Table 14. Interpretation of both of these matrices indicates that the employee outcomes can be grouped into three conceptually meaningful categories. In cases where an outcome loaded above .40 on two components, the loading with the highest magnitude was used to determine the category into which the outcome should be grouped. Using this rule of thumb, both rotations resulted in the identification of the same three categories of outcomes. The first category of outcome measures, defined by the first component, consists exclusively of what were previously referred to as job-related outcomes. None of these outcomes had appreciable loadings on either of the other two components, ln summary, the job-related outcomes are Organizational Commitment, Role Ambiguity, Role Conflict, Job Satisfaction, and Acceptance by Co-workers. The outcome measures previously discussed as career-related outcomes loaded on two separate components. The first of these had high loadings for the three aspects of skill development (job, interpersonal, and conceptual). None of these measures loaded appreciably (i.e., above .40) on either of the other two components. Thus, the second component can be viewed as a skill development component. The third component had high loadings for the other three career-related outcomes of Satisfaction with Progression, Average Annual Salary Increase, and Average Number of Promotions per Year. Although Satisfaction with Progression also had high loadings on the first component, the loadings on the third component were higher in both the orthogonal and oblique solutions. For this reason, Satisfaction with Progression was grouped into the third category of outcomes which, for present purposes, will be referred to as promotional outcomes. The nature of these categories of outcomes deserves further comment. It was suggested earlier that each of the outcomes could be viewed as either a job-related outcome or a career-related outcome. The results of the components analysis suggest that, while the job-related outcomes related highly enough to one another to form a single component, the career-related outcomes fell into two distinct categories. The fact that the job-related outcomes formed a distinct group might be explained by the fact that they all relate to aspects of the situation that employees are experiencing presently. T a b l e 14 R e s u l t s f o r P r i n c i p a l Components A n a l y s i s of Employee Outcome Measures Orthogonal Component Loadings O b l i q u e Component L o a d i n g s Sea l e I II I I I I II I I I O r g a n i z a t i o n a l Commitment 66 31 21 65 21 13 Role Ambiguity -75 -20 02 -77 - 10 11 Role C o n f l i c t -69 38 00 -73 48 05 Job S a t i s f a c t i o n 74 14 20 73 03 12 S a t i s f a c t i o n w i th P r o g r e s s i o n 53 07 56 47 -04 52 Co-worker Acceptance 57 25 01 58 18 -06 Job S k i l l Development 29 58 31 24 52 26 I n t e r p e r s o n a l S k i l l Development 00 74 -07 -02 76 - 1 1 C onceptual S k i l l Development 39 63 16 36 57 09 Average Annual S a l a r y I n c r e a s e -02 14 74 - 1 3 08 76 Average Number of Promotions/Yr. 08 -02 76 -01 - 10 78 Note. The f i r s t t h r e e e i g e n v a l u e s were 3 .67 (33. 4%), 1. .33 (12. 1%), and 1.22 (11.1%). H i g h e s t c o r r e l a t i o n among components was between I and I l l (r = . 24). Decimal p o i n t s omitte 116 A second point worth noting is that method variance is an unlikely explanation for emergence of the three components because measures that were assessed in different ways (e.g., Satisfaction with Progression and Average Annual Salary Increase) loaded on the same component, and measures that were assessed in the same way (e.g., Acceptance by Co-workers and Conceptual Skill Development) loaded on different components. The interesting aspect of the career-related outcomes is that, with the exception of the measure of Interpersonal Skill Development, they all refer explicitly to the time since employees first joined their present organizations. Despite this communality, the career-related outcomes split into the empirically distinct categories of Skill Development Outcomes and Promotional Outcomes. In order to facilitate analyses concerning the relationship between the three categories of outcomes and the measures associated with the sources of supportive behaviours, the scores on all outcome measures within a particular category were combined into a composite score for that category. For each of the categories, the first principal component for the measures in that category was extracted and component scores were calculated using the regression method (Harman, 1976). For instance, in order to arrive at a composite score for the Skill Development Outcomes, the 3 x 3 correlation matrix for the measures of Job Skill Development, Interpersonal Skill Development, and Conceptual Skill Development was submitted to a principal component analysis that retained only the first component. On the basis of the loadings on this component, the three measures of skill development were combined (using the regression method) to yield the Skill Development Composite. The same procedure was used for the other two categories of outcomes. The composite scores were standardized to have a mean of zero and a standard deviation of one. The first principal component was used because weighting measures by the loadings on this component results in the most internally consistent linear combination of the measures under consideration. The internal consistencies of the Job-Related, Skill Development, and Promotional Composites were .77, .58, and .57, respectively. Given that the latter two composites each are based on only three measures, the level of internal consistency of the three composites can be judged as acceptable. Relationship between Employee Outcomes and Supportive Behaviours Documentation of the relationship between employee outcomes and supportive behaviours is complicated somewhat by the fact that respondents rated differing numbers 117 of sources of support on the eight supportive behaviour scales. For the sake of completeness, Appendix F contains the correlations among all employee outcomes (including the three composites) and all supportive behaviour scales for the first, second, and third sources of supportive behaviours, respectively. In what follows no attempt will be made to discuss all of the correlations presented in Appendix F. Of primary interest are the correlations between the TSBS (Total Supportive Behaviour Score) and the Job-Related, Skill Development, and Promotional Composites. The TSBS is used because, in addition to being a reliable composite of the eight supportive behaviour scales, the correlation between the TSBS and any particular outcome or outcome composite is generally similar in magnitude to the highest correlation between any one of the eight scales and the outcome under consideration. The employee outcome composites are used because, as shown in the previous section, they represent reliable indices of three conceptually distinct employee outcome categories. Also, not much interpretative generality is lost by focusing on the correlations between the composites and the TSBS rather than on the correlations between the individual employee outcomes and the TSBS. Although one of the outcomes included in a composite is, at times, more highly related to the TSBS than is the composite itself, the magnitudes of the correlations for the individual outcomes and the composites are quite similar. The most general conclusion based on the correlations presented in Appendix F is that the TSBS associated with each of the three rated sources related positively and, with one exception, significantly to all three employee outcome composites. Although this evidence is compelling, these correlations obscure information pertaining to the relationship between supportive behaviours and employee outcomes for a number of subgroups of respondents identified earlier. The three subgroups of interest are (a) respondents who rated only one source on the supportive behaviour scales, (b) respondents who rated two sources, and (c) respondents who rated three sources. Table 15 provides the complete set of correlations between the TSBS and the employee outcome composites for each source of supportive behaviours for each of the subgroups of respondents. Viewed as a group, these correlations reinforce the general conclusion that supportive behaviours are positively related to employee outcomes; the level of supportive behaviours received from any source, regardless of the number of sources rated, was positively related to all three Table 15 Correlations between Employee Outcome Composites and Supportive  Behaviours from Different Sources F i r s t source of supportive behaviours Total Sample n: Outcome Composite J-R S P .25*** 383 .47** * 372 .28 355 Respondents with 1 source n: .15 103 ,45** * 101 .24 93 Respondents with 2 sources n: .32*** .49*** .29 124 123 116 Respondents with 3 sources n: .14" 155 .36*** 148 .17 146 Second source of supportive behaviours Respondents with at least 2 sources n: .25*** .40*** .22 277 269 261 Respondents with 2 sources n: .32*** .45*** .28 121 120 114 Respondents with 3 sources n: .19** .35*** .18 156 149 147 Th i r d source of supportive behaviours Respondents with 3 sources n: .16" 154 . 23** 148 .12 145 Note. J-R = Job-Related, S = S k i l l Development, P = Promotional. *p < .05, one-tailed. **p < .01, one-tailed. ***p_ < .001, one - t a i l 119 employee outcome composites. Only two correlations were not significant at the .05 level. The evidence indicates that employees who received higher levels of supportive behaviours had more positive attitudes toward their present organizational situation, had higher levels of skill development, and exhibited higher levels of outcomes associated with organizational advancement than did employees who received lower levels of supportive behaviours. Of particular interest in Table 15 are the differences in the correlations between supportive behaviours and the employee outcome composites for the three different composites. Because of the similarity of the correlations for subgroups of respondents who rated different numbers of sources, differences among correlations for the different outcome composites will be discussed in reference to the total sample. For TSBS-1 the total sample consists of all respondents, for TSBS-2 the sample consists of respondents who rated at least two sources, and for TSBS-3 the sample consists of respondents who rated three sources. For all three sources of supportive behaviours the correlation between the level of supportive behaviours and the Skill Development Composite was higher than the correlations between the level of supportive behaviours and both the Job-Related and Promotional Composites. The correlations with the latter two composites were similar in magnitude and did not differ from one another by more than .04 for any source of supportive behaviours. Dependent samples tests of significance indicated that the correlation between TSBS-1 and the Skill Development Composite was significantly higher than the correlations between TSBS-1 and the other two employee composites; z = 4.40 for the Skill Development/Job-Related difference and z = 3.46 for the Skill Development/Promotional difference. For TSBS-2 the same differences were also significant at the .05 level; z = 2.52 for the Skill Development/Job-Related difference and z = 2.71 for the Skill Development/Promotional difference. While the differences in correlations for the third source of supportive behaviours were in the same direction as the differences for the first two sources of support, they were not significant at the .05 level; z = 0.83 for the Skill Deve!opment/)ob-Related difference and z = 1.17 for the Skill Development/Promotional difference. In summary, the above comparisons suggest that supportive behaviours, while positively and significantly related to all three employee outcome composites, were more highly related to employee outcomes concerned with skill development than to outcomes 120 pertaining to either the present organizational situation or organizational advancement. It was shown earlier that the level of supportive behaviours received by respondents (i.e., TSBS) was positively and significantly related to whether these behaviours were received from people identified as mentors (see Appendix E). Of interest is whether measures based on the definitions of mentors are related to the outcome composites in the same way in which the supportive behaviours were found to be related to the outcome composites. This question, being somewhat exploratory, was addressed for the first source of supportive behaviours only. Three definitionally based measures were used: (a) Definition 1 coded as no = 0 and yes = 1, (b) Definition 2 coded in the same way, and (c) a combination measure that was merely a count of the number of 'yes' responses to the two definitions (range of 0 to 2). The correlations between these three definitionally based measures and the three employee outcome composites are presented in Table 16; the correlations for TSBS-1 and the number of functions received are provided for comparison purposes. All correlations presented in Table 16 were significant at the .001 level. This means that the definitionally based measures, the supportive behaviours, and the supportive functions all explained a significant amount of variance in all three employee outcome composites. Investigation of differences in the correlations within the definitionally based measures revealed that the relatively broad definition (Definition 1) explained somewhat less of the variance in employee outcomes than did either of the other two definitionally based measures. Dependent samples tests of the differences in these correlations revealed that the combination measure was more highly related to all three composites than was the measure based on the broad definition (z = 2.41, p < .01 for the job-Related Composite; z = 3.20, p < .001 for the Skill Development Composite; z = 1.96, p < .05 for the Promotional Composite). None of the other differences in correlations among the definitionally based measures was significant at the .05 level. Based on the pattern of correlations between the definitionally based measures and the employee outcome composites it can be concluded that all measures are significantly related to employee outcomes, but that this relationship is highest for the measure that combines information from both definitions of mentors. The differences in correlations between the supportive behaviour scale scores and the definitionally based measures also are of interest. For the Job-Related and Promotional Composites the correlations with the supportive behaviour scale score did not differ Table 16 Correlations between Employee Outcome Composites and D e f i n i t i o n s of  Mentors and Supportive Behaviours Employee Outcome Composite n TSBS-1 NF Def 1 Def2 Def 1 + 2 Job-Related 364 .26 .19 .20 .26 .26 S k i l l Development 353 .49 .31 .24 .31 .32 Promotional 338 .28 .22 .26 .29 .31 Note. TSBS-1 = Total Supportive Behaviour Score f o r the f i r s t source. NF = Number of supportive functions f o r the f i r s t source. Defl = Does (did) t h i s person take a personal i n t e r e s t i n your career and guide or sponsor you? (no = 0, 1 = yes). Def2 = Does (did) t h i s person serve as a career r o l e model and a c t i v e l y advise, guide, and promote your career and training? (no = 0, yes = 1). Defl+2 = Count of the number of 'yes' responses to the two d e f i n t i o n s . A l l c o r r e l a t i o n s are s i g n i f i c a n t at the .001 l e v e l . 122 significantly from the correlations with all three definitionally based measures (p > .05 in all cases). This suggests that the definitionally based measures explained approximately the same amount of variance in these two employee outcomes as did the supportive behaviours. For the Skill Development Composite, however, the supportive behaviours explained a significantly higher proportion of the variance than did any of the three definitionally based measures (z = 5.02, p < .001 for Definition 1; z = 3.87, p < .001 for Definition 2; z = 3.82, p < .001 for the combination measure). In summary, the definitionally based measures, the supportive behaviours, and the supportive functions exhibited approximately the same level of relationship with the job-Related and Promotional Composites. The relationships between the definitionally based measures and the employee outcome composites were relatively uniform across the three composites. This was also found for the supportive functions. The supportive behaviours, on the other hand, exhibited a higher relationship with the Skill Development Composite than with the other two employee outcome composites. Also, the relationship between the Skill Development Composite and supportive behaviours was significantly higher than the relationships between this outcome composite and the measures based on the definitions of mentors. This finding suggests that knowledge of the level of supportive behaviours received by individuals can explain more of the variance in employee outcomes than can (a) knowledge concerning whether or not these behaviours were received from people who were explicitly seen as mentors and (b) knowledge concerning the number of functions provided. The Impact of Multiple Sources of Support on Employee Outcomes Hypothesis 2 stated that the level of supportive behaviours received from the highest source would be less positively related to employee outcomes than would the total level of supportive behaviours received from all sources combined. The results concerning the relationship between supportive behaviours received from different sources and the employee outcome composites indicate that the behaviours received from any source were positively related to the employee outcomes (see Table 15). The present analysis investigates the relationship between employee outcomes and supportive behaviours in terms of the incremental effects that additional sources of supportive behaviours have on employee outcomes, over and above the effect that one source has on these outcomes. 123 Initial evaluation of this hypothesis used an hierarchical multiple regression approach in which each of the three outcome composites served as dependent variables. For respondents who rated two sources of supportive behaviours, the TSBS associated with the source who provided the highest level of behaviours was entered at the first step; the TSBS for the other source was entered at the second step. The same procedure was fol lowed for respondents who rated three sources except that, at the second step, the TSBS for both additional sources was entered. Support for the incremental effects of additional sources of supportive behaviours would be indicated by (a) a significant increase in the R 2 obtained at the second step, and (b) positive beta weights for the TSBS for the source(s) entered at the second step. In general, the hypothesis was not supported; in only one of the six analyses was the increase in the R 2 obtained at the second step significant at the .05 level (see Table 17). Failure to support the hypothesis using the hierarchical multiple regression approach appears to be largely artifactual. Even though the relationship between employee outcome composites and supportive behaviours was positive for all sources of supportive behaviours (see Table 15), the results in Table 17 fail to show incremental effects associated with additional sources. This most likely is due to the high correlations among the supportive behaviours received from the various sources. For respondents who rated two sources, the lowest correlation between the two sources in any of the analyses was .78. Among respondents who rated three sources, the lowest correlations between supportive behaviours received from the highest source and the other two sources were .87 and .65 for the second and third highest sources, respectively. Because of this multicollinearity, the regression approach makes it appear that the supportive behaviours associated with some sources are unrelated, or even negatively related, to the employee outcomes. This conclusion does not make intuitive sense, however, because of the positive zero-order relationship between employee outcomes and supportive behaviours received from each of the individual sources. In order to overcome the problems associated with multicollinearity it was decided to assess the relationship between multiple sources of support and employee outcomes in a different way. Rather than focusing on the level of supportive behaviours received from the sources rated, it was decided to concentrate on the number of supportive functions with which respondents were provided by different members of their organization. One major advantage of focusing on the supportive functions rather than on the behaviours received T a b l e 17 H i e r a r c h i c a l M u l t i p l e R e g r e s s i o n R e s u l t s f o r Employee Outcomes and S u p p o r t i v e B e h a v i o u r s Two Sources Step 1: S u p p o r t i v e b e h a v i o u r s from h i g h e s t s o u r c e : M u l t i p l e R (and R squared): Step 2: S u p p o r t i v e b e h a v i o u r s from o t h e r s o u r c e : M u l t i p l e R (and R squared): Change in R squared from Step 1: Beta weights: Step 1 v a r i a b l e : Step 2 v a r i a b l e : Number of respondents in a n a l y s i s : Employee Outcome Composite Jo b - R e l a t e d Sk i 11 Development P r o m o t i o n a l .30 ( .09)*** .36 ( . 13)** * .04* .03 .34* 121 .50 (.25)' .51 (.26)' .01 . 32* . 22 120 .31 (.10)* *' .32 ( . 10)* * .00 .21. . 13 1 14 Three Sources Step 1: S u p p o r t i v e b e h a v i o u r s from h i g h e s t s o u r c e : M u l t i p l e R (and R squared): Step 2: S u p p o r t i v e b e h a v i o u r s from o t h e r s o u r c e s : M u l t i p l e R (and R squared): Change in R squared from Step 1: Beta weights: Step 1 v a r i a b l e : Step 2 (second h i g h e s t s o u r c e ) : Step 2 ( t h i r d h i g h e s t s o u r c e ) : Number of res p o n d e n t s in a n a l y s i s : J o b - R e l a t e d S k i l l Development Promot iona1 .12 ( .01) .23 (.05) .04 - . 14 .16 . 18 152 .35 ( . 12)' .37 ( . 13)' .01 . 29 - .06 .17 146 15 (.02) 17 ( .03) .01 . 10 - .01 . 10 143 *E < .05. **p_ < .01 . ***p_ < .001 . 125 from multiple sources is that it permits inclusion of all respondents in the analysis; the 106 respondents who rated only one source could not be included in analyses concerning multiple sources of supportive behaviours. This group of respondents did, however, provide information concerning the supportive functions that were provided by different members of their organization. It should be noted that the number of supportive functions received from an individual was positively related to the ievel of supportive behaviours received from that individual. The correlations between the number of functions and level of supportive behaviours received were .60, .46, and .39 for the first, second, and third sources of supportive behaviours, respectively. This indicates that the number of supportive functions received can serve as a proxy for the level of supportive behaviours received in examining the relationship between multiple sources of support and employee outcomes. The analyses concerning number of supportive functions also used an hierarchical multiple regression approach. As before, the three employee outcome composites served as the dependent variables. At the first step, the number of supportive functions received from the person who provided the highest number of functions was entered. At the second step, the number of supportive functions received from all other individuals was entered. An example is offered to clarify the method used. Respondents could have listed up to 24 initials in response to the statements concerning who provided them with supportive functions (up to three initials for each of the eight functions). A count was made of the number of functions provided by the person who provided the highest number. This total was entered at the first step of the regression. The number of functions provided by all other people was entered at the second step. As in the analyses concerning supportive behaviours, support for the incremental effect on employee outcomes of sources of supportive functions other than the highest source would be indicated by (a) a. significant increase in the R2 obtained at the second step, and (b) a positive beta weight for the number of supportive functions entered at the second step. The results for each of the three employee outcome composites are presented in Table 18. For all three composites, the number of supportive functions received from individuals other than the highest source of functions resulted in a significant increase in the R2. Further, unlike the analyses concerning supportive behaviours, the beta weights for both variables in the regression equation were positive and, with one exception, significant at the .05 level, in all three analyses the beta T a b l e 18 H i e r a r c h i c a l M u l t i p l e R e g r e s s i o n R e s u l t s f o r Employee Outcomes and S u p p o r t i v e F u n c t i o n s Step 1: S u p p o r t i v e f u n c t i o n s from h i g h e s t source: M u l t i p l e R (and R squared): Step 2: S u p p o r t i v e f u n c t i o n s from o t h e r s o u r c e s : M u l t i p l e R (and R squ a r e d ) : Change in R squared from Step 1: Beta w e i g h t s : Step 1 v a r i a b l e : Step 2 v a r i a b l e : Number of respondents in a n a l y s i s : Employee Outcome Composite J o b - R e l a t e d Sk i11 Development Promot iona1 .18 ( .03)** ' .26 ( .07)** ' .04*** . 10 .21*** 394 .30 ( .09)*** .21 (.04)** 37 (.14) ' .05*** .20*** .25* * * 381 .25 (.06) .02** .15** .15** 364 •p. < .05. * * a < .01. "'a < ooi 127 weights for the number of supportive functions received from individuals other than the highest source were significant at the .01 level. These results suggest that the supportive functions received from all individuals are significantly related to the employee outcome composites. This relationship is higher when the focus is on supportive functions received from all individuals than when only the supportive functions received from one individual (the highest source) are considered. This result implies that supportive functions received from any source are positively related to employee outcomes; one individual need not be the source of a high number of functions. The Impact of Dispersion of Support on Employee Outcomes Discussion to this point has focused on the relationship between employee outcomes and the extent to which employees have been the recipients of supportive behaviours and functions from single and multiple sources. In this section attention is turned to an investigation of the amount of additional variance in employee outcomes that can be explained by the degree to which the supportive behaviours and functions are distributed across sources. Based on the literature concerning the hazards associated with being provided with a high number of supportive functions by a single source (i.e., by a mentor), it was hypothesized that the more evenly a given level of support is distributed across sources, the higher would be the employee outcomes. Two hypotheses concerning this notion were formulated. The first (Hypothesis 3) concerned the extent to which supportive behaviours were concentrated in a single source and the second (Hypothesis 4) concerned the distribution of supportive behaviours across all sources. Analyses concerning the proportion of behaviours and functions provided by the highest source will be discussed first. This will be followed by evaluation of the hypothesis concerning the total distribution of behaviours across sources. Concentration of behaviours and functions in the highest source. For supportive behaviours separate analyses were conducted for respondents who rated two sources and for respondents who rated three sources. The hypothesis stated that, for a given level of supportive behaviours, the proportion of behaviours received from the highest source would relate negatively to employee outcomes. For both subgroups an hierarchical multiple regression approach was used in which the outcome composites served as the dependent variables, the total level of supportive behaviours received from all sources was entered at 128 the first step, and the proportion of these behaviours received from the highest source was entered at the second step. Support for the hypothesis would be indicated by a significant increment in the R2 obtained at the second step along with a negative beta weight for the variable entered at the second step. As the results in Table 19 show, the hypothesis was not supported. In no case was the increment in R2 obtained after adding the proportion of behaviours associated with the highest source significant; the beta weights associated with this variable, which were not consistently positive or negative, were not significant at the .05 level in any of the six analyses. These results provide consistent evidence that employee outcomes are significantly related to the level of supportive behaviours received from all sources combined (the R2 obtained at the first step was significant at the .05 level in all six analyses), but that the proportion of behaviours received from the highest source explains little additional variance in the employee outcomes. This finding suggests that the relationship between supportive behaviours and employee outcomes is more related to the aggregate level of behaviours than to the distribution of the behaviours across sources. It appears that the more supportive behaviours a person receives, the better, regardless of their source. Evaluation of this hypothesis in relation to the number of functions (as opposed to behaviours) received supports this conclusion. Again, an hierarchical multiple regression approach was used. The total number of supportive functions received from all sources was entered at the first step of the multiple regression. At the second step, the proportion of supportive functions provided by the person who provided the highest number was entered. This analysis involved the total sample of respondents. The results, presented in Table 20, are quite similar to the results concerning proportion of supportive behaviours. In all three analyses, the total number of supportive functions, entered at the first step, were significantly related to the employee outcome composites. Adding the proportion of functions associated with the highest source to the regression equation at the second step resulted in a significant increment in R2 only for the Skill Development Composite. While the increment of .01 is statistically significant (p < .05), this will not be viewed as support for the hypothesis because it is due mostly to the large sample size of 381 respondents. In summary, the results pertaining to the proportion of behaviours and functions received from the source who provided the most (Hypothesis 3) suggest that the level of functions and behaviours received from all sources is more important in understanding T a b l e 19 H i e r a r c h i c a l M u l t i p l e R e g r e s s i o n R e s u l t s f o r Employee Outcomes and  P r o p o r t i o n of S u p p o r t i v e B e h a v i o u r s Received from the Highest Source Two Sources Step 1: T o t a l s u p p o r t i v e b e h a v i o u r s : M u l t i p l e R (and R squared): Step 2: P r o p o r t i o n of b e h a v i o u r s from h i g h e s t source: M u l t i p l e R (and R squared): Change in R squared from Step 1: Beta w e i g h t s : Step 1 v a r i a b l e : Step 2 v a r i a b l e : Number of respondents i n a n a l y s i s : Employee Outcome Composite Job-Re 1ated S k i l l Development Prompt i o n a l .35 (. 12) " * .36 ( . 13)*"* .01 .31** - . 10 121 .51 (.26)*** .51 (.26)*** .00 .52* * * .03 120 32 (.10)' 32 ( .11)' .01 .35* * * .07 114 Three Sources Step 1: T o t a l s u p p o r t i v e b e h a v i o u r s : M u l t i p l e R (and R squared): Step 2: P r o p o r t i o n of b e h a v i o u r s from h i g h e s t source: M u l t i p l e R (and R squared): Change in R squared from Step 1: Beta w e i g h t s : Step 1 v a r i a b l e : Step 2 v a r i a b l e : Number of respondents in a n a l y s i s : J o b - R e l a t e d S k i l l Development Prompt iona 1 .19 (.04)' .21 (.04)' .00 . 15 - . 10 152 36 (.13)*** . 17 ( .03)' .37 (.13)*** .00 .39* * * . 10 146 17 (.03) .00 . 16 - .01 143 *e < .05. **e < .01. * * *e < .001 T a b l e 20 H i e r a r c h i c a l M u l t i p l e R e g r e s s i o n R e s u l t s f o r Employee Outcomes and  P r o p o r t i o n of S u p p o r t i v e F u n c t i o n s Received from the Highest Source Step 1: T o t a l s u p p o r t i v e f u n c t i o n s : M u l t i p l e R (and R squared): Step 2: P r o p o r t i o n of f u n c t i o n s from h i g h e s t source: M u l t i p l e R (and R squared): Change i n R squared from Step 1: Beta w e i g h t s : Step 1 v a r i a b l e : Step 2 v a r i a b l e : Number of respondents in a n a l y s i s : Employee Outcome Composite J o b - R e l a t e d S k i l l Development Prompt iona1 .26 (.07)' .27 (.07)' .00 .29* * * .05 394 .36 ( . 13)' 38 (.14)' .01 * .44* * * .13* 381 .24 ( .06)* * .25 ( .06)** .00 . 28* * * .06 364 •p. < .05. **p_ < .01 . ***p_ < .001 131 the relationship between employee outcomes and supportive behaviours and functions than the way in which the functions and behaviours are distributed across sources. The following section addresses this issue in more detail because it concerns the way in which behaviours are distributed across all sources. Distribution of behaviours across sources. Hypothesis 4 stated that, for a given level of supportive behaviours, the more uniformly the behaviours are distributed across the sources rated, the higher will be the employee outcomes. This hypothesis was evaluated for supportive behaviours only because the range of unique people who provided supportive functions (1 to 13) would have made analysis of the distribution of functions unwieldy. Separate analyses were conducted for respondents who rated two sources and for respondents who rated three sources. In all analyses an hierarchical multiple regression approach was used in which the outcome composites served as dependent variables, the total level of supportive behaviours received from all sources rated was entered at the first step, and a dispersion index was entered at the second step. The dispersion index, which was discussed in detail under Hypothesis 4, will be reviewed briefly here. The dispersion index used is essentially a measure of the variance (or the uneveness of the distribution) of the supportive behaviours across sources rated. It was calculated by summing the squared differences between the observed level of behaviours and the expected level of behaviours (assuming an even distribution) for each of the sources rated and, then, dividing this sum by one less than the number of sources rated. The expected level of behaviours is the total level received from all sources divided by the number of sources. Defined in this way, the higher the dispersion index, the less uniformly the supportive functions are distributed across the sources. This means that the present hypotheses (of greater uniformity being related to higher levels of outcomes) will be supported if (a) the R2 obtained at the second step differs significantly from the R2 obtained at the first step, and (b) the beta weight for the dispersion index is negative. A significantly positive beta weight for the dispersion index would be viewed as evidence that higher levels of outcomes are related to the extent to which the supportive behaviours are concentrated in one or two sources. As the results in Table 21 indicate, adding the dispersion index to the regression equation at the second step did not result in a significantly higher R2 than was obtained with the total level of supportive behaviours alone. While the beta weights for the total T a b l e 21 H i e r a r c h i c a l M u l t i p l e R e g r e s s i o n R e s u l t s f o r Employee Outcomes and  the D i s t r i b u t i o n of S u p p o r t i v e Behaviours a c r o s s Sources Two Sources Step 1: T o t a l s u p p o r t i v e b e h a v i o u r s : M u l t i p l e R (and R s q u a r e d ) : Step 2: D i s p e r s i o n of b e h a v i o u r s : M u l t i p l e R (and R s q u a r e d ) : Change i n R squared from Step 1: Beta weights: Step 1 v a r i a b l e : Step 2 v a r i a b l e : Number of respondents i n a n a l y s i s : Employee Outcome Composite Jo b - R e l a t e d S k i l l Development Prompt iona 1 .35 (.12)*** .51 (.26)*** .32 (.10)**' .36 (.13)**' .01 . 35* * - .08 121 .51 (.26)' .00 .51*** .04 120 32 ( . 11)' .01 .32* * * .06 114 Three Sources Step 1: T o t a l s u p p o r t i v e b e h a v i o u r s : M u l t i p l e R (and R s q u a r e d ) : Step 2: D i s p e r s i o n of b e h a v i o u r s : M u l t i p l e R (and R s q u a r e d ) : Change i n R squared from Step 1: Beta weights: Step 1 v a r i a b l e : Step 2 v a r i a b l e : Number of respondents in a n a l y s i s : J o b - R e l a t e d S k i l l Development Prompt iona1 .19 ( .04)' .21 (.04)' .00 . 19" - .09 152 .36 (. 13)* *' .37 ( . 13)'*' .00 .35*** .08 146 17 (.03)' 17 (.03) .00 .17* .01 143 '0. < .05. **p. < .01 . ***p_ < .001 133 level of supportive behaviours were consistently positive and significant, none of the beta weights for the dispersion index was significant at the .05 level. These results, when considered in conjunction with the results pertaining to the negligible additional outcome variance accounted for by the proportion of behaviours received from the highest source (as found in relation to Hypothesis 3), again suggest that the relationship between employee outcomes and supportive behaviours is accounted for more by the level of these behaviours than by the way in which these behaviours are distributed across sources. How the behaviours are distributed across sources appears to explain little of the variation in employee outcomes. The major reason for postulating that there would be a negative relationship between employee outcomes and the extent to which a given level of supportive behaviours is concentrated in one or two sources was because it was assumed that the hazards associated with supportive behaviours would be highest in relationships with high levels of supportive behaviours. Failure to find significant relationships between employee outcomes and the way in which supportive behaviours are distributed across sources may be explained by the pattern of relationships among hazards, supportive behaviours, and employee outcomes. These relationships are examined in the following section. The Relationships among Hazards, Supportive Behaviours, and Employee Outcomes The relationship between hazards and supportive behaviours is of interest for a number of reasons. Other than providing empirical support for the often-discussed negative aspects of mentor-protege relationships, a positive relationship could be viewed as evidence that respondents in this study were not merely responding in a socially desirable manner. A socially desirable response pattern would be one in which respondents with high levels of such positive factors as friendship or supportive behaviours would also have low levels of negative factors such as hazards. Such a response pattern would lead to a negative relationship between the positive and negative factors. If the relationship between supportive behaviours and hazards is positive, however, then it is unlikely that socially desirable responding is in operation. The empirical evidence supports the hypothesis (Hypothesis 5a) that hazards and supportive behaviours are positively related. The correlations between level of supportive behaviours and the associated hazards (see Table 22) were positive and significant at the .001 level of all sources of supportive behaviours. It was previously argued, however, that if the hazards and supportive behaviours are linearly (as opposed to logarithmically) Table 22 Correlations among Hazards, Supportive Behaviours, and  Employee Outcome Composites Measures HZ-1 HZ-2 HZ-3 HZ-12 HZ-123 Job-Related Composite -.20*** -.13* -.05 -.21*** -.16* S k i l l Development Composite .16** .19** .16* .20*** .21* Promotional Composite -.04 .02 .07 -.02 .04 TSBS-1 .28*** TSBS-2 .24*** TSBS-3 .26*** TSBS-12 .31*** TSBS-123 .37* Note. TSBS refers to the Total Supportive Behaviour Score. HZ ref e r s to Hazards. The numbers following the dash r e f e r to the sources over which the sum has been taken (e.g., HZ-12 refer s to the sum of the hazards f o r the f i r s t two sources. *p < .05, two-tailed. **p < .01, two-tailed. ***p_ < .001, two-tailed 135 related, support for the hypotheses concerning the distribution of behaviours across sources might not be found. A logarithmic relationship between behaviours and hazards would be evidenced by larger differences in hazards for unit changes in behaviours at high levels of behaviours than for unit changes in behaviours at low levels of behaviours. In the event of such a relationship, the total level of hazards would be lower for a given level of supportive behaviours the more evenly the behaviours are distributed across sources. Support for a logarithmic relationship between hazards and behaviours was not found. For each of the three sources of behaviours, a regression equation was estimated for predicting level of behaviours from hazards. Using the observed level of hazards, the expected level of behaviours was estimated using the estimated regression equation. A plot of the residuals (observed less expected level of behaviours) indicated that the residuals were normally distributed.3 Such a result would not have been obtained for a logarithmic relationship because, in such a case, the regression equation would fit the observations at some levels of behaviours better than at others. Given the positive relationship between hazards and supportive behaviours and the normality of the residuals, it must be concluded that the relationship between hazards and behaviours is a linear one. The results concerning the nature of the relationship between hazards and supportive behaviours indicate that the relationship is significant, positive, and linear for all three sources rated; as the level of supportive behaviours received from a source increases, the level of hazards associated with that source increases as well. This finding is important because it provides empirical support for the notion that there are. negative factors associated with the provision of behaviours often attributed to mentors. However, the present results suggest that these negative factors are not limited to relationships with high levels of supportive behaviours. Table 22 also includes the correlations concerning the relationship between the level of hazards and the employee outcome composites. Hypothesis 5b stated that the hazards would be negatively related to employee outcomes. This hypothesis received only partial support. Although the hazards associated with each source of support were 3 The nature of the relationship between supportive behaviours and hazards was also evaluated using a hierarchical regression analysis in which the level of hazards was entered at the first step and the square root of the hazards was entered at the second step. For each of the three sources, the beta weight for the square root of the hazards was not significant at the .05 level. 136 consistently negatively related to the Job-Related Composite, the hazards were consistently positively related to the Skill Development Component. There was virtually no relationship between hazards and the Promotional Composite. These results suggest that the hazards may have to be present in order for skills to develop effectively, but, at the same time, that the hazards will have negative consequences for employees in day-to-day matters (i.e., as reflected in the measures in the Job-Related Composite). In order to more fully understand the relationship among hazards, supportive behaviours, and employee outcomes one other set of analyses was performed. The hypotheses concerning the effects of dispersion on outcomes (Hypotheses 3 and 4) were based on the assumption that the level of hazards associated with a given level of supportive behaviours would be reduced by dispersing behaviours uniformly across sources. Implicit in this assumption is the notion that, for a given level of behaviours, employee outcomes will be higher the lower the level of hazards. This suggests that analyses concerning hazards and outcomes should focus directly on the hazards rather than on the hypothetical link between hazards and the dispersion of behaviours across sources. To this end, three hierarchical multiple regression analyses were performed for each of the three employee outcome composites. The first of these involved the supportive behaviours and hazards associated with the first source only; this analysis included all respondents. The second analysis was based on the supportive behaviours and hazards associated with the first two sources combined; this analysis included only respondents who rated at least two sources. The final analysis was performed for only those respondents who rated three sources; this analysis was based on the supportive behaviours and hazards for all three sources combined (i.e., the sum of supportive behaviours across all three sources and the corresponding sum for hazards). In each of the analyses, the employee outcome composites served as the dependent variables, the total level of supportive behaviours received from the sources under consideration was entered at the first step, and the level of hazards associated with these sources was entered at the second step. Support for the notion of hazards being associated with lower levels of employee outcomes for a given level of supportive behaviours would be indicated by a significant increase in the R2 obtained at the second step, along with a negative beta weight for the level of hazards. 137 This pattern was obtained in five of the nine analyses, the results of which are presented in Table 23. The strongest effects were associated with the Job-Related Composite; the lowest increment in R2 for the three Job-Related analyses was .07. The increments in R2 obtained for the Promotional Composite, while significant in two of the three analyses, were relatively small (e.g., .02). None of the analyses concerning the Skill Development Composite indicated a significant increment in the R2 after adding the hazards measure at the second step. In all five of the analyses that showed significant increases in the R2 obtained after adding the level of hazards to the regression equation, the beta weight for hazards was negative and significant at the .05 level. This suggests that, for a given level of supportive behaviours, the highest levels of employee outcomes can be expected when the level of hazards is at a minimum. However, because the supportive behaviours are positively related to all employee outcome composites, the hazards should not be minimized at the expense of supportive behaviours. This would have the effect of lowering the overall level of employee outcomes. The Relationship between Supportive Behaviours and Friendship Friendship has been viewed as a psychosocial mentoring function (e.g., Kram, 1985) as well as an outcome of mentor-protege relationships (Klopf & Harrison, 1981; Missirian, 1980; Reich, 1986). This suggests that friendship, regardless of how it is conceptualized, should be positively related to the level of supportive behaviours received from an individual (Hypothesis 6). The hypothesis was supported; the correlations between the level of supportive behaviours received from the three sources and the level of reported friendship with these sources were .50, .48, and .37 (p < .001 in all cases) for the first, second, and third sources of support, respectively. The relationship between level of friendship and supportive behaviours was assessed in more detail in order to determine whether the relationship was, in fact, linear. This was done because of the possibility that the level of supportive behaviours might increase relatively rapidly once a certain level of friendship was present than at lower levels of friendship. As with the hazards, this analysis investigated the distribution of the residuals obtained using the estimated regression equation for predicting level of supportive behaviours from level of friendship. For all three sources the residuals (observed less expected level of supportive behaviours) were normally distributed. This result suggests that the relationship between supportive behaviours and friendship is linear; the higher the T a b l e 23 H i e r a r c h i c a l M u l t i p l e R e g r e s s i o n R e s u l t s f o r Employee Outcomes and  the Hazards A s s o c i a t e d with S u p p o r t i v e Behaviours F i r s t Source Step 1: T o t a l s u p p o r t i v e b e h a v i o u r s : M u l t i p l e R (and R s q u a r e d ) : Step 2: Hazards M u l t i p l e R (and R s q u a r e d ) : Change in R squared from Step 1: Beta w e i g h t s : Step 1 v a r i a b l e : Step 2 v a r i a b l e : Number of respondents in a n a l y s i s : Employee Outcome Composite Jo b - R e l a t e d S k i l l Development Promot i o n a l .25 (.06)*** . 3 8 ( . 1 4 ) * * * . 0 8 * * * . 3 3 * " - . 3 0 * * * 381 .46 (.21 ) * * * .28 ( .08) ' . 4 6 ( . 2 1 ) * * * .00 . 4 5 * * * . 0 3 371 .31 (.09)' .02* . 3 1 * * * - . 1 3 * 3 5 5 F i r s t Two Sources Step 1: T o t a l s u p p o r t i v e b e h a v i o u r s : M u l t i p l e R (and R squared): Step 2 : Hazards: M u l t i p l e R (and R squared): Change in R squared from Step 1: Beta w e i g h t s : Step 1 v a r i a b l e : Step 2 v a r i a b l e : Number of respondents in a n a l y s i s : J o b - Related S k i l l Development Promot i o n a l . 2 6 ( . 0 7 ) ' .41 ( . 1 7 ) " . 1 0 * * * . 3 7 * * * - . 3 3 * * * 2 7 5 . 4 5 ( . 2 0 ) * * * . 2 5 ( . 0 6 ) ' . 4 6 ( . 2 1 ) * * * .00 . 4 4 * * * .05 267 . 2 7 ( . 0 8 ) ' . 01 * . 2 8 * * * - . 1 3 * 2 5 9 * £ < . 0 5 . **p_ < . 0 1 . * * * e < 001 T a b l e 23 c o n t i n u e d H i e r a r c h i c a l M u l t i p l e R e g r e s s i o n R e s u l t s f o r Employee Outcomes and  the Hazards A s s o c i a t e d with S u p p o r t i v e Behaviours A l l Three Sources Step 1: T o t a l s u p p o r t i v e f u n c t i o n s : M u l t i p l e R (and R s q u a r e d ) : Step 2: Hazards: M u l t i p l e R (and R s q u a r e d ) : Change in R squared from Step 1: Beta w e i g h t s : Step 1 v a r i a b l e : Step 2 v a r i a b l e : Number of res p o n d e n t s in a n a l y s i s : Employee Outcome Composite Job-Re 1ated S k i l l Development Prompt i o n a l .19 (.04)* .33 ( . 11)*** .07*** .30* ** -.29*** 152 .36 (.13)**' 36 (.13) **' .00 .33*** .07 146 .17 ( .03)* .17 (.03) .00 . 19* - .05 143 *e < . 0 5 . **a < . 0 1 . ***p. < .001 140 level of supportive behaviours, the higher the level of friendship. It was suggested earlier that the level of friendship might be related to the organizational position of the source of supportive behaviours relative to the position of the recipient. In particular, it was argued that the level of friendship with co-workers would be higher than the level with supervisors because the latter might be hesitant to enter into close relationships with their subordinates. This issue was evaluated by classifying the sources of supportive behaviours into the four groups of (a) supervisor, (b) higher level than supervisor, (c) co-worker, and (d) other on the basis of the question 'What is (was) your relationship with this person?' Sources were classified as 'other' if they could not be classified into any of the other three groups. While the grouping procedure is not as unambiguous as might be desired because of potential changes in relationships over time (i.e., a supervisor could have become a co-worker), a comparison of the first three groups on mean level of friendship was informative. The 'other' group was excluded from this analysis because it contained sources that could have been placed in one of the other three groups if respondents had provided more information. For example, a source that was described as 'a manager in another department' conceivably could be a co-worker or someone of higher organizational level. For each of the three sources of supportive behaviours a oneway A N O V A was used to compare the three groups (supervisors, higher level than supervisors, co-workers) on reported level of friendship. The means relevant to these comparisons are given in Table 24. Multiple comparisons (Student Newman-Keuls) revealed that the level of friendship with co-workers was significantly higher than the level of friendship with supervisors or with people at hierarchical levels higher than supervisors; the latter two groups did not differ significantly. This pattern was found for all three sources of supportive behaviours. These findings are discussed in more detail after consideration is given to how supportive relationships that differ in terms of the organizational position, age, and gender of the parties to the relationships can be characterized according to the level of supportive behaviours received, whether mentors are involved, and the level of employee outcomes exhibited. Relationship between Characteristics of Sources of Support and Mentor Definitions,  Supportive Behaviours, and Employee Outcomes It was noted earlier (in relation to Research Question 3) that mentors usually are male, senior to their proteges in terms of age and organizational position, and first 141 Table 24 Mean Levels of Friendship f o r Co-workers, Supervisors, and Organizational  Members at Higher Levels than Supervisors Higher than F i r s t Source Co-workers Supervisors Supervisors Number of respondents 55 270 30 Mean l e v e l of fri e n d s h i p 17.89 14.98 14.77 F(2, 352) = 9.86, p < .001. Multiple comparisons: Co-workers > Supervisors, Higher than Supervisors. Higher than Second Source Co-workers Supervisors Supervisors Number of respondents 66 142 37 Mean l e v e l of fri e n d s h i p 17.45 13.59 13.19 F(2, 242) = 17.53, p < .001. Multiple comparisons: Co-workers > Supervisors, Higher than Supervisors. Higher than Third Source Co-workers Supervisors Supervisors Number of respondents 51 67 22 Mean l e v e l of friendship 16.78 ' 13.16 11.32 F(2, 137) = 13.75, p < .001. Multiple comparisons: Co-workers > Supervisors, Higher than Supervisors. 142 encountered during the early career stages (i.e., establishment) or in the early years of organizational tenure. The purpose of this section is (a) to describe the present sample in terms of the factors just mentioned, and (b) to determine whether employees who have supportive relationships that differ in relation to these factors receive different levels of supportive behaviours and exhibit different levels of employee outcomes. The different types of supportive relationships also will be related to the two definitions of mentors used in this study. Table 25 contains information concerning (a) the organizational position of the sources of supportive behaviours relative to respondents, (b) gender differences between the sources of supportive behaviours and respondents, (c) the stage of organizational tenure at which respondents first met their sources of supportive behaviours, and (d) age discrepancies between the sources of supportive behaviours and respondents. All of the sources described in Table 25 were most likely to be supervisors, male, to have been met during the establishment stage of organizational tenure, and older than respondents. This characterization of the sources of supportive behaviours is highly similar to the description of mentors provided by Henderson (1985) and Roche (1979). Of interest is how the factors of organizational position, gender differences, organizational tenure, and age relate to the level of supportive behaviours received, the definitions of mentors, and employee outcomes. Because of the generally similar pattern of percentages for a particular factor across the three sources of supportive behaviours (see Table 25), the following analyses are based only on the first source of supportive behaviours. Organizational position. Analyses concerning organizational position included the three groups of (a) co-workers, (b) supervisors, and (c) organizational members at higher levels than supervisors. Sources classified as 'other' were removed from consideration because these sources potentially could have been classified into one of the other three groups (e.g., manager in another department) or because they fell into a group not widely represented in the sample (e.g., subordinates, n = 2). In relation to definitions of mentors, supportive behaviours, and employee outcomes, most of the differences that were significant were between supervisors and co-workers (see Table 26). Although higher level organizational members usually were found to be more similar to supervisors than to co-workers, the relatively low number of such higher level members (n = 30) affected the probability of finding significant differences. Table 25 C h a r a c t e r i s t i c s of Sources of Supportive Behaviours Organizational P o s i t i o n Co-workers: n (%) Supervisors Higher than Supervisors Other Unknown F i r s t Source Second Source Third Source 55 (13.9) 66 (22.7) 51 (31.5) 273 (68.8) 142 (48.8) 67 (41.4) 30 ( 7.6) 37 (12.7) 22 (13.6) 36 ( 9.1) 42 (14.4) 22 (13.6) 3 ( .8) 4 ( 1.4) 0 ( 0.0) Gender Breakdown Male Source, Male Respondent Male Source, Female Respondent Female Source, Male Respondent Female Source, Female Respondent Unknown 277 (69.8) 197 (67.7) 103 (63.6) 72 (18.1) 65 (22.3) 41 (25.3) 9 ( 2.3) 6 ( 2.1) 5 ( 3.1) 38 ( 9.6) 22 ( 7.6) 13 ( 8.0) 1 ( .3) 1 ( .3) 0 ( 0.0) When Source was F i r s t Met  Before j o i n i n g the organization 0 - 2 years a f t e r j o i n i n g 3 - 1 0 years a f t e r j o i n i n g More than 10 years a f t e r j o i n i n g Unknown 43 (10.8) 29 (10.0) 200 (50.4) 128 (44.0) 134 (33.8) 111 (38.1) 14 ( 3.5) 18 ( 6.2) 6 ( 1.5) 5 ( 1.7) 17 (10.5) 76 (46.9) 61 (37.7) 7 ( 4.3) 1 ( .6) Source-Respondent Age Difference  Source younger than respondent 52 (13.1) 43 (14.8) 30 (18.5) Source les s than 6 yrs. older 109 (27.5) 72 (24.7) 34 (21.0) Source 6 to 10 years older 89 (22.4) 58 (19.9) 30 (18.5) Source 11 to 20 years older 103 (25.9) 89 (30.6) 44 (27.2) Source over 20 years older 40 (10.1) 27 ( 9.3) 22 (13.6) Unknown 4 ( 1 . 0 ) 2 ( . 7 ) 2 ( 1 . 2 ) 144 Table 26 Relationship between Organizational P o s i t i o n of Source and Mentor  D e f i n i t i o n s , Supportive Behaviours, and Employee Outcomes Higher than F i r s t Source Co-workers Supervisors Supervisors D e f l : Yes 31 (57.4) 185 (71.7) 20 (71.4) No 23 (42.6) 73 (28.3) 8 (28.6) Chi-squared (2 df) = 4.36, ns. Def2: Yes 24 (44.4) 169 (63.3) 14 (51.9) No 30 (55.6) 98 (36.7) 13 (48.1) Chi-squared (2 df) = 7.33, p. < -05. TSBS-1: Mean 76.25 89.22 93.03 s.d. 34.74 30.55 34.00 F(2, 342) = 4.09, p < -05. Multiple comparisons: Supervisors > Co-workers. J-R: Mean -.33 .06 -.04 s.d. 1.07 .95 1.18 F(2, 354) = 3.59, p < .05. Multiple comparisons: Supervisors > Co-workers. S: Mean s.d. F(2, 342) = 9.89, E < - .48 .87 .001. .07 1.00 .42 .88 Multiple comparisons: Supervisors, Higher than Supervisors > Co-workers. P: Mean s.d. F(2, 325) = 4.36, p < Multiple comparisons: -.37 .93 .05. Supervisors .05 .91 > Co-workers. -.06 1.12 Note. Defl = Does (did) t h i s person take a personal i n t e r e s t i n your career and guide or sponsor you? Def2 = Does (did) t h i s person serve as a career r o l e model and a c t i v e l y advise, guide, and promote your career and tra i n i n g ? TSBS-1 = Total Supportive Behaviour Score f or the f i r s t source. J-R = Job Related Composite. S = S k i l l Development Composite. P = Promotional Composite. 145 The results in Table 26 indicate that supervisors were more likely to be viewed as mentors than were co-workers (but only when the specific definition of mentors was used), and that supervisors provided higher levels of supportive behaviours than did co-workers. Because supportive behaviours and employee outcomes were positively related, it was not surprising to find that employees who identified co-workers as their first source of supportive behaviours exhibited significantly lower levels of employee outcomes on all three outcome composites than did employees who identified supervisors as their first source of supportive behaviours. It is surprising, however, that even though supervisors (a) provided higher levels of supportive behaviours than did co-workers, and (b) were more likely to be viewed as mentors than were co-workers, the level of friendship with co-workers was significantly higher than the level of friendship with supervisors (see Table 24 - first source). This pattern of results suggests that friendship does not necessarily imply a high level of supportive behaviours and that a high level of supportive behaviours does not necessarily imply friendship, ln other words, it is possible to be the recipient of a high level of supportive behaviours from a person without having a high level of friendship with that person (as with a supervisor who is viewed as a mentor). It also is possible to be the recipient of a relatively low level of supportive behaviours from a person and still have a high level of friendship with that person (as with a co-worker who is not viewed as a mentor). This suggests that high levels of friendship need not characterize mentor-protege relationships. Gender differences. Four groups of respondents were compared on the definitions of mentors, level of supportive behaviours received, and level of employee outcomes: (a) male source (of supportive behaviours) - male respondent, (b) male source - female respondent, (c) female source - male respondent, and (d) female source - female respondent. As the results in Table 27 indicate, these four groups of respondents were very similar in relation to the likelihood of the source being viewed as a mentor, the level of supportive behaviours received, and the level of employee outcomes. The only significant difference among the four groups was for the Promotional Composite; women with male sources of support had a significantly higher Promotional Composite than did men with male sources of support and women with female sources of support. This result should be interpreted with caution because of the small number of women with female sources of support (n = 38). The primary reason for the significant differences on the Promotional Composite appears to be due to the fact that women in the present study had significantly higher Promotional Composites (M = .41) 146 Table 27 Relationship between Source-Respondent Gender Differences and Mentor  D e f i n i t i o n s , Supportive Behaviours, and Employee Outcomes F i r s t Source MS-MR MS-FR FS-MR FS-FR D e f l : Yes 185 (70.6) 46 (66.7) 6 (66.7) 26 (70.3) No 77 (29.4) 23 (33.3) 3 (33.3) 11 (29.7) Chi-squared (3 df) = .45, ns. Def2: Yes 154 (57.5) 43 (61.4) 4 (44.4) 28 (73.7) No 114 (42.5) 27 (38.6) 5 (53.6) 10 (26.3) Chi-squared (3 df) = 4.59, ns. TSBS-1: Mean 86.89 88.51 83.89 87.72 s.d. 32.25 28.87 28.40 30.18 F(3, 378) = .09, ns. J-R: Mean -.04 .10 .10 .05 s.d. 1.01 .95 .79 1.08 F(3, 391) = .42, ns. S: Mean .01 .02 .01 -.17 s.d. .98 1.05 1.05 1.06 F(3, 378) = .35, ns. P: Mean -.15 .59 .14 .04 s.d. .91 1.22 .72 .92 F(3, 361) = 10.29, 2 < -001. Multiple comparisons: MS-FR > MS-MR and MS-FR > FS-FR. Note. Defl = Does (did) t h i s person take a personal in t e r e s t i n your career and guide or sponsor you? Def2 = Does (did) t h i s person serve as a career r o l e model arid a c t i v e l y advise, guide, and promote your career and training? TSBS-1 = Total Supportive Behaviour Score f or the f i r s t source. J-R = Job Related Composite. S = S k i l l Development Composite. P = Promotional Composite. MS-MR = Male Source, Male Respondent. MS-FR = Male Source, Female Respondent. FS-MR = Female Source, Male Respondent. FS-FR = Female Source, Female Respondent. 147 than did men (M = -.15), r(364) = 4.82, p < .001. This difference likely is explained by the higher rate of promotions for women who had their jobs reclassified on the basis of job evaluation in the company in the energy sector. (This information was obtained from a personnel representative employed by the company in the energy sector.) These results are of interest for a variety of reasons. First, men and women in the present sample were provided with virtually identical levels of supportive behaviours and were equally likely to have received these behaviours from someone who was viewed as a mentor (see Table 27). Second, whether these supportive behaviours were received from a man or from a woman appeared to have little effect on (a) whether this person was viewed as a mentor, (b) the level of supportive behaviours received from this person, and (c) the level of employee outcomes (in two of three outcomes). While there may be special problems associated with cross-gender relationships in the work place, these problems do not appear to be reflected in the situations of respondents in the present study. Organizational Tenure. The establishment, advancement, and maintenance stages used in the present analysis were defined in terms of organizational tenure. Respondents were classified into one of these stages on the basis of when, during their tenure, they first encountered their first source of supportive behaviours: establishment (0 to 2 years tenure), advancement (3 to 10 years tenure), and maintenance (over 10 years tenure). Comparisons of respondents who first met their sources of supportive behaviours at different stages revealed striking similarities (see Table 28). Although the majority of sources were first encountered during the establishment stage, sources met at any one stage were just as likely to be viewed as mentors as sources met at other stages. Also, the level of supportive behaviours received from the first source of supportive behaviours was not related to the stage in which the source was first met. In relation to employee outcomes, employees who first met their sources of supportive behaviours during the establishment stage did not have higher employee outcome composites than did employees who first met their sources of supportive behaviours at either of the two other stages. These results suggest that employees in any stage of organizational tenure can benefit from the supportive behaviours received from other individuals. It is not necessarily more important, as far as employee outcomes are concerned, to be the recipient of supportive behaviours during the early years of organizational tenure than at Table 28 Relationship between Stage i n which Source was Met and Mentor  D e f i n i t i o n s , Supportive Behaviours, and Employee Outcomes Stage i n which Source was Met F i r s t Source E s t a b l i s h . Advancement Maintenance De f l : Yes 129 (67.5) 90 (70.3) 11 (91.7) No 62 (32.5) 38 (29.7) 1 ( 8.3) Chi-squared (2 df) = 3.17, ns. Def2: Yes 113 (58.5) 79 (59.4) 10 (76.9) No 80 (41.5) 54 (40.6) 3 (23.1) Chi-squared (2 df) = 1.71, ns . TSBS-1: Mean 84.65 90.59 93.50 s.d. 31.63 30.84 28.66 F(2, 334) = 1.69, ns. J-R: Mean .01 -.02 .19 s.d. 1.02 .99 .54 F(2, 344) = .27, ns. S: Mean -.08 .17 -.19 s.d. 1.02 .98 1.03 F(2, 333) = 2.60, ns. P: Mean s.d. F(2, 318) .01 1.14 .04 .77 .22 .74 .36, ns. Note. Defl = Does (did) t h i s person take a personal i n t e r e s t i n your career and guide or sponsor you? Def2 = Does (did) th i s person serve as a career r o l e model and a c t i v e l y advise, guide and promote your career and t r a i n i n g ? TSBS-1 = Total Supportive Behaviour Score for the f i r s t source. J-R = Job Related Composite. S = S k i l l Development Composite. P = Promotional Composite. 149 later stages. Age differences. In order to understand how age discrepancies between respondents and their sources of support were related to the definitions of mentors, level of supportive behaviours received, and employee outcomes, the five age discrepancy groups described in Table 25 were compared (see Table 29). In general, sources of supportive behaviours who were between 11 and 20 years older than respondents were more likely to be viewed as mentors than were other sources. Although this group of sources also provided higher levels of supportive behaviours than did the sources in the other age discrepancy groups, this difference was not significant (p > .05). Respondents with sources less than 10 years their senior had lower scores on the employee outcome composites than did the two groups of respondents whose sources were more than 10 years their senior. However, there was only one significant difference between any two of the five groups; respondents with sources between 11 and 20 years older than themselves scored higher on the Job-Related Composite than did respondents who had sources who were less than 6 years older than themselves. Taken as a group, the findings concerning age discrepancy suggest that employees with sources of supportive behaviours who are at least 10 years older than themselves will be the recipients of higher levels of supportive behaviours and exhibit higher levels of employee outcomes than will employees who have less senior sources of supportive behaviours. Given that many supervisors are over 10 years older than their subordinates, the present findings reinforce those presented earlier; namely, employees who had supervisors as sources of supportive behaviours were the recipients of high levels of supportive behaviours and exhibited high levels of employee outcomes. T a b l e 29 R e l a t i o n s h i p between Source-Respondent Age D i s c r e p a n c i e s and Mentor  D e f i n i t i o n s , S u p p o r t i v e B e h a v i o u r s , and Emplpyee Outcomes Age of Source r e l a t i v e to Respondent F i r s t Source D e f l : Yes No Younger 32 (66.7) 16 (33.3) 0 to 5 v r s . o l d e r 68 (65.4) 36 (34.6) C h i - s q u a r e d (4 d f ) = 13.57, p. < .01. Def2: Yes 26 (53.1) 57 (54.3) No 23 (46.9) 48 (45.7) C h i - s q u a r e d (4 d f ) = 8.12, p. < .10. TSBS-1: Mean 84.02 s.d. 32.46 F(4, 374) = 2.27, p. < .10. 83.90 33. 24 6 to 10 v r s . o l d e r 53 (60.2) 35 (39.8) 50 (56.8) 38 (43.2) 83. 76 30.48 11 to 20 v r s . o l d e r 81 (83.5) 16 (16.5) 72 (71.3) 29 (28.7) 95.06 28. 36 J-R: Mean .03 -.19 -.07 s.d. 1.06 1.01 .99 F(4, 387) = 2.70, p. < 05. M u l t i p l e comparisons: 11 to 20 y r s . o l d e r > 0 to 5 y r s . o l d e r 24 91 over 20 v r s . o l d e r 26 (70.3) 11 (29.7) 23 (59.0) 16 (41.0) 88. 18 32.87 .08 1 .00 O T a b l e 29 c o n t i n u e d R e l a t i o n s h i p between S o u r c e - R e s p o n d e n t Aae D i s c r e p a n c i e s and Mentor D e f i n i t i o n s . S u p p o r t i v e B e h a v i o u r s , and Employee Outcomes Age of S o u r c e r e l a t i v e to Respondent F i r s t S o u r c e Younger S : Mean - . 0 9 s . d . 1.18 F ( 4 , 375) = 2 . 9 4 , n s . P: Mean -.1.3 s . d . .94 F ( 4 , 358) = 2 . 0 7 , a < -10 . 0 to 5 y r s . o l d e r - .06 1 .05 . 12 .82 6 to 10 y r s . o l d e r - . 13 .98 .08 .97 11 to 20 v r s . o l d e r .23 .81 .22 1.18 o v e r 20 v r s . o l d e r .05 1 .05 . 12 1.01 N o t e . Def1 = Does ( d i d ) t h i s p e r s o n take a p e r s o n a l i n t e r e s t in your c a r e e r and g u i d e or s p o n s o r you? Def2 = Does ( d i d ) t h i s p e r s o n s e r v e as a c a r e e r r o l e model and a c t i v e l y a d v i s e , g u i d e , and promote your c a r e e r and t r a i n i n g ? TSBS-1 = T o t a l S u p p o r t i v e B e h a v i o u r S c o r e f o r the f i r s t s o u r c e . J -R = J o b - R e l a t e d C o m p o s i t e . S = S k i l l Development C o m p o s i t e . P = P r o m o t i o n a l C o m p o s i t e . 152 Chapter 6  Discussion and Conclusions The primary purpose of this study was to investigate the relationship between the receipt of mentoring (renamed as supportive) behaviours and employee outcomes. The study assessed employee outcomes (job- and career-related) and the quantity of different types (career and psychosocial) of supportive behaviours received by employees from different sources. A secondary purpose was to investigate the relationship between the employee outcomes and the way in which the supportive behaviours were dispersed across the sources of supportive behaviours. The results of this study lead to the general conclusions (a) that the supportive behaviours received by employees, from whatever source, are positively and significantly related to employee outcomes; (b) that the way in which the supportive behaviours are dispersed across sources explains little of the variation in employee outcomes over and above that explained by the level of supportive behaviours received (i.e., more is better); and (c) that the hazards commonly associated with mentor-protege relationships are negatively related to employee outcomes (job-related outcomes, in particular) for given levels of supportive behaviours. Based on a review of the mentoring literature, eight functions that mentors commonly provide proteges were identified: Sponsoring, Exposure and Visibility, Teaching the Job, Teaching the Informal System, Protection, Role Modeling, Encouragement, and Personal Counselling. The first five of these functions can logically be grouped into what Kram (1985) termed career functions. Kram (1985) suggested that these functions could be instrumental for career development (e.g., advancement in an organization). The latter three functions are similar to what Kram (1985) termed psychosocial functions. Kram (1985) suggested that these functions were more related to personal development in terms of competence and effectiveness. Respondents in this study were asked to indicate the extent to which people with whom they had worked had provided them with behaviours associated with the functions. Five behaviours identified in the mentoring literature were used to assess each function. The mentoring functions and behaviours described above were renamed supportive  functions and supportive behaviours. This was done because (a) a review of the work-related social support literature indicated that what some writers have termed 153 mentoring functions are conceptually similar to different kinds of social support (e.g., structural, informational, and emotional), and (b) the mentoring literature indicated that employees can be the recipients of mentoring functions from people who may not be viewed as mentors (e.g., co-workers). By using the term 'supportive' instead of 'mentoring' both of these points are recognized explicitly. Structure of Supportive Behaviours The first question that was addressed empirically in this study was the structure of the behaviours associated with the supportive functions. It was found that the supportive functions did not form two distinct categories corresponding to the career and psychosocial functions identified by Kram (1985). In each of eight analyses for different subgroups of respondents, the first principal component accounted for over half of the variance and the second component accounted for little additional variance. Because of these findings, most analyses in this study collapsed across type of supportive function and considered only the total level of supportive behaviours received from any particular source. It is important to note that failure to identify empirically distinct categories of supportive behaviours does not imply that the supportive behaviours under consideration are not conceptually distinct. The analyses presented earlier suggest that respondents who received relatively high levels of behaviours associated with a particular function also received high levels of behaviours associated with the other functions relative to other  respondents. In other words, the high correlations among the eight supportive behaviour scales imply that there is a great deal of overlap in how respondents are rank-ordered from one function to the next. The high correlations do not imply, however, that respondents failed to distinguish among the behaviours associated with the different functions. Such a possiblity would have had to be entertained if the means and standard deviations for the different supportive behaviour' scales had been highly similar. However, the pattern of means for the different scales across the three different sources (see Tables 6 to 8) suggests that neither halo nor central tendency can be used to explain the finding of one general category of supportive behaviours. The means for the different supportive behaviour scales, differed by well over one standard deviation for all three sources of support. Further, the rank-ordering of the means within the three sources of support was virtually identical. This suggests that some supportive behaviours (e.g., those associated with Encouragement) 154 are more likely to be provided in work environments than are others (e.g., those associated with Teaching the Informal System). The high correlations among the supportive behaviour scales, as noted before, indicate that respondents who received relatively high levels of behaviours associated with one function also received relatively high levels of behaviours asscociated with the other functions. The high correlations do not necessarily imply that the behaviours associated with one function are conceptually similar to the behaviours associated with other functions. Structure of Employee Outcomes In order to investigate the relationship between supportive behaviours and employee outcomes, two types of employee outcomes (job-related and career-related) were assessed. Most of the job-related outcomes were suggested by the literature that indicates that different outcomes may be important to employees at different career stages (job satisfaction, role ambiguity, role conflict, acceptance by co-workers, organizational commitment). The career-related outcomes (skill development, rates of salary increase and promotions, satisfaction with progression) were included because they were concerned with development over time. These outcomes were considered to be especially important because the purported beneficial effects of being the recipient of supportive behaviours usually are conceptualized in terms of individual (personal) or career development as well (Kram, 1985). The conceptual distinction between the job-related and career-related outcomes received only partial empirical support. Although the job-related outcomes formed a distinct group of outcomes, the career-related outcomes were found to fall into two conceptually distinct groups: skill development outcomes and promotional outcomes. Although one can only speculate on why the career-related outcomes did not load together on one component, a possible explanation lies in the nature of the reward systems in the organizations included in the sample. Of particular interest is the organization in the energy industry because this organization employed over 80% of the survey respondents. Approximately 45% of the employees surveyed in this organization were unionized. Salary increases for these employees are based on a lock-step system in which increases are given at predetermined times until employees attain the maximum salary for their particular job categories; this takes approximately four years. Salary increases are withheld only for unsatisfactory performance. After employees have reached 155 the top of their respective job categories, the only salary increases they receive are based on settlements negotiated yearly. Salary increases for unionized employees are never based on merit. This means that good performance can be rewarded only by promotion to higher job categories. There is, however, little opportunity for this because this organization substantially reduced the number of full-time staff over the past seven years. This has resulted in many employees being 'top-ended' in their particular job categories. The situation is similar for nonunionized personnel. For this group of employees, salary increases are based on yearly negotiated settlements in conjunction with potential merit increases. For any particular job category, the difference between the highest and lowest salary is approximately 15%. Depending on the merit category assigned on the basis of a yearly performance review, employees receive either (a) no increase, (b) a 2% to 3% increase, or (c) a 5%, 6%, or 7% increase. After employees have reached the top of the salary scale for their job categories, all increases are based solely on the yearly negotiated settlement. Employees with the highest merit category receive lump sum payments, but their annual salary does not increase. As is the case with unionized personnel, many nonunionized employees are 'top-ended' in their job categories with little opportunity for advancement because of reduction in the work force in this organization. Given such a situation, the failure to identify one component for the career-related outcomes may not be surprising. Since the organization in the energy industry-which contributed substantially to the present sample-is constrained in the way it rewards employees, skill development is not necessarily rewarded with promotions and salary increases. Empirically, this would lead to the obtained result of separate components for skill development outcomes and for promotional outcomes. A review of mentoring studies conducted to date indicated that employee outcomes have been poorly conceptualized. Although both job-related and career-related outcomes have been assessed previously, the conceptual distinction between outcomes pertaining to current situations and outcomes pertaining to development has not been made. The results of this study indicate that an even finer distinction among various types of outcomes may be warranted. In particular, it appears that personal development can be achieved with or without career development. This finding suggests that future work on the relationship between supportive behaviours and employee outcomes should distinguish among the different types of outcomes. Such an emphasis seems to be particularly important because different mentoring functions have been assumed to be 156 related to different types of developmental outcomes (Kram, 1985). Kram (1985) was somewhat vague about how outcomes related to career functions differed from outcomes related to psychosocial functions. It appears, however, that the major difference revolves around whether the outcomes concern career development (i.e., like the promotional outcomes) or whether they concern personal development (i.e., in terms of skill development). Until the relationship between supportive behaviours and different types of employee outcomes is more fully understood, future work should consider explicitly which outcomes are of primary interest. Relationship between Supportive Behaviours and Employee Outcomes It was previously suggested that supportive behaviours should be less highly related to job-related outcomes than to career-related outcomes. The argument was made that behaviours that are assessed in relation to a certain period of time (in this case, time since joining the present organization) should be more highly related to outcomes that also concern this time period than to outcomes that are concerned with employees' present situations. For this reason, the finding that supportive behaviours were more highly related to the skill development outcomes than to the job-related outcomes is not surprising. However, the promotional outcomes, which were assessed in relation to the same period of time as were the supportive behaviours and the skill development outcomes, were less highly related to the supportive behaviours than were the skill development outcomes. The reason for the relatively low, albeit significant, relationship between the promotional outcomes and supportive behaviours may, again, be related to the nature of the reward systems in the organizations studied. As noted previously, because of constraints in the way employees can be rewarded for good performance, promotions and salary increases may be largely institutionalized (i.e., yearly negotiated settlements with little opportunity for advancement). This suggests that individual employee behaviour and initiative may have little impact on at least two of the promotional outcomes-promotion rate and rate of salary increase. This may explain why the relationship between supportive behaviours and the promotional outcomes is lower than the relationship between supportive behaviours and the skill development outcomes; the latter are likely within the control of the individual employee, while the former are likely determined largely, at least in this study, by organizational and economic factors. 157 The failure to find a higher relationship between supportive behaviours and the promotional outcomes than between supportive behaviours and the job-related outcomes, points to one possible limitation of the present study. Many of the employees sampled in this study were employed by their organizations in a recessionary period that began around 1981 and that, to some extent, continues to the present day. During this recessionary period many organizations in British Columbia, the organizations in the presented study included, were forced to lay off significant proportions of their work forces. This has the effect of limiting the promotional opportunities of the remaining employees over the course of the recessionary period. In a statistical sense, the variance that can be expected in promotional outcomes during a recessionary period is less than the variance that might be expected in a growing economy. For this reason, the relationship between promotional outcomes and supportive behaviours obtained in the present study might be lower than the relationship that would be obtained in a similar study conducted during a period of economic growth. The supportive behaviours assessed in this study were suggested by the literature concerning the functions that mentors provide on behalf of their proteges. One of the major reasons behaviours were assessed in this study was that it was felt that employee outcomes could be better explained by knowledge of the level of supportive behaviours received by employees than by the knowledge of whether the providers of these behaviours were simply recognized as mentors. For comparative purposes, the variance in employee outcomes explained (a) by supportive behavious received by employees and (b) by employee responses to two definitions of mentors were examined. Although the level of supportive behaviours received and the knowledge of whether the sources of these behaviours were recognized as mentors were related approximately equally to the job-related and promotional outcomes, skill development outcomes were more highly related to the level of supportive behaviours received than to knowledge of whether these behaviours were received from a mentor. This pattern of findings (for supportive behaviours and for knowledge about mentors) suggests that the assessment of supportive behaviours on a continuum represents a refinement over the assessment of mentoring as an all-or-none phenomenon for understanding employee outcomes. Knowledge of the level of supportive behaviours received by employees is more highly related to employee outcomes than is the knowledge of whether the behaviours were received from someone who is perceived as a mentor. 158 Knowledge of the level of supportive behaviours received by employees also is of more practical importance than is knowledge of whether employees have mentors. Mere knowledge of whether or not employees have mentors does not indicate the level of support that such employees receive or do not receive in relation to the individual supportive functions. As will be discussed later, knowledge of which functions and behaviours are being provided can be used for diagnostic and remedial purposes. In contrast, knowledge of whether or not employees have mentors is of more limited practical utility. Incremental Effect of Multiple Sources of Behaviours on Employee Outcomes Many writers (e.g., Darling, 1985; Halcomb, 1980) have noted that individuals who provide a wide range of supportive functions and behaviours are relatively rare in modern organizations. This has led some writers (e.g., Kram, 1985; Missirian, 1980; Phillips-Jones, 1982) to suggest that employees who look to a variety of people for functions and behaviours can derive the same benefits as can employees who rely on a single provider of high levels of these functions and behaviours. For this reason, this study assessed the level of supportive behaviours and the number of functions received by employees from multiple sources. It was hypothesized that the level of employee outcomes would be positively related to the level of supportive behaviours received from secondary and tertiary sources after the level of supportive behaviours received from the primary source was taken into consideration. Even though the level of supportive behaviours received from each source was positively related to all three types of employee outcomes, the level of supportive behaviours received from sources other than the primary one explained virtually no additional variance in employee outcomes. In relation to supportive functions, however, the hypothesis was supported. Employee outcomes were positively related to the number of supportive functions provided by the primary source, but even more highly related to the number of supportive functions provided by all sources combined. It appears that failure to find incremental effects due to supportive behaviours received from secondary and tertiary sources was due to a methodological shortcoming of the present study. Speculation concerning why analyses based on supportive functions exhibited incremental effects on employee outcomes, while analyses based on supportive behaviours failed to do so, illustrates this flaw. 159 It appears that the major reason for the obtained pattern of results was due to methodological artifact. The correlations between supportive behaviours received from the highest source and behaviours received from the other source(s) were in the .70 to .80 range. These high correlations were most likely due to the instructions concerning whether additional sources of support should be rated on the supportive behaviours. In an effort to ensure that respondents completed the survey, additional sources were rated only if they had provided three or more functions. This was requested of respondents because the pilot study, which had an extremely low response rate, indicated that few supportive behaviours were provided by people who provided fewer than three functions. However, because of these instructions, additional sources of supportive behaviours could have been rated on the supportive functions only if the highest source of functions provided three or more functions (i.e., if the highest source provided fewer than three functions, no additional sources were rated). This procedure, then, resulted in an artifactually high relationship between the behaviours provided by the highest source and behaviours provided by additional sources. For the supportive functions, this was not the case. The correlations between the number of supportive functions received from the highest source and the number of functions received from all other sources were in the .40 range. These intercorrelations were lower than those for the behaviours because additional sources of functions were identified regardless of how many functions were provided by the highest source. Because of this, problems of multicollinearity were not present in the analyses concerning functions. This likely explains why analyses concerning the incremental effects of functions were significant, while those for behaviours were not significant. In order to overcome the problem of multicollinearity associated with the level of supportive behaviours provided by the different sources of supportive behaviours, future studies should assess the level of supportive behaviours received from sources who provided only one or two supportive functions as well as the level of supportive behaviours received from sources who provided a higher number of functions. In this way a broader range of supportive behaviours would likely be realized, and the level of supportive behaviours received from one source would not likely be artifactually related to 160 the level of supportive behaviours received from other sources. Effects of Dispersion of Supportive Behaviours on Employee Outcomes Based on the literature concerning the hazards and disadvantages of mentor-protege relationships, it was hypothesized that the more a given level of supportive behaviours received by employees emanates from only one or two sources, the lower would be the employee outcomes. The results indicated that the way in which supportive behaviours were dispersed across sources explained virtually no additional variance in employee outcomes beyond that explained by the level of supportive behaviours received. These findings suggest that the relationship between supportive behaviours and employee outcomes was accounted for more by the level of the behaviours than by the way the behaviours were distributed across sources. The practical conclusion that might be drawn from these results is that employees should seek supportive behaviours without concern about their source. Because of the methodological problem concerning which sources of supportive functions were rated on the supportive behaviours, such a conclusion may not be warranted. Failure to find a relationship between employee outcomes and the dispersion of supportive behaviours across sources may be due to restricted range in the dispersion of behaviours across sources in the present sample. Analyses concerning dispersion of behaviours involved only those respondents who were provided with three or more functions by each of at least two people. It was noted earlier that, among respondents who rated more than one source, the level of supportive behaviours received from any one source was highly related to the level of supportive behaviours received from other sources. This finding, taken in conjunction with the fact that mean levels of supportive behaviours received from each of the three sources were highly similar, suggests that the supportive behaviours were distributed across sources relatively evenly. The distribution of dispersion in this sample of respondents may have had a wider range had respondents been asked to rate additional sources of supportive behaviours regardless of how many functions these sources provided. If this had been done, more respondents may have had higher levels of dispersion of behaviours across sources. For example, if sources had been rated regardless of the number of functions provided, the analyses involving dispersion would have included respondents who had been provided with a high number of functions by the first source, but only one or two functions by additional sources. Because the level of supportive behaviours was positively related to the number of 161 functions provided, such respondents likely would have had the highest levels of dispersion. To more fully evaluate the relationship between dispersion of supportive behaviours across sources and employee outcomes, future studies should assess the level of supportive behaviours received from multiple sources regardless of the number of functions provided by these sources. If these studies replicate the present findings, the distribution of behaviours across sources may not be important. For now, this is a tentative conclusion. Relationship between Hazards and Employee Outcomes A substantial literature exists that suggests that employees who are the recipients of high levels of supportive behaviours can expect these behaviours to be accompanied by a number of hazards that may be negatively related to employee outcomes. For this reason, it was hypothesized that the hazards associated with mentor-protege relationships would be positively related to the level of supportive behaviours received and negatively related to employee outcomes. The hypothesis concerning supportive behaviours was supported; high levels of supportive behaviours were associated with high levels of hazards. The hypothesis concerning employee outcomes was only partially supported. Although hazards were negatively related to the job-related outcomes (as hypothesized), they were positively related to the skill development outcomes and virtually unrelated to the promotional outcomes. It was suggested earlier that the hazards may have to be present for skill development to occur, but that the hazards may have negative consequences for employees in their day-to-day matters. Another possible explanation for the obtained pattern of results suggests itself. Because most of the hazards assessed in this sudy concerned negative attributions of the focal employee made by other organizational members, it is possible that hazards arise only after employees have developed skills. In other words, other employees may be envious and attribute negative characteristics to employees who, unlike themselves, have been able to develop personally. Such envy and negative attributions may lead other employees to not accept the recipient of high levels of supportive behaviours. This, in turn, might lead to low levels of job satisfaction and organizational commitment, and high levels of role conflict and ambiguity for the recipient of the high levels of support. While it is not possible to infer causality based on the responses to this study, the results suggest that employees who are the recipients of 162 high levels of support and who, at the same time, are able to avoid the hazards associated with mentor-protege relationships, may be able to realize higher levels of employee outcomes than will employees who are unable to avoid the hazards. This conclusion is supported by the findings that showed that hazards explained significant amounts of incremental variance in employee outcomes (the job-related ones, in particular) for given levels of supportive behaviours (see Table 23). The practical implication of these findings may well be that, in order to realize the highest level of employee outcomes, attention should be directed toward minimizing the level of hazards and, simultaneously, maximizing the level of supportive behaviours. As will be discussed later, the most effective way to accomplish these goals may be to deliver the supportive behaviours in a way that reduces perceptions of favouritism on the part of other organizational members. Relationships among Friendship, Characteristics of Sources of Supportive Behaviours, and  Employee Outcomes In an effort to understand how the characteristics of sources of supportive behaviours were related to employee outcomes, respondents were asked to report each source's age, gender, and organizational position. The level of friendship with these sources was also assessed. In general, gender differences between respondents and their sources and the stage of organizational tenure in which the sources were first met were unrelated to the level of employee outcomes and the level of supportive behaviours provided by the sources. On the other hand, the organizational position of the source of supportive behaviours relative to the respondent's position was significantly related to the level of friendship with that source, the level of supportive behaviours received, and employee outcomes. These analyses involved comparisons among sources of support who were co-workers, supervisors, and organizational members at higher levels than respondents' supervisors. Although the level of friendship with co-workers was higher than with .the other two groups of sources, respondents who received their supportive behaviours from supervisors received higher levels of support and reported higher levels of employee outcomes than did respondents who received their supportive behaviours from co-workers. Respondents who received their supportive behaviours from higher level organizational 163 members were more like respondents who received their behaviours from supervisors than they were like respondents who received their behaviours from co-workers. However, because relatively few sources were higher level members, few comparisons involving this group were significant. These results suggest that, although friendship is positively related to the level of supportive behaviours received, high levels of friendship do not necessarily imply high levels of supportive behaviours. In other words, it appears that the organizational position of a source of supportive behaviours is more informative in relation to level of supportive behaviours received and employee outcomes than is level of friendship. This interpretation does not imply that employees should look only to supervisors as sources of supportive behaviours. The relationship between supportive behaviours and employee outcomes was approximately the same for co-workers as a group as for supervisors as a group; for any particular outcome the correlations differed by less than .06. This result suggests that employees should look to supervisors for support first because supervisors can be expected to provide the highest levels of supportive behaviours. Employees who are unable to receive supportive behaviours from supervisors probably benefit from seeking these behaviours from co-workers. Limitations of the Study Sample. Of the 397 respondents whose responses were used in the analysis 325 (81.9%) were employed by one organization (see Table 4). This suggests that this study comes close to what might be viewed as a case study. Because of this, generalizations and recommendations based on the findings should be made with considerable caution. The relatively small numbers of respondents from each of the other four organizations (maximum sample size was 29) made it impossible to perform separate analyses for each organization. To address this issue of generalizability, the correlations between the supportive behaviours received from the first source (i.e., TSBS-1) and the three types of employee outcomes were examined more closely. These correlations were computed for all respondents employed by the company in the energy industry (n = 325) and, separately, for all of the remaining respondents (n = 72). The correlations between supportive behaviours and the three employee outcome composites were (energy industry first): .24 and .21 for the Job-Related Composite, .47 and .39 for the Skill Development Composite, and .23 and .30 for the Promotional Composite. The pattern of correlations is highly 164 similar for the two subgroups of respondents and similar to the pattern of relationships discussed earlier (i.e., a higher relationship between supportive behaviours and the skill development outcomes than between supportive behaviours and the other two types of employee outcomes). While this is only one comparison, it involves the finding that is, perhaps, most central to the present study; namely, the relationship between supportive behaviours and employee outcomes. The results suggest that this relationship is a fairly general one. However, additional research is needed in order to document that supportive behaviours and employee outcomes are positively related in different organizational settings. Causality. Implicit in much of the mentoring literature and, indeed, in much of the material discussed in this study, is the notion that supportive functions and behaviours lead to, or cause, employee outcomes. Rarely has it been suggested, for instance, that people might provide Sponsoring and Encouragement to an individual precisely because that individual had demonstrated a high level of skill development in the past. Because this study was cross-sectional, the issue of the directionality of the relationship between supportive behaviours and employee outcomes cannot be addressed. As noted by Thoits (1982), in order to assess the notion of causality explicitly, it would be necessary, at the very least, to assess supportive behaviours and employee outcomes at two separate points in time. Because supportive behaviours may be received at any point over a relatively long period of time (e.g., perhaps ten years), assessment would probably be required at numerous time points. It then would be possible to relate the level of supportive behaviours received within each of a number of time periods to employee outcomes evidenced both before and after these time periods. Depending on the obtained pattern of relationships, inferences could be drawn concerning the directionality of the supportive behaviour-employee outcome relationship. r Studies of this nature are clearly needed. In an experimental study that used written materials to describe the performance of hypothetical subordinates (proteges), Carroll, Olian, and Ciannantonio (1988) found that subordinates with high levels of past performance were more likely to be provided with friendship and career enhancing (i.e., mentoring) behaviours than were subordinates with moderate levels of past performance. This finding needs to be replicated in a field setting. In order to evaluate how past performance is related to the level of supportive behaviours employees receive, future studies should incorporate the timing of performance 165 appraisals into their design. In the case of yearly performance appraisals, employees might be asked, just prior to each appraisal, about the extent to which they had been the recipients of supportive behaviours in the past year. If performance leads to supportive behaviours, then the relationship between behaviours received in the past year and the previous year's performance rating should be higher than the relationship between the behaviours received in the past year and the current performance rating. In contrast, if supportive behaviours lead to performance, the relative magnitudes of these relationships should be reversed (i.e., a higher relationship between past behaviours and current performance than between present behaviours and past performance). A study designed to assess such relationships would represent a first step in understanding the directionality of the relationship between support and employee outcomes. To date, the mentoring phenomenon is not a very well understood phenomenon. The relationship between mentoring (supportive) behaviours and employee outcomes has not been well documented in the studies that have been undertaken. The results of this study indicate quite clearly that supportive behaviours and employee outcomes are positively and significantly related. Thus, the results of this study contribute to the understanding of the nature of supportive relationships in the work place. Additional research is clearly needed in order to understand the precise relationship between supportive behaviours and employee outcomes. Retrospective rationality. Results of mentoring studies have been criticized because they often rely on the, perhaps, imperfect or distorted memories of the respondents involved (Speizer, 1981). This clearly presents a problem for accurately interpreting the results of such studies. In Roche's (1979) study, for instance, the average age of respondents was approximately 48 years. Given that most of the respondents first encountered their mentors during the first five years of their careers, a number of respondents would have been commenting on events that transpired over 20 years earlier. Memory decay and distortion may have affected what was reported. In this study, the problem of retrospective rationality was limited (but clearly not overcome) by restricting the sample to employees who had been employed by their organizations for 15 years or fewer. This procedure had the effect of limiting recall of events to a maximum of 15 years. Because respondents had known their sources of supportive behaviours for an average of approximately seven years, the time between recall of events (February 1988) and when the events took place was often much less 166 than 15 years. For this reason, memory decay and distortion should be of less influence in this study than in many previous studies. It might be argued that people who retrospectively make sense of their experiences would do so in a fashion that is logically consistent. Empirically, this might result in measures being related to each other based primarily on the desirability of the factors assessed by these measures. In such a case, desirable and undesirable factors would be negatively related, and conceptual distinctions among factors of like desirability would not be demonstrated empirically. The pattern of relationships found among measures in this study suggests that retrospective rationality is of limited utility for explaining the major results. One interesting pattern of relationships that suggests some degree of discriminant validity concerns hazards, supportive behaviours, and employee outcomes. First, the hazards, which have been viewed as negative aspects of mentor-protege relationships, were positively related to the level of supportive behaviours received. If the respondents in this study had been responding in terms of social desirability, one would have expected a negative relationship between hazards and level of support because respondents would have been hesitant to ascribe negative outcomes to the providers of clearly positive behaviours. Second, given (a) the positive relationship between hazards and support, and (b) the positive relationship between support and employee outcomes, a positive relationship between hazards and employee outcomes might be expected. This was found for only the skill development outcomes. There was a consistently negative relationship between hazards and the job-related outcomes and virtually no relationship between hazards and the promotional outcomes. This pattern of relationships was interpreted previously. If the hazards had been found to relate to all employee outcomes in a similar way, then arguments concerning retrospective rationality would have to be entertained. As it is, it is difficult to imagine how retrospective rationality would lead to the obtained pattern of results. Other evidence of discriminant validity is suggested by the structure of employee outcomes and by the way in which supportive behaviours were related to these outcomes. On the one hand, relatively objective outcomes that were assessed in one part of the questionnaire (salary and promotions) loaded on the same component as did a relatively subjectively assessed outcome (satisfaction with progression) that was placed in a 167 different part of the questionnaire. On the other hand, outcomes that were assessed in the same part of the questionnaire and that used identical response formats (interpersonal skill development and role ambiguity) were found to load on different components. These findings suggest that neither halo nor response style can be used as explanations for the obtained pattern of relationships among the outcomes assessed in this study. Further, with the exception of the relationship between supportive behaviours and the promotional outcomes (which was not as high as expected), the supportive behaviours and employee outcomes were related in ways that were both conceptually meaningful and hypothesized a priori. It is difficult to see how halo, response style, or retrospective rationality would result in the obtained pattern of findings. Considering the foregoing indicators of discriminant validity in this study, it appears that the use of a self-report questionnaire that concerned previously experienced events does not represent a major limitation of this study. Spurious correlation. The primary concern in this study was the assessment of the relationship between employee outcomes and the extent to which employees had been the recipients of supportive behaviours and functions. In order to understand these relationships more fully, other factors were considered that may be directly related to both supportive behaviours and employee outcomes. Consideration of these factors allows one to assess whether the relationship between supportive behaviours and employee outcomes was due to spurious correlation. Three such factors suggested themselves: social competence, self-esteem, and technical competence. Reis (1984) suggested that socially competent people could more easily develop meaningful relationships with others as well as cope with stress more effectively. In relation to supportive behaviours and work-related outcomes, a similar situation may exist. First, social competence, because it is required in interactions at work, may be related to outcomes such as promotions. In other words, socially competent people advance and socially incompetent people do not. Second, socially competent people may become the recipients of more supportive behaviours than socially incompetent people because the socially competent have more contact with other people. Thus, social competence may result in both more supportive behaviours and more positive outcomes. The possibility then exists that any relationship between supportive behaviours and employee outcomes can be explained by the factor of social competence. Although not a direct measure of social competence, this study included a measure of extraversion. If it is assumed that 168 social competence is, at least in part, a function of the extent to which people interact with others (as reflected in the extraversion measure), then the pattern of relationships between extraversion and employee outcomes and between extraversion and supportive behaviours should provide an indication of the extent to which the relationship between employee outcomes and supportive behaviours can be explained by the factor of social competence. In relation to self-esteem, it may be the case that individuals who are successful in work-related matters (e.g., performance, promotions) derive a certain sense of self from these accomplishments. In such a case, an individual's self-esteem would be directly related to a variety of work-related outcomes. As for the supportive behaviours, potential providers of such behaviours may recognize a positive sense of self in others and, because of this, be more likely to provide behaviours to employees with high levels of self-esteem than to individuals who have not come to accept themselves. In such a scenario, an employee's level of self-esteem would be directly related to the level of supportive behaviours provided by others. Thus, self-esteem may be positively related to both supportive behaviours and work-related outcomes and could, potentially, explain the relationship between behaviours and outcomes. A similar argument can be made for the level of technical competence that an individual brings to the organization. If ones assumes that individuals who are technically competent when they join an organization would be more likely to succeed than individuals who are not, then technical competence would be related to employee outcomes. This scenario seems likely. It is also possible that technically competent people would more likely be the recipients of supportive behaviours than would technically incompetent people. A variety of writers have suggested that mentors, in choosing proteges, look for high levels of competence (e.g., Kram, 1985; Reich, 1985, 1986; Zey, 1984). This possibility suggests that the relationship between supportive behaviours and employee outcomes might be explained by the third variable of technical competence. Because technical competence was not assessed directly in this study, educational attainment was used as the proxy for initial level of technical competence; educational level was coded on a 5-point scale ranging from 1 = up to high school graduation to 5 = graduate degree (see Appendix D). The correlations between employee outcomes and supportive behaviours would be spurious to the extent that other factors were found to be positively related both to 169 employee outcomes and supportive behaviours. The way in which the three factors of extraversion, self-esteem, and educational level were related to supportive behaviours and employee outcomes (see Table 30) suggests that neither extraversion, self-esteem, nor educational level can be used to explain the obtained relationships between supportive behaviours and employee outcomes in this study. (Only the first source of supportive behaviours was considered in these analyses because the three factors were related similarly to the level of supportive behaviours received from the other two sources). Although each of the three factors was significantly related to the supportive behaviours and/or to one or more of the employee outcomes, the correlations were not high. Because extraversion, self-esteem, and educational level did not explain significant amounts of the variation in the supportive behaviour-employee outcome relationships, the positive and significant relationships between supportive behaviours and employee outcomes documented earlier do not appear to be artifactually inflated. The limitation in this study lies in the possibility that the relationship between supportive behaviours and employee outcomes might be explained by factors other than those included in the present study. For this reason, future studies might assess a wider range of individual difference measures in an attempt to more fully understand the relationship between supportive behaviours and employee outcomes. Quality of supportive behaviours. This study was limited to the assessment of the quantity of supportive behaviours received by employees from various people in their organizations. Although the quantity of supportive behaviours received was positively and significantly related to employee outcomes, it is possible that consideration of the quality of supportive behaviours received might explain additional variance in these outcomes. The possibility exists that some employees may have been the recipients of low levels of supportive behaviours, but that these behaviours had a tremendous impact on their careers (i.e., they were of high quality). An example of such a situation is one in which an employee is sponsored for a position only once, receives that position, and advances steadily thereafter. In the present study, the provider of the sponsoring may not have been rated on the supportive behaviours because only one function was provided. In an attempt to get at the notion of quality, one final exploratory analysis was conducted. The goal of this analysis was to identify and compare respondents who reported receiving high (low) levels of supportive behaviours and functions, but who did not (did) view the providers of these behaviours and functions as mentors according to Table 30 Correlations of Extraversion, Self-esteem, and Educational Level  with Employee Outcome Composites and Supportive Behaviours P o t e n t i a l l y Explanatory Factor n TSBS-1 J-R S _E Extraversion 343 .15** .19*** .27*** .05 Self-esteem 346 .07 .34*** .22*** .09* Educational l e v e l 346 -.09 -.04 .03 .10* Note. TSBS-1 = Total Supportive Behaviour Score f o r the f i r s t source J-R = Job-Related Composite. S = S k i l l Development Composite. P = Promotional Composite. *p < .05, one-tailed. **p_ < .01, one-tailed. ***p < .001, one-tailed 171 the two definitions used in this study. For such respondents, the answers to the definitions would be somewhat contradicted by the level of supportive functions and behaviours received. A discriminant function analysis was used to identify these 'inconsistent' responders. To begin with, two groups of respondents were formed on the basis of the two definitions of mentors for the first source of supportive behaviours: (a) respondents who answered 'yes' to both definitions (n = 198), and (b) respondents who answered 'no' to both definitions (n = 90). The discriminant function analysis used (a) the level of supportive behaviours received from the first source and (b) the number of supportive functions provided by the first source in order to discriminate between the two groups. Discriminant function scores were derived on the basis of the standardized discriminant function coefficents for the two measures (i.e., behaviours and functions). The discriminant function scores were used to classify as many of the respondents as possible into two groups that corresponded to the 'yes-yes' and 'no-no' groups formed on the basis of the definitions of mentors. Of the 288 respondents used in this analysis, 235 (81.6%) were correctly classified. Of interest in this exploratory analysis are the respondents who were incorrectly classified. There were 35 respondents who answered 'no' to both definitions who, according to the level of supportive behaviours and functions received, were predicted to have been in the 'yes-yes' group, ln other words, employees in this group received relatively high levels of supportive behaviours and functions, but did not view the providers of these behaviours and functions as mentors. This group of respondents is referred to as the 'Should be Mentors' (but are not) group. Also misclassified were 18 respondents who answered 'yes' to both definitions of mentors, but who received relatively low levels of supportive behaviours and functions. This group of respondents is referred to as the 'Should not be Mentors' (but are) group. It is this latter group that is of primary interest. Because this group received low levels of supportive behaviours and functions but still viewed the providers of support as mentors, it is possible that the support received was of high quality. Such quality, one might argue, should be reflected in the level of employee outcomes for this group. In an effort to understand the inconsistencies between the level of supportive functions and behaviours received and the responses to the definitions of mentors, the Should be Mentors group was compared to the Should not be Mentors group on age, 172 organizational tenure, age and gender differences between the respondent and the source of support, the three employee outcome composites, self-esteem, and extraversion. The two groups did not differ significantly on any of these measures. The comparisons on the employee outcome measures are of interest because individuals who have had mentors (however defined) are assumed to exhibit higher levels of outcomes than are individuals who have not had mentors. The fact that the Should be Mentors and Should not be Mentors groups did not differ on the employee outcome composites and, further, that both groups scored relatively close to the means for the entire sample on these measures, suggests that the two groups cannot be distinguished in terms of the quality of supportive behaviours received. It should be noted that this analysis and discussion of quality confounds quality with level of employee outcomes (i.e., high quality support is assumed to be reflected in high levels of outcomes). Because this study was primarily concerned with the quantity of supportive behaviours received, this is not a major limitation of the study. However, additional work is clearly needed on how the quality of supportive behaviours differs from the quantity of supportive behaviours in relation to employee outcomes. Focus on organizational tenure. This study assessed the extent to which employees had been the recipients of supportive behaviours during their tenure within one organization. Similarly, the employee outcomes concerned with development (e.g., skill development) and change (e.g., average annual salary increase) were assessed in relation to the time when employees first began working for their current organizations. This focus means that this study investigated the relationship between supportive behaviours and within-organization development. Such within-organization development is equivalent to career development only for those respondents who were not previously employed by other organizations. Given that mentoring is often associated with career development (e.g., Hunt & Michael, 1983; Kram, 1985), the focus on organizational tenure represents a limitation of the present study. Various writers (e.g., Bova & Phillips, 1981; Roche, 1979) have noted that proteges often first meet mentors during early career stages or during the early years of organizational tenure. By focusing on organizational tenure, this study did not assess the level of supportive behaviours received during the early career stages for those respondents who spent the early years of their careers working for organizations other than the current one. If the behaviours received during the early career stages are of 173 primary importance to subsequent employee development, then a lower relationship between supportive behaviours and employee outcomes might be expected for (a) respondents who had spent the early years of their careers working for organizations other than the current one than for (b) respondents who began their careers with their current organization. This notion was addressed empirically. Respondents were classified into one of two groups on the basis of the length of time they-had been employed full-time by organizations other than their current one since the age of eighteen. The relationship between level of supportive behaviours received and the three employee outcome composites was evaluated, separately, (a) for respondents with less than five years tenure with other organizations (i.e., respondents in early career) (n = 181), and (b) for respondents with five or more years tenure with other organizations (n = 183). The correlations between supportive behaviours (from the first source) and the employee outcome composites were (early career group first): .29 and .22 for the Job-Related Composite, .48 and .46 for the Skill Development Composite, and .28 and .27 for the Promotional Composite. The pattern of correlations is highly similar for the two subgroups of respondents. These results indicate that the supportive behaviours received over the course of organizational tenure are related to employee development within an organization regardless of career stage. Despite this finding, additional research is required before the results of this study can be generalized to suggest that supportive behaviours received over the course of a career are positively related to career development. While such a generalization might be valid for employees who have spent their entire careers with one organization, supportive behaviours and employee outcomes need to be assessed in relation to careers that span a variety of organizations. Implications and Directions for Future Research In the early stages of the research reported here it was felt that it would be important to document that mentors had beneficial effects on the personal and career development of employees. As the research progressed, it became apparent that the term 'mentor' meant different things to different people and that definitional consensus had not been achieved. It also became apparent, however, that there was considerable overlap in the literature concerning the functions, roles, and behaviours that have been attributed to mentors. For this reason, this study concentrated not on whether employees had mentors, but, rather, on whether employees had been provided with the behaviours that 174 have been associated with mentor roles and functions. These behaviours were referred to as supportive behaviours in order to dissociate them from a particular source (i.e., mentors). The results of this study suggest that employees who receive supportive behaviours, regardless of whether they are received from mentors, can be expected to have positive attitudes toward their present organizational situations, high levels of skill development, and high levels of organizational advancement. Such outcomes clearly are desirable from both an individual and an organizational point of view. The practical implications of these findings may well be (a) that employees should be encouraged to seek out supportive behaviours, and (b) that organizations should encourage employees to provide supportive behaviours to others. There are, however, a number of theoretical isssues concerning supportive behaviours that are in need of further investigation. Theoretical issues. The results of this study suggest that the more supportive behaviours employees receive, the better, as far as employee outcomes are concerned. However, this study did not take into account either the desired level of supportive behaviours, or the discrepancy between received and desired levels of supportive behaviours. If all employees desire as many supportive behaviours as possible, then desired levels and discrepancies between received and desired levels may not be important. If, on the other hand, there are individual differences in desired levels of supportive behaviours, then additional research needs to address (a) what these individual differences are, and (b) how discrepancies between received and desired levels are related to employee outcomes. It is possible, for instance, that such discrepancies will explain more of the variance in employee outcomes than that explained by the level of supportive behaviours alone. In such an event, it may be important to tailor the provision of supportive behaviours to the needs and desires of individual employees. The hazards associated with mentor-protege relationships also require further investigation. In this study, the hazards were negatively related to employee outcomes (the job-related ones, in particular) for given levels of supportive behaviours. This finding suggests that employee outcomes can be maximized by reducing the level of hazards. There a number of ways in which this might be accomplished. In general, the hazards associated with being the recipient of supportive behaviours are due to the perceptions held by organizational members external to the focal relationship; these appear to be mostly perceptions of favouritism. Clawson and Kram 175 (1984) suggested that attributions of favouritism might be reduced by making other organizational members aware of the abilities and accomplishments of the focal employee. The possibility remains, however, that employees external to a supportive relationship may still be envious of the special attention afforded other employees through the provision of supportive behaviours (i.e., why them and not me?). This suggests that other methods should be used to reduce perceptions of favouritism. The most straightforward method may be to ensure that all employees are afforded relatively equal treatment in terms of the supportive behaviours with which they are provided. Because employee outcomes were positively related to level of support received, such a strategy should attempt to ensure that all employees are provided with high (or desired) levels of support. The highest aggregate level of employee outcomes may be achieved when the variance of support across employees in a particular work group is at a minimum. Research on this notion is clearly required. It may also be possible to reduce perceptions of favouritism by providing supportive behaviours in a more discrete fashion. If employees external to a supportive relationship are unaware of the support that is being provided, then these employees may be more likely to attribute the successes and accomplishments of a person to his or her personal qualities (e.g., intelligence, hard work) rather than to the actions of a particular provider of supportive behaviours. If discretely provided supportive behaviours lead to higher levels of employee outcomes than do openly provided supportive behaviours, then one must question the wisdom of establishing formal mentoring programs that are directed at only a small subgroup of employees. In such programs, favouritism would be obvious to all employees and the hazards might be expected to be even higher than in informal relationships in which favouritism is only suspected. Additional research concerning the relative benefits and drawbacks of formal mentoring programs is needed. Another avenue of future research concerns the differences and similarities between the factors that have been associated with effective leadership and the types of supportive behaviours that have been discussed here. Of particular interest is that the styles of leadership (e.g., consideration, employee-centred, relationship-oriented) that include a variety of the supportive behaviours discussed in this paper have not been related consistently to high levels of employee satisfaction and, in particular, performance (see Bass, 1981 for a review). Although the overlap between supportive behaviours and the 176 behaviours associated with any particular leadership style is far from complete, the results of this study may suggest why the findings in the leadership area are plagued by inconsistency. Typically, leadership studies focus on the level of behaviours provided by one supervisor (leader) at one point in time in order to understand employee outcomes at that point in time. The results of this study suggest that supervisors may not be the only sources of the behaviours that are assumed to be related to employee outcomes. For instance, although supervisors were identified by 69% of respondents as providing the highest number of supportive functions, 31% reported receiving the highest number of supportive functions from organizational members who were not supervisors. Had this study focused only on supervisors, it is likely that the relationships between supportive behaviours and employee outcomes would have been lower than those obtained; because of the positive relationships between employee outcomes and supportive behaviours received from whatever source, the level of employee outcomes would have been underestimated for respondents who received supportive behaviours from sources other than their supervisors. In terms of leadership behaviours, it is important to recognize that leadership can be shared (Bass, 1981) and that the sum total of leadership behaviours received by employees might be more highly related to employee outcomes than might the level of leadership behaviours received from any particular individual (i.e., a supervisor). This notion clearly deserves further study. The foregoing theoretical issues aside, the assessment of supportive behaviours in work settings should be of some practical benefit. This point will be discussed next. Practical applications. As part of this study, a scale was developed that can be used to assess the level of eight types of supportive behaviours received by an employee from each of several different people. Although this scale needs to be evaluated in a variety of settings, initial indications are that this scale is highly reliable (i.e., internally consistent) and both content and construct valid. Thus, this scale seems well-suited for use as a diagnostic tool. Knowledge of the extent to which any particular employee is being provided with each of the eight types of supportive behaviours (e.g., Teaching the Job, Encouragement) either (a) by anyone, or (b) by his or her supervisor is useful. As far as an employee is concerned, the source of supportive behaviours may be less important than knowing the level of behaviours being provided relative to work group, organizational, or industry 177 norms (additional studies are required to establish these norms). By comparing the received level of supportive behaviours to the appropriate norm, an employee can isolate areas of deficiency and develop strategies to overcome these deficiencies. For instance, an employee who receives fewer behaviours associated with Teaching the Job than do most of his or her co-workers may want to discuss this matter with his or her supervisor. Knowledge of the level of supportive behaviours provided to employees by their supervisors also has potential uses. First, supervisors can be compared on the average level of supportive behaviours provided to subordinates. Such information can be used to indicate areas of deficiency. This might be followed by training programs in which supervisors are taught how to deliver the supportive behaviours they are not providing. Second, the extent to which supervisors provide supportive behaviours can be incorporated into their performance appraisals. If the provision of supportive behaviours results in positive employee outcomes, then supervisors should be encouraged to provide as many supportive behaviours as possible. However, because the provision of behaviours may require considerable time investments, supervisors may be hesitant to provide the behaviours unless they are rewarded for doing so. By incorporating the level of supportive behaviours provided into performance appraisals, supervisors can be motivated to provide the behaviours. Because such a procedure represents a step toward the formalization of the provision of supportive behaviours, care must be taken to ensure that the behaviours are provided in an equitable fashion. Favouritism and perceptions of favouritism should be minimized. Conclusion Much literature suggests that support is related to well-being (e.g., Shumaker & Brownell, 1984) and that support promotes change and development in novel situations (e.g., Walter & Marks, 1981). This study represents a first step in measuring the quantity of support received in work settings and assessing the value of such support. The findings indicate quite clearly that quantity of supportive behaviours is positively related to a variety of desirable employee outcomes. These supportive behaviours need not, necessarily, be received from someone who is viewed as a mentor. 178 References Abdel-Halim, A.A. (1982). Social support and managerial affective responses to job stress. Journal of Occupational Behaviour, 3, 281-295. Adams, j.S. (1963). Toward an understanding of inequity. Journal of Abnormal and Social Psychology, 67, 422-436. Adler, S., & Aranya, N. (1984). A comparison of the work needs, attitudes, and preferences of professional accountants at different career stages. Journal of Vocational Behavior, 25, 45-57. Alleman, E. (1985). Manual for the Alleman Leadership Development Questionnaire. Mentor, OH: Leadership Development Consultants, Inc. Alleman, E., Cochran, J., Doverspike, j., & Newman, I. (1984). Enriching mentoring relationships. Personnel and Guidance Journal, 62, 329-332. Bales, R.F. (1958). Task roles and social roles in problem-solving groups. In E.E. Maccoby, T.H. Newcomb, & E.L. Hartley (Eds.), Readings in social psychology. New York: Holt, Rinehart, & Winston. Bass, B.M. (1981). Stogdill's handbook of leadership. New York: Free Press. Berry, P. (1983). Mentors for women managers: Fast-track to corporate success. Supervisory Management, 28(8), 36-40. Bolton, E.B. (1980). A conceptual analysis of the mentor relationship in the career development of women. Adult Education, 30, 195-207. Bova, B.M., & Phillips, R.R. (1981). The mentor relationship: A study of mentors and proteges in business and academia. University of New Mexico. (ERIC Document Reproduction Service No. ED 208 233). Bowen, D.D. (1985). Were men meant to mentor women? Training and Development Journal, 39(2), 31-36. Brayfield, A.H., & Rothe, H.F. (1951). An index of job satisfaction. Journal of Applied Psychology, 35, 307-311. Burke, R.J. (1984). Mentors in organizations. Group and Organizational Studies, 9, 353-372. Carroll, S.J., Olian, J.D., & Giannantonio, CM. (1988). Mentor reactions to proteges: An experiment with managers. Proceedings of the 48th Annual Meeting of the Academy of Management, 273-276. Clawson, J.G. (1980). Mentoring in managerial careers. In C.B. Derr (Ed.), Work, family and the career. New York: Praeger. 179 Clawson, J.C. (1985). Is mentoring necessary? Training and Development Journal, 39(4), 36-39. Clawson, J.G., & Kram, K.E. (1984). Managing cross-gender mentoring. Business Horizons, 27(3), 22-32. Cohen, P.A. (1984). College grades and adult achievement: A research synthesis. Research in Higher Education, 20, 281-293. Cook, J.D., Hepworth, S.J., Wall, T.D., & Warr, P.B. (1981). The experience of work. London: Academic Press. Cosier, R., & Dalton, D. (1983). Equity theory and time: A reformulation. Academy of Management Review, 8, 311-319. Corulla, W.J. (1987). A psychometric investigation of the Eysenck Personality Questionnaire (Revised) and its relationship to the 1.7 Impulsiveness Questionnaire. Journal of Personality and Individual Differences, 8, 651-658. Darling, L.W. (1985). The mentoring dimension. The Journal of Nursing Administration, 15, 41-42. Demo, D.H. (1985). The measurement of self-esteem: Refining our method. Journal of Personality and Social Psychology, 98, 1490-1502. Douglas, C.A., & Schoorman, F.D. (1988, August). The impact of career and psychosocial mentoring by supervisors and peers. Paper presented at the Academy of Management meetings, Anaheim, CA. Edwards, K.L. (1980). The influence of management function and perceived environmental support on perceived stress and job satisfaction of black females in managerial and professional positions in industry. Unpublished doctoral dissertation, University of Cincinnati. Eysenck, H.J., & Eysenck, S.B.G. (1975). Manual of the Eysenck Personality Inventory. London: University of London Press. Eysenck, S.B.G., Eysenck, H.J., & Barrett, P. (1985). A revised version of the psychotocism scale. Journal of Personality and Individual Differences, 6, 21-29. Farren, C, Gray, J.D., & Kaye, B. (1984). Mentoring: A boon to career development. Personnel, 61(6), 20-24. Feldman, D.C. (1976). A contingency theory of socialization. Administrative Science Quarterly, 21, 433-452. Fitt, L.W., & Newton, D. A. (1981). When the mentor is a man and the protege is a woman. Harvard Business Review, 59(2), 56, 58, 60. 180 Ford, L.D. (1985). Facets of work support and employee work outcomes: An exploratory analysis. Journal of Management, 11(3), 5-20. Fury, K. (1980). Mentor mania. Savvy, 1(1), 42-47. Cray, W.A. (1986). Components for developing a successful formalized mentoring program in business, the professions, education, and other settings. Proceedings of the First International Conference on Mentoring, 1, 15-22. Gould, S., & Hawkins, B.L. (1978). Organizational career stage as a moderator of the satisfaction-performance relationship. Academy of Management Journal, 3, 434-450. Hackman, J.R., & Oldham, G.R. (1975). Development of the Job Diagnostic Survey. Journal of Applied Psychology, 60, 135-139. Halatin, T.J., & Knotts, R.E. (1982). Becoming a mentor: Are the risks worth the rewards? Supervisory Management, 27(2), 27-29. Halcomb, R. (1980). Mentors and the successful woman. Across the Board, 17(2), 13-18. Hall, D.T. (1976). Careers in Organizations. Santa Monica, CA: Goodyear Publishing Company. Harman, H.H. (1976). Modern factor analysis. (3rd ed.). Chicago: University of Chicago Press. Harpaz, l. (1985). Meaning of working profiles of various occupational groups, journal of Vocational Behavior, 26, 25-40. Henderson, D.W. (1985). Enlightened mentoring: A characteristic of public management professionalism. Public Administration Review, 45, 857-863. House, J.S. (1981). Work stress and social support. Reading, MA: Addison-Wesley. House, J.S., & Kahn, R.L. (1985). Measures and concepts of social support. In S. Cohen & S.L. Syme (Eds.), Social support and health. New York: Academic Press. Hunt, D.M., & Michael, C. (1983). Mentorship: A career training and development tool. Academy of Management Review, 8, 475-485. Kanter, R.M. (1977). Men and women of the corporation. New York: Basic Books. Katz, R. (1980). Time and work: Toward an integrative perspective. In L.L. Cummings & B. Staw (Eds.), Research in organizational behaviour, Vol. 2, (pp. 81-127). Katz, R.L. (1955). Skills of an effective administrator. Harvard Business Review, 33(1), 33-42. Klopf, G.J., & Harrison, J. (1981). Moving up the career ladder: The case for mentors. Principal, 61(1), 41-43. 181 Kram, K.E. (1983). Phases of the mentor relationship. Academy of Management Journal, 26, 608-625. Kram, K.E. (1985). Mentoring at work: Developmental relationships in organization life. Clenview, IL: Scott, Foresman. La Caipa, ).). (1977). Testing a multidimensional approach to friendship. In S. Duck (Ed.), Theory and practice in interpersonal attraction. New York: Academic Press, (pp. 249-270). Levinson, D.J., Darrow, C.N., Klein, E.B., Levinson, M.H., & McKee, B. (1978). The seasons of a man's life. New York: Knopf. Louis, M.R., Posner, B.Z., & Powell, G.N. (1983). The availability and helpfulness of socialization practices. Personnel Psychology, 36, 857-866. Luthans, F. (1988). Successful vs. effective real managers. The Academy of Management Executive, 2, 127-132. Merriam, S. (1983). Mentors and proteges: A critical review of the literature. Adult Education Quarterly, 33, 161-173. Missirian, A.K. (1980). The process of mentoring in the career development of female managers. Unpublished doctoral dissertation, University of Massachusetts. Mossholder, K.W., Bedeian, A.G., Touliatos, )., & Barkman, A.l. (1985). An examination of intraoccupational differences: Personality, perceived work climate, and outcome preferences. Journal of Vocational Behavior, 26, 164-176. Mount, M.K. (1984). Managerial career stage and facets of job satisfaction. Journal of Vocational Behavior, 24, 340-354. Mowday, R.T., Steers, R.M., & Porter, L.W. (1979). The measurement of organizational commitment. Journal of Vocational Behavior, 14, 224-247. Noe, R.A. (1988). Women and mentoring: A review and research agenda. Academy of Management Review, 13, 65-78. Norbeck, J.S., Lindsey, A.M., & Carried, V.L. (1981). The development of an instrument to measure social support. Nursing Research, 30, 264-269. Nunnally, ).C. (1967). Psychometric theory. New York: McGraw Hill. Orth, CD., & Jacobs, F. (1971). Women in management: Patterns for change. Harvard Business Review, 49(4), 139-147. Phillips-Jones, L. (1982). Mentors and proteges. New York: Arbour House. Pinder, CC. (1982). Survey of transfer-related attitudes. Vancouver, BC: University of British Columbia. 182 Price, J.L., & Mueller, CW. (1986). Handbook of organizational measurement. Marshfield, MA: Pitman Publishing, Inc. Reber, R.W., & Van Gilder, G. (1982). Behavioral insights for supervision. Englewood Cliffs, N): Prentice-Hall. Reich, M.H. (1985). Executive views from both sides of mentoring. Personnel, 62(3), 42-46. Reich, M.H. (1986). The mentor connection. Personnel, 63(2), 50-56. Reis, H.T. (1984). Social interactions and well-being. In S.W. Duck (Ed.), Personal relationships 5: Repairing personal relationships. New York: Academic Press. Rhodes, S.R. (1983). Age-related differences in work attitudes and behavior: A review and conceptual analysis. Psychological Bulletin, 93, 328-367. Rizzo, J.R., House, R.J., & Lirtzman, S.l. (1970). Role conflict and ambiguity in complex organizations. Administrative Science Quarterly, 15, 150-163. Robinson, J., & Shaver, P. (1973). Measures of social psychological attitudes. Ann Arbor, Ml: Institute for Social Research. Roche, G.R. (1979). Much ado about mentors. Harvard Business Review, 57(1), 14-28. Rosenberg, M. (1965). Society and the adolescent self-image. Princeton, NJ: Princeton University Press. Rubin, Z. (1970). Measurement of romantic love. Journal of Personality and Social Psychology, 16, 265-273. Samson, G.E., Graue, M.E., Weinstein, T., & Walberg, H.J. (1984). Academic and occupational performance: A quantitative synthesis. American Educational Research Journal, 21, 311-321. Scandura, T.A., & Katerberg, R.J. (1988, August). Much ado about mentors and little about measurement: Development of an instrument. Paper presented at the Academy of Management meetings, Anaheim, CA. Schein, E.H. (1978). Career dynamics: Matching individual and organizational needs. Reading, MA: Addison Wesley. Schockett, M.R., & Haring-Hidore, M. (1985). Factor analytic support for psychosocial and vocational mentoring functions. Psychological Reports, 57, 627-630. Shapiro, E.C, Haseltine, F.P., & Rowe, M.P. (1978). Moving up: Role models, mentors, and the "patron system." Sloan Management Review, 19(3), 51-58. Shumaker, S.A., & Brownell, A. (1984). Toward a theory of social support: Closing the conceptual gaps. Journal of Social Issues, 40, 11-36. 183 Slocum, J.W., & Cron, W.L. (1985). Job attitudes and performance during three career stages, journal of Vocational Behavior, 26, 126-145. Smith, P.C, Kendall, L.M., & Hulin, C.L.. (1969). The measurement of satisfaction in work and retirement. Chicago: Rand McNally. Speizer, J.J. (1981). Review essay-role models, mentors, and sponsors: The elusive concepts. Signs, 6, 692-712. Statistics Canada. (1988). The consumer price index. Ottawa: Ministry of Supply and Services Canada. Stodgill, R.M. (1963). Manual for the Leader Behavior Description Questionnaire-Form XII. Bureau of Business Research, Ohio State University, Columbus, OH. Stumpf, S.A., & Rabinowitz, S. (1981). Career stage as a moderator of performance relationships with facets of job satisfaction and role perceptions. Journal of Vocational Behavior, 18, 202-218. Thoits, P.A. (1982). Conceptual, methodological, and theoretical problems in studying social support as a buffer against life stress, journal of Health and Social Behavior, 23, 145-159. Tsui, A.S., & Gutek, B.A. (1984). A role set analysis of gender differences in performance, affective relationships, and career success of industrial middle managers. Academy of Management Journal, 27, 619-635. Vance, CN. (1982). The mentor connection. The Journal of Nursing Administration, 12, 7A3. Vroom, V.H. (1964). Work and motivation. New York: Wiley. Walter, G.A., & Marks, S.E. (1981). Experiential learning and change. New York: Wiley. Whitely, W., Dougherty, T.W., & Dreher, G.F. (1988). The relationship of mentoring and socioeconomic origin to managers' and professionals' early career progress. Proceedings of the 48th Annual Meeting of the Academy of Management, 58-62. The Woodlands Group. (1980). Management development roles: Coach, sponsor, and mentor. Personnel Journal, 59, 918-921. Zey, M.C. (1984). The mentor connection. Homewood, IL: Dow Jones-Irwin. 184 Appendix A  Items for Measures Developed for this Study 185 Satisfaction with Progression How satisfied are you with your present salary? How satisfied are you with the rate at which your salary has increased since you joined your present organization? How satisfied are you with the rate at which you have been promoted since you joined your present organization? How satisfied are you with the kinds of assignments you have received since you joined your present organization? How satisfied are you with the rate at which you have developed professionally since you joined your present organization? Job Skill Development Respondents were asked to indicate how much they had developed each of the following skills: (a) verbal skill, (b) numerical skill, (c) technical skill, (d) writing skill, (e) diplomacy and tact, (f) long range planning skill, (g) short range planning skill, (h) group decision making skill, (i) independent decision making skill, (j) information processing skill, (k) public speaking skill, (I) negotiating skill, (m) mediation/peacemaking skill (n) abstract reasoning skill, (o) legal skill, (p) resource acquisition skill, (q) resource allocation skill, (r) budgetary skill, (s) coordinating skill, (t) political skill, (u) creative skill, (v) interpersonal skill, (w) counselling skill, (x) liaison skill, and (y) work assignment skill. Interpersonal Skill Development I believe that it is important to look at situations from the point of view of other employees when making decisions. I believe that it is important to consider the impact my behaviour has on other employees. I believe that it is important to understand the specific concerns of other employees. I believe that it is important to understand the particular attitudes and values of other employees. I believe that it is important to consider the individual needs of other employees when making decisions. Conceptual Skill Development I know how all the parts of the firm fit together. I understand the reasons behind company policies. I am aware of the problems faced by other members of the organization. I understand the 'bigger picture' facing the firm. I understand the organization's overall goals. Acceptance by Co-workers People at work value my opinions. I lack credibility among my co-workers, (reverse scored) People at work listen to my ideas. I feel that I am accepted by my co-workers. People at work do not respect my judgment, (reverse scored) Supportive Behaviours The items listed below are grouped according to the supportive function with which they are associated. The numbers in parentheses indicate the order of the items the pilot study and the final questionnaire, respectively. Sponsoring Recommended you to other members of the organization for a particular position, assignment, or opportunity. (1, 1) Went out of his or her way to tell others about how you were performing. (10, 9) Told others about your potential or competence. (19, 17) Attempted to ensure that you were not passed over for desirable tasks or positions by telling others about you. (28, 25) Helped you get assigned to special project teams, task forces, committees, etc. by making sure that people knew about you. (37, NA) Made sure that other members of the organization became aware of your accomplishments. (46, 33) Exposure and Visibility Helped you meet important people outside of your organization. (5, 5) Included you in meetings with other people whenever possible. (14, NA) Saw to it that you got to know as many people in the organization as possible. (23, 21) Made opportunities available that provided high visibility for you. (32, 29) Arranged for you to attend functions where you could meet influential members of you organization. (41, 37) Introduced you to more senior members of your organization. (50, 13) 187 Teaching the Job Cave you assignments that developed your skills. (3, 3) Taught you about the technical aspects of your job. (12, NA) Provided opportunities for you to learn different aspects of your job. (21, 19) Ensured that you learned from your assignments. (30, 27) Helped you learn your job by discussing hypothetical problems. (39, 35) Provided the guidance necessary to help you learn your job. (48, 11) Teaching the Informal System Taught you how to personally treat different people in the organization. (8, 8) Told you who the powerful members of the organization were. (17, 16) Made sure you had an understanding of the informal aspects of the organization. (26, 24) Told you about the political aspects of the organization. (35, 32) Told you about who could be trusted. (44, 40) Told you about unwritten rules and procedures. (53, NA) Protection by Prevention Made sure that you were not assigned tasks that you might not have been able to accomplish. (7, NA) Made sure that you did not work too closely with others until you could perform well in their presence. (16, 15) Shielded you from situations where you would not look your best. (25, 23) Stopped you from taking on tasks that might have led to criticism if you did not do them well. (34, NA) Stopped you from doing things that might have led to criticism from your superiors. (43, NA) Attempted to stop you from taking on work that you might not have been able to do. (52, NA) Protection by Absorption Took the blame for some of your actions. (9, 7) Prevented your mistakes from becoming public knowledge. (18, 31) Absorbed criticism that was directed at you. (27, NA) Took responsibility for your mistakes. (36, NA) Made sure that other organizational members did not find out about your mistakes. (45, 188 39) Made sure that you did not suffer negative consequences for your actions. (54, NA) Role Modeling Observing him or her in meetings helped you learn how to manage people better. (2, 2) Watching him or her helped you deal more effectively with people. (11, 10) Watching him or her helped you improve your management skills. (20, 18) His or her behaviour provided you with an example of how to do things better. (29, 26) By watching how he or she did things, you learned how to perform your job more easily. (38, 34) You learned aspects of your work by observing him or her. (47, NA) Encouragement Encouraged you to make the most of opportunities that came your way. (6, 6) Expressed confidence in you when you faced difficult decisions. (15, 14) Encouraged you to carry on even when you had your doubts. (24, 22) Made you feel that you were a worthwhile member of the organization. (33, NA) Made you feel .confident that you would succeed at difficult tasks. (42, 38) Expressed confidence in your abilities. (51, 30) Personal Counselling Could be counted on to discuss your personal concerns. (4, 4) You were able to discuss personal matters with him or her. (13, 12) Served as a confidant when you needed one. (22, 20) Took an interest in your personal concerns. (31, 28) Could be counted on to give you personal guidance. (40, NA) Acted as a sounding board for your personal concerns. (49, 36) Hazards of Mentor-Protege Relationships Do you think that your relationship with this person is (was) of such a nature that some members of the organization are (were) envious of your relationship? Do you think that some members of the organization attribute(d) your successes and failures too much to this person's actions? Do you think that your relationship with this person has kept you from attaining any positions, assignments, or opportunities that you may have wanted? 189 Do you think that your relationship with this person has kept you from assuming responsibility for matters that you should have been responsible for? Friendship Your work relationship aside, how close, personally, do (did) you feel to this person as a friend? How much do (did) you support and encourage each other when one of you feels (felt) unhappy? How personally close do (did) you feel to this person most of the time? How often do (did) you show that you like (liked) each other as individuals? How personally satisfying is (was) your relationship with this person? Characteristics of Relationships What is (was) your relationship with this person? - i.e., is (was) this person your first supervisor, a subordinate, a co-worker, your supervisor's boss, etc. How old (in years) is this person now? Estimate if not sure. Is this person male (M) or female (F)? How old were you when you first met this person? 190 Appendix B Instructions to judges for Classifying Supportive Behaviours 191 Mentoring has been defined as a relationship between two individuals in which one individual (usually called the mentor) contributes to the career development of the other by acting as a teacher, guide, or coach. This definition of mentoring is viewed as a very broad or general one. In order to gain more insight into the phenomenon of mentoring, a literature review was conducted in order to identify functions that mentors perform for other employees. The functions identified by this review are listed and defined on the following pages. Following identification of the functions, a number of statements were written that were intended to be indicative of the behaviours that mentors might engage in when providing each of the functions. Your task is to indicate which of the statements are indicative of which functions. The purpose of this task is to provide me with an indication of the degree to which different people agree that the statements are indicative of particular functions. The functions are divided into two groups-career functions and psychosocial functions. The definitions for the career functions are provided on the following page. Please read these definitions CAREFULLY and feel free to ask questions about them. After you have familiarized yourself with these definitions, I will provide you with a list of statements. For each statement, please indicate with which function you believe it to be associated. FEEL FREE TO REFER BACK TO THE DEFINITIONS WHEN SORTING THE STATEMENTS This procedure will then be repeated for the psychosocial functions. Career Functions Career functions "...enhance learning the ropes and preparing for advancement in an organization" (Kram, 1985, p. 22). Sponsoring Communicating information to others, either verbally or nonverbally, concerning the potential or competence of an employee. Exposure and Visibility Providing an employee with tasks or opportunities that ensure that others will become personally acquainted with that employee. Teaching the Job Providing knowledge or experiences that help an employee learn technical and managerial skills. Teaching the Informal System Imparting of knowledge to an employee concerning informal aspects of the organization such as norms, mores, and politics. Protection by Prevention Preventing an employee from being exposed to tasks, positions, or situations in which the employee may be at risk or unlikely to succeed. Protection by Absorption Taking the blame for, or not telling others about an employee's actions or behaviours that may be detrimental to the employee. 193 Career Functions Please indicate your response by placing the appropriate letter to the right of the statement number: A = Sponsoring B = Exposure and Visibility C = Teaching the Job D = Teaching the Informal System E = Protection by Prevention F = Protection by Absorption 01... 19.. 02... 20.. 03... 21.. 04... 22.. 05... 23.. 06... 24.. 07... 25.. 08... 26.. 09... 27.. 10... 28.. 11... 29.. 12... 30., 13... 31., 14... 32. 15... 33. 16... 34. 17... 35. 18... 36. 194 Psychosocial Functions Psychosocial functions "...enhance a sense of competence, clarity of identity, and effectiveness in a professional role" (Kram, 1985, p. 22). Role Modeling Being a model or example for an employee thereby allowing the employee to learn by observation. Encouragement Providing encouragement or support that helps an employee feel competent, confident, or worthwhile. Personal Counselling Discussing an employee's concerns involving self, career, or family. 195 Psychosocial Functions Please indicate your response by placing the appropriate letter to the right of the statement number: A = Role Modeling B = Encouragement C = Personal Counselling 01 10... 02 11... 03 12... 04 13... 05 14... 06 15... 07. 16... 08 17... 09 18... Appendix C  Questionnaire Instructions and Item Order 197 Measures were included in the order listed in this Appendix. The scales used (i.e., 5-point Likert) are described in the Method section. The items for measures developed for use in this study are given in Appendix A. Instructions are provided verbatim. Job Skill Development Instructions: The particular skills that people develop depend, to a large extent, on assignments received and positions held. Please indicate how much you have been able to develop each of the skills listed below since joining your present organization. Role Ambiguity, Role Conflict, Acceptance by Co-workers, Interpersonal Skill Development,  and Conceptual Skill Development The items for all of these measures were randomly arranged in the second section of the first part of the questionnaire. Instructions: The following statements concern conditions that may apply to you or your job. Please indicate the degree to which the statements are true of your particular situation by circling the appropriate number between 1 and 7 below each statement. Satisfaction with Progression Instructions: The following statements concern your level of satisfaction with such things as promotion and salary. Please indicate your degree of satisfaction by circling the appropriate number between 1 and 5 below each statement. Organizational Commitment Instructions: Listed below are a series of statements that concern the feelings you may have about the organization for which you work. Please indicate the degree of your agreement or disagreement with each statement by circling the appropriate number between 1 and 7 below each statement. ]ob Satisfaction Instructions: Some jobs are more interesting and satisfying than others. Below each statement circle the appropriate number between 1 and 5 which best describes how you feel about your present job. 198 Supportive Functions The statements used to elicit the initials of providers of supportive functions were presented in Table 5. The following instructions preceded the statements in Table 5. Instructions: In this part of the questionnaire we would like you to: (a) identify (using initials only) people in your organization who have provided you with a variety of organizational experiences, and (b) answer a number of questions concerning your relationship with some of these people. Since joining your present organization, have any people you have worked with (either currently or in the past), provided you with any of the following experiences? If yes, please place the INITIALS of up to 3 of these people in the spaces to the right of each statement. If no, go on to the next statement. We should note that the number of people you list will depend largely on the positions you have held and the length of time you have worked for your present organization. For this reason, you may or may not list very many people. Determination of which People to Rate on Supportive Behaviours Instructions: Please consider all of the people you listed on the previous page. In the yellow space under Person 1 at the top right of this page, place the INITIALS of the person you listed mosr often in the eight statements on the previous page. List this person's initials regardless of the number of times you listed this person on the previous page. In the yellow space under Person 2 at the top right of this page, place the INITIALS of the person you listed next most often, but only if you listed that person 3  or more times. Otherwise, leave the space under 'Person 2' blank. Finally, in the yellow space under Person 3, place the INITIALS of the person you listed next most often, but, again, only if you listed that person 3 or more times. Characteristics of Relationships, Friendship, Hazards, Mentor Definitions, and Effect on  Career Development Items concerning these factors were included in one section and were presented in the following order: (a) the four questions concerning relationships, (b) the five 199 friendship items, (c) the four items for hazards, (d) the two mentor definitions (broad definition first), and (e) the item concerning overall effect on career development. Instructions: We would like you to briefly describe the people you have just identified by answering a number of questions concerning your relationship with them. For each of the people you have listed please give your answer in the column under that person's initials. Supportive Behaviours Instructions: The following statements refer to activities that the people you have listed may have engaged in on your behalf. The activities are similar to the experiences we listed earlier in the questionnaire, but are somewhat more specific. For each activity, please indicate the extent to which each person you listed engaged in that activity on your behalf by entering the appropriate number between 0 and 4 in the column under each person's initials. Extraversion Instructions: Listed below are a number of questions related to how you might describe yourself. Please answer each question by putting a circle around the 'YES' or the 'NO' following each question. There are no right or wrong answers. Self-Esteem Instructions: The following statements may or may not be descriptive of you. Please indicate the degree to which you agree or disagree with each statement by circling the appropriate number between 1 and 4 below each statement. Demographic Variables There were no specific instructions used. In order, respondents were asked to provide information concerning (a) age, (b) sex, (c) extent of formal education, (d) name of present employer, (e) years of organizational tenure, (f) years of full-time work since the age of 18, (g) number of different positions held with present employer, (h) number of promotions with present employer, (i) current salary, (j) initial salary with present employer, and (k) type of job category. 200 Appendix D  Description of Survey Respondents Age, Gender, and Tenure of Respondents and P o t e n t i a l Respondents Age: Mean s.d. Respondents 37.99 7.06 P o t e n t i a l Respondents 38.47 7.23 Gender: Male Female 286 (72.0%) 111 (28.0%) 455 (72.9%) 169 (27.1%) Years of Tenure: Mean s .d. 10.31 4.13 9.23 4.34 202 Level of Education and Job Category of Survey Respondents Number of Highest l e v e l of education Respondents Up to high school graduation 33 Some college or un i v e r s i t y 59 College diploma 99 Uni v e r s i t y degree/some graduate work 120 Graduate degree 86 Totals 397 Percent of Respondents 8.3 14.9 24.9 30.2 21.7 100.0 Job category Number of Respondents Single categories Senior executive 3 Middle manager 19 F i r s t l i n e supervisor 68 Profe s s i o n a l 124 Technical 92 Other or not s p e c i f i e d 21 Multi p l e categories P r o f e s s i o n a l / t e c h n i c a l 24 Pro f e s s i o n a l middle manager 6 Pro f e s s i o n a l f i r s t l i n e supervisor 13 Technical middle manager 1 Technical f i r s t l i n e supervisor 7 Some other combination of job categories 19 Totals 397 Percent of Respondents 0.7 4.8 17.1 31.2 23.2 5.3 6.0 1.5 3.3 0.2 1.8 4.8 100.0 Appendix E Means and (Standard Deviations) of Supportive Behaviours according to Definitions of Mentors 204 Means and (Standard Deviations) of Supportive Behaviours according to D e f i n i t i o n s of Mentors D e f i n i t i o n 1 D e f i n i t i o n 2 Scales for F i r s t Source Yes No Yes No Sponsoring 13.49 8.51 13.73 9.41 ( 4.19) ( 4.67) ( 4.19) ( 4.65) Exposure and V i s i b i l i t y 10.23 5.24 „ 10.56 5.93 ( 5.03) ( 4.32) ( 5.09) ( 4.28) Teaching the Job 13.58 9.91 14.38 9.67 ( 4.33) ( 4.91) ( 4.05) ( 4.41) Teaching the Informal System 10.39 6.81 10.57 7.30 ( 5.08) ( 4.81) ( 4.92) ( 5.00) Protection 6.97 4.25 7.25 4.45 ( 4.69) ( 3.96) ( 4.83) ( 3.76) Role Modeling 13.08 9.76 13.96 9.21 ( 4.82) ( 5.16) ( 4.42) ( 4.85) Encouragement 14.99 10.98 15.43 11.36 ( 3.59) ( 4.75) ( 3.45) ( 4.40) Personal Counselling 13.36 9.11 13.59 9.72 ( 4.78) ( 5.13) ( 4.70) ( 5.16) TSBS-1 96.42 64.71 99.97 67.01 (27.51) (28.92) (26.87) (26.45) Minimum n 256 109 223 149 Note. D e f i n i t i o n 1 = Does (did) t h i s person take a personal i n t e r e s t i n your career and guide or sponsor you? D e f i n i t i o n 2 = Does (did) t h i s person serve as a career role model and a c t i v e l y advise, guide, and promote your career and training? TSBS-1 = Total Supportive Behaviour Score for the f i r s t source. A l l differences between the Yes and No groups are s i g n i f i c a n t at the .001 l e v e l (one-tailed). 205 Means and (Standard Deviations) of Supportive Behaviours according to D e f i n i t i o n s of Mentors D e f i n i t i o n 1 D e f i n i t i o n 2 Scales f o r Second Source Yes No Yes No Sponsoring 12.71 7.72 13.11 8.33 (4.52) (4.27) (4.58) ( 4.29) Exposure and V i s i b i l i t y 9.77 5.21 10.21 5.58 ( 4.56) ( 4.30) ( 4.60) ( 4.17) Teaching the Job 12.81 8.67 13.56 8.86 ( 4.14) ( 4.52) ( 4.23) ( 3.98) Teaching the Informal System 9.70 7.17 9.86 7.51 ( 4.80) ( 4.62) ( 4.66) ( 4.77) Protection 7.17 3.78 7.06 4.51 ( 4.62) ( 3.53) ( 4.68) ( 3.94) Role Modeling 12.93 9.78 13.70 9.76 ( 4.37) ( 5.01) ( 4.19) ( 4.69) Encouragement 14.41 10.48 14.78 10.99 ( 3.80) ( 4.76) ( 3.81) ( 4.51) Personal Counselling 12.37 8.50 12.50 9.21 ( 5.09) ( 5.01) ( 4.92) ( 5.27) TSBS-2 92.11 61.92 94.93 65.15 (26.77) (25.40) (27.01) (24.76) Minimum n 158 108 135 136 Note. D e f i n i t i o n 1 = Does (did) t h i s person take a personal i n t e r e s t i n your career and guide or sponsor you? D e f i n i t i o n 2 = Does (did) t h i s person serve as a career r o l e model and a c t i v e l y advise, guide, and promote your career and tra i n i n g ? TSBS-2 = Total Supportive Behaviour Score f o r the second source. A l l differences between the Yes and No groups are s i g n i f i c a n t at the .001 l e v e l (one-tailed). 206 Means and (Standard Deviations) of Supportive Behaviours according  to D e f i n i t i o n s of Mentors D e f i n i t i o n 1 D e f i n i t i o n 2 Scales f or Third Source Yes No Yes No Sponsoring 12.22 8.08 12.89 8.91 ( 4.83) ( 4.34) ( 4.92) ( 4.45) Exposure and V i s i b i l i t y 8.92 5.59 10.03 5.85 ( 5.06) ( 4.02) ( 5.24) ( 3.83) Teaching the Job 11.99 9.33 13.39 9.27 ( 4.31) ( 4.49) ( 3.99) ( 4.19) Teaching the Informal System 9.23 7.61* 9.48 7.98* (4.98) (4.52) (4.78) (4.78) Protection 5.90 4.25* 6.87 4.09 ( 4.75) ( 3.97) ( 4.77) ( 3.90) Role Modeling 11.71 9.68** 12.65 9.65 ( 4.11) ( 4.82) ( 4.26) ( 4.24) Encouragement 14.16 10.70 14.54 11.50 ( 3.58) ( 4.49) ( 3.57) ( 4.33) Personal Counselling 11.39 8.66 11.97 9.08 ( 4.96) ( 5.35) ( 4.64) ( 5.40) TSBS-3 86.05 64.43 91.89 67.15 (25.95) (24.44) (27.01) (22.56) Minimum n 88 60 62 88 Note. D e f i n i t i o n 1 = Does (did) t h i s person take a personal i n t e r e s t i n your career and guide or sponsor you? D e f i n i t i o n 2 = Does (did) t h i s person serve as a career r o l e model and a c t i v e l y advise, guide, and promote your career and training? TSBS-3 = Total Supportive Behaviour Score f o r the t h i r d source. A l l differences between the Yes and No groups are s i g n i f i c a n t at the .001 l e v e l (one-tailed) except *p < .05 and **p < .01 (one-tailed). 207 Appendix F Correlations between Employee Outcomes and Measures for the Three Sources of Support C o r r e l a t i o n s between Employee Outcomes and Measures f o r the F i r s t Source of Support Measures f o r the F i r s t Source of Support Employee Outcomes SP EX u PR RM EN PC TSBS FR HZ NF O r g a n i z a t i o n a l Commitment 35 38 35 20 15 39 35 25 39 20 -06 28 Role Ambiguity -16 -16 -22 05 03 -18 - 18 -08 - 16 13 15 -13 Role C o n f l i c t 03 04 -07 14 1 1 -09 -02 03 03 02 31 04 Job S a t i s f a c t i o n 18 21 18 08 -01 23 17 11 20 07 -12 14 S a t i s f a c t i o n w i th P r o g r e s s i o n 29 30 27 13 05 27 28 12 28 10 -06 23 Co-worker Acceptance 18 13 . 19 01 00 17 20 07 15 11 - 1 1 13 Job S k i l l Development 37 43 41 35 32 45 42 35 50 22 15 29 I n t e r p e r s o n a l S k i l l Development 13 10 18 06 06 17 16 13 16 15 08 14 Conceptual S k i l l Development 29 35 26 27 16 36 27 21 34 14 1 1 20 Average Annual S a l a r y Increase 10 16 06 10 15 15 12 01 14 07 01 08 Average Number of Promotions/Yr. 21 16 19 12 06 19 15 07 19 05 -02 19 J o b - R e l a t e d Composite 25 25 29 07 01 30 26 14 25 15 -20 18 S k i l l Development Composite 37 42 39 33 26 46 40 33 47 24 16 30 Pro m o t i o n a l Composite 28 28 24 15 13 28 26 08 28 04 -04 21 Note. SP = S p o n s o r i n g . EX = Expos ure and V i s ib i 1 i ty , TJ = Teaching the Job, TI = Teac h i n g the I n f o r m a l S PR = P r o t e c t i o n , RM = Role Modeling, EN = Encouragement, PC = P e r s o n a l C o u n s e l l i n g , TSBS = T o t a l S u p p o r t i v e Behaviour Score, FR = F r i e n d s h i p , HZ = Hazards, NF = Number of F u n c t i o n s P r o v i d e d . C o r r e l a t i o n s g r e a t e r than .16 in a b s o l u t e magnitude are s i g n i f i c a n t at the .001 l e v e l , o n e - t a i l e d . C o r r e l a t i o n s between .12 and .16 are s i g n i f i c a n t at the .01 l e v e l and c o r r e l a t i o n s between .09 and .11 are s i g n i f i c a n t at the .05 l e v e l . Minimum n f o r a c o r r e l a t i o n = 355, maximum = 395 (decimal p o i n t s o m i t t e d ) . o co C o r r e l a t i o n s between Employee Outcomes and Measures f o r the Second Source of Support Measures f o r the Second Source of Support Employee Outcomes SP EX I i u PR RM EN PC TSBS FR HZ NF O r g a n i z a t i o n a l Commitment 32 40 37 20 17 31 26 11 35 08 05 19 Role Ambiguity -14 -15 -22 -02 06 -15 -14 -08 - 14 13 09 -06 Role Conf 1 i c t -03 01 -15 09 06 -05 -07 -02 -04 00 29 07 Job S a t i s f a c t i o n 22 22 22 11 00 17 22 08 21 1 1 -07 12 S a t i s f a c t i o n w i th P r o g r e s s i o n 24 22 24 08 04 11 31 13 23 15 02 11 Co-worker Acceptance 17 13 13 -02 -02 06 20 06 12 13 - 10 04 Job S k i l l Development 37 38 40 30 33 39 32 19 43 15 21 18 I n t e r p e r s o n a l S k i l l Development 10 08 05 03 -03 07 14 10 08 15 02 12 Conceptual S k i l l Development 20 31 23 27 15 29 27 20 31 24 16 12 Average Annual S a l a r y I n c r e a s e 16 17 07 14 15 13 18 13 18 05 05 07 Average Number of Promotions/Yr. 11 05 06 -02 -02 08 11 02 06 00 01 01 J o b - R e l a t e d Composite 26 27 32 07 01 22 26 10 25 13 - 13 11 S k i l l Development Composite 32 37 33 30 23 36 35 24 40 26 19 20 Pr o m o t i o n a l Composite 23 18 15 09 08 15 29 15 22 1 1 02 06 Note. SP = S p o n s o r i n a . EX = Exposure and V i s i b i l i t y , TJ = Teaching the Job, TI - Teach i ng the I n f o r m a l ' PR = P r o t e c t i o n , RM = Role Modeling, EN = Encouragement, PC = P e r s o n a l C o u n s e l l i n g , TSBS = T o t a l S u p p o r t i v e Behaviour Score, FR = F r i e n d s h i p , HZ = Hazards, NF = Number of F u n c t i o n s P r o v i d e d . C o r r e l a t i o n s g r e a t e r than .18 in a b s o l u t e magnitude are s i g n i f i c a n t at the .001 l e v e l , o n e - t a i l e d . C o r r e l a t i o n s between .14 and .18 are s i g n i f i c a n t at the .01 l e v e l and c o r r e l a t i o n s between .10 and .13 are s i g n i f i c a n t at the .05 l e v e l . Minimum n f o r a c o r r e l a t i o n = 261, maximum = 291 ( d e c i m a l p o i n t s o m i t t e d ) . o v£> C o r r e l a t i o n s between Employee Outcomes and Measures f o r the T h i r d Source of Support Measures f o r the T h i r d Source of Support Emplpvee Outcomes SP EX I J I I PR RM EN PC TSBS FR HZ NF O r g a n i z a t i o n a l Commitment 27 25 08 08 15 14 28 04 22 19 02 00 Role Ambiguity -06 -05 -10 08 09 - 17 -06 -08 -06 -19 -03 -06 Role C o n f l i c t 05 08 -14 22 09 -01 -09 03 05 -01 20 04 Job S a t i s f a c t i o n 21 18 17 06 07 14 27 11 22 04 -06 13 S a t i s f a c t i o n w i th P r o g r e s s i o n 18 10 11 03 -02 01 23 04 13 08 -02 04 Co-worker Acceptance 15 07 07 00 -06 -03 10 05 07 04 00 02 Job S k i l l Development 30 24 16 16 24 21 24 04 28 05 19 04 I n t e r p e r s o n a l S k i l l Development 06 15 01 00 -03 -02 -04 -13 00 -02 03 -04 Conceptual S k i l l Development 18 16 15 16 08 17 15 00 19 17 12 08 Average Annual S a l a r y I n c r e a s e 13 27 11 12 09 07 20 -02 16 -02 16 13 Average Number of Promotions/Yr. 08 06 01 07 -11 -04 01 -14 -04 - 10 00 -02 J o b - R e l a t e d Composite 19 14 16 04 00 13 23 08 16 14 -05 06 S k i l l Development Composite 25 25 16 15 15 17 15 -06 23 06 16 05 Promotional Composite 17 19 10 02 -01 01 18 -07 12 -03 07 07 Note. SP = S p o n s o r i n g . EX = Exposure and V i s i b i 1 i t y , TJ = Teaching the Job, TI = T e a c h i n g the Informal I PR = P r o t e c t i o n , RM = Role Modeling, EN = Encouragement, PC = P e r s o n a l C o u n s e l l i n g , TSBS = T o t a l S u p p o r t i v e Behaviour Score, FR = F r i e n d s h i p , HZ = Hazards, NF = Number of F u n c t i o n s P r o v i d e d . C o r r e l a t i o n s g r e a t e r than .24 i n a b s o l u t e magnitude are s i g n i f i c a n t at the .001 l e v e l , o n e - t a i l e d . C o r r e l a t i o n s between .19 and .24 are s i g n i f i c a n t at the .01 l e v e l and c o r r e l a t i o n s between .14 and .18 are s i g n i f i c a n t at the .05 l e v e l . Minimum n f o r a c o r r e l a t i o n = 145, maximum = 162 ( d e c i m a l p o i n t s o m i t t e d ) . 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0098183/manifest

Comment

Related Items