Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The reletionship of television viewing and demographic variables with knowledge and attitudes about sexual… Johnston, Lorri 1994

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

Item Metadata

Download

Media
831-ubc_1994-0463.pdf [ 7.91MB ]
Metadata
JSON: 831-1.0054038.json
JSON-LD: 831-1.0054038-ld.json
RDF/XML (Pretty): 831-1.0054038-rdf.xml
RDF/JSON: 831-1.0054038-rdf.json
Turtle: 831-1.0054038-turtle.txt
N-Triples: 831-1.0054038-rdf-ntriples.txt
Original Record: 831-1.0054038-source.json
Full Text
831-1.0054038-fulltext.txt
Citation
831-1.0054038.ris

Full Text

THE RELATIONSHIP OF TELEVISION VIEWING AND DEMOGRAPHIC VARIABLES WITH KNOWLEDGE AND ATTITUDES ABOUT SEXUAL HARASSMENT by Lorri Johnston B.A., Simon Fraser University, 1991 A THESIS IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS in THE FACULTY OF GRADUATE STUDIES Department of Counselling Psychology We accept this Thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA September 1994 © LORRI JOHNSTON, 1994 In presenting this thesis in partial fulfillment 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. (Signature) Department of Counselling Psychology The U n i v e r s i t y of B r i t i s h Columbia Vancouver, Canada Date September 30, 1994 11 ABSTRACT The purpose of this study was to explore whether television could be used as a meaningful way of raising people's awareness and knowledge about a social issue, namely, sexual harassment. The studied explored whether there was a relationship with those who watched television programs dealing with sexual harassment from those who did not watch such programs in terms of their knowledge and attitudes about sexual harassment in the workplace. Additionally, whether or not people perceived that their own attitudes changed after watching television programs about sexual harassment was also explored. A random sample of 500 men and women was drawn by computer from a large union in British Columbia. Of those randomly sampled, 239 (47.8%) returned useable surveys. Standard multiple regression analyses were performed to answer the six research questions that examined the relationship of age, gender, education level, socioeconomic status, ethnicity, experience of being sexually harassed, number of television programs watched about sexual harassment, and sources of knowledge (newspaper, radio, television, magazines, friends, information from union, other) with attitudes about and knowledge of sexual harassment and perceived change of view about sexual harassment. Simultaneous regression analysis indicated that there were four variables predictive of attitudes about sexual harassment for the entire group of survey respondents. These variables were gender, number of television programs watched, age, and information. These four predictor variables indicated that male respondents would be more tolerant and accepting of sexual harassment than female respondents; older respondents would be more tolerant and accepting of sexual harassment; respondents who watched television programs i i i about sexual harassment, and respondents who got information from friends would be less accepting and tolerant of sexual harassment. For the knowledge criterion, simultaneous regression analysis indicated that there were two variables predictive of knowledge about sexual harassment, gender and information from friends, with gender being the strongest predictor. Being female and getting information from friends was predictive of being more knowledgeable about sexual harassment. For the perceived change of view criterion, only one independent variable, getting information about sexual harassment from magazines, was predictive of perceiving a change of view about sexual harassment associated with watching television programs about sexual harassment. Separate regression analyses run by gender were also reported in the study. Although the findings of the study were characterized by weak relationships, results may be suggestive of important implications for future training in terms of sexual harassment. iv TABLE OF CONTENTS ABSTRACT ii LIST OF TABLES ix ACKNOWLEDGEMENTS x CHAPTER I. INTRODUCTION 1 A. Nature of the Problem 1 B. Purpose of the Study 6 C. Significance of the Study 7 D. Defintion of Terms 7 E. Research Questions 9 CHAPTER H. REVIEW OF THE LITERATURE 10 A. Introduction to the Chapter 10 Media Introduction 10 B. Review of Television Literature 11 Media: The law associated with sex role socialization 11 Agenda Setting 14 Cultivation Analysis . 19 Perceived Reality 20 Creation of awareness of social problems through mass media 21 Knowledge Gap 22 Message rJiscrimination 23 C. Review of Sexual Harassment Literature 23 Demographic Variables Associated with Sexual Harassment 23 Personal Experience of Being Sexually Harassed 25 Perceptions and Attitudes About Sexual Harassment 27 V D. Research Questions 27 CHAPTER UI. METHODOLOGY 30 A. Design of the Study 30 B. Subjects 31 C. Procedure 31 D. Instrumentation 32 a. Attitude Scale: Section 1 32 b. Knowledge About Sexual Harassment: Section 2 34 c. Experience of Being Sexually Harassed: Section 3 35 d. Sources of Knowledge About Sexual Harassment: Section 4 36 e. Number of Television Programs Watched: Section 5 37 f. Perceived Change of View About Sexual Harassment: Section 6 .... 38 g. Demograpahic Variables as Independent Variables 39 h. Demographic Variables as Descriptive Information 41 E. Data Analysis 42 Rationale for Choice of Simultaneous Multiple Regression 44 Treatment of Missing Data 45 CHAPTER IV. RESULTS 47 A. Return Rate of the Sample 47 B. Summary of Descriptive Statistics 48 C. Descriptive Statistics of the Survey Respondents by Section 53 Total Attitude Sum 53 Total Knowledge Sum 54 Total Perceived Change of View Mean 59 vi Total Experience Sum 60 Sources of Knowledge 63 Number of Television Programs Watched 65 Union Component 66 D. Multiple Regression 66 Assumptions 67 Results of Multiple Regression for Dependent Variable: Mean Total Attitude Sum 70 Results of Multiple Regression for Mean Total Knowledge Sum 74 Results of Regression for Total Perceived Change of View Mean 77 Regression Analyses for Women 81 Regression Analyses for Men 86 E. Summary of Results 90 Attitudes About Sexual Harassment 90 Knowledge of Sexual Harassment 92 Perceived Change of View 94 CHAPTER V. DISCUSSION 96 Purpose of the Study 96 Restatement of the Research Questions 96 Discussion of the Survey Return Rate 97 Representativeness of the Sample 98 Experience of Being Sexually Harassed 99 Attitudes About Sexual Harassment 104 Knowledge of Sexual Harassment 112 Perceived Change of View About Sexual Harassment 117 Limitations of the Study 125 vii Implications for Further Research 132 Implications for Counselling 134 Conclusions 135 REFERENCES 137 APPENDIX A: Introduction Letter and survey 146 APPENDIX B: Follow Up Letters 152 APPENDIX C: Sources of Knowledge: OTHER 154 APPENDIX D: Table D-l: Ethnicity Responses Exactly as Subjects Answered 155 APPENDIX E: Description of Missing Data by Section 156 APPENDIX F: Table F-l: Survey Return Rates by Mailout 157 APPENDIX G: Table G-1: Perceived Change of View Means for Women and Men with Typing Errors as Strongly Agree 158 APPENDIX H: Table H-1: Descriptive Statitstics of Respondents with the Highest Total Experience Sums 159 Table H-2: Experience of Sexual Harassment for All Survey Respondents in Terms of Frequency for Each Item of Section 3 160 Table H-3: Experience of Being Sexually Harassed: Percentage of Males and Females responding to Each Item by Category 162 APPENDIX I: Table 1-1: Sources of Knowledge: Mean and Standard Deviation Scores 164 Table 1-2: Sources of Knowledge: Frequencies and Percentages of Women and Men Responding to Each Category 165 APPENDIX J: Table J-l: Number of Television Programs Watched by Episode 166 Table J-2: Total Number of Programs Watched 166 APPENDIX K: Table K-l: Number and Percentage of Survey Respondents in Each Union Component 167 Table K-2: Comparison of Survey Respondents to Union Population by Component 167 Table K-3: Comparison of Percentage of Subjects in the Respondent Sample to the Entire Union Population by Union Component Adjusted for Inclusion Error 168 APPENDIX L: Table L-1: Correlation Matrix for Attitude and Knowledge Regressions 169 Table L-2: Correlation Matrix for Perceived Change of View Regression 170 APPENDIX M: Table M-l: Correlation Comparisons of Men and Women 171 APPENDIX N: Table N-1: Combined Percentage of Agreement and Disagreement with Attitude Statements of Section 1 for Men and Women 172 ix LIST OF TABLES Table 1: Mean Age of Survey Respondents and Number of Respondents in Age Groups of Ten Year Increments 50 Table 2 : Education Levels of Survey Respondents 50 Table 3 : Socioeconomic Status of Survey Respondents 51 Table 4 : Current Work Status of Survey Respondents 51 Table 5 : Marital Status of Survey Respondents by Category 52 Table 6 : Household Income of Survey Respondents 52 Table 7 : Mean Attitude Scores of Women and Men 56 Table 8 : Mean Knowledge Scores of Women and Men 57 Table 9 : Mean Perceived Change of View Scores for Women and Men 61 Table 10: Standard Multiple Regression: Mean Total Attitude Sum 73 Table 11: Standard Multiple Regression: Mean Total Knowledge Sum 75 Table 12: Standard Multiple Regression: Perceived Change of View 79 Table 13: Standard Multiple Regression for Females: Mean Total Attitude Sum 83 Table 14: Standard Multiple Regression for Females: Mean Total Knowledge Sum 83 Table 15: Standard Multiple Regression for Females: Total Perceived Change of View Mean 85 Table 16: Standard Multiple Regression for Males: Total Attitude Sum 87 Table 17: Standard Multiple Regression for Males: Mean Total Knowledge Sum 87 Table 18: Multiple Standard Regression for Males: Total Perceived Change of View Mean 89 X ACKNOWLEDGMENTS I would like to thank my advisor, Dr. Richard Young, for his support and guidance. Additionally, I also wish to thank Dr. Bonnie Long and Dr. Charles Curtis for their ideas and expertise. A very special thanks goes to Chan Choon Hian for his technical assistance and particularly for his patience. Finally, I would like to express my deepest appreciation to my friends and family who have been supportive throughout this entire process and who have never waivered in their belief that I would complete my goal. I CHAPTER I. INTRODUCTION A. Nature of the Problem Sexual harassment is a social problem that has only been identified and recognized as a problem since the 1970s (Farley, 1978; MacKinnon, 1979). Surveys have shown that sexual harassment is pervasive in our society. Prevalence estimates range from 42 to 90% of working women (Terpstra & Baker, 1989) and 20-30% of female college students (Dzeich & Weiner, 1984). Surveys of sexual harassment to date have found that the vast majority of sexual harassers are male, parallelling the great majority of women as the victims (Pryor, 1987). Although sexual harassment generally occurs more often to females, it has been shown that males also experience it but to a much lesser degree (U. S. Merit Systems Protection Board, 1981). There is no way of knowing whether the incidence of sexual harassment that has been demonstrated in the past two decades reflects an increase or decrease of sexual harassment in our society. Women's early writings indicate that as long as women have worked outside of the home, they have experienced what we now name as "sexual harassment" (Bularzik, 1978). It was in the naming that finally gave women a common term to speak about a common experience. Without a name, social problems cannot be defined, analyzed, nor brought to the attention of society (MacKinnon, 1979). There are two main problems associated with sexual harassment. The first of these is its harmful effects to the individual as well as the employer. Individual costs of sexual harassment are high and perhaps immeasurable in the sense that the victim may suffer from decreased concentration, loss of ambition, depression, denial of promotion, loss of job, and loss of self-esteem (Farley, 1978; MacKinnon, 1979). Changing or transferring jobs because of sexual harassment reduces the likelihood of promotion or further training based on 2 experience, thus affecting a woman's long-term career prospects (Hemming, 1985). In addition, sick pay and pension rights linked to years of service are lost. With our present economic climate, finding comparable work may be difficult and may lead a woman to taking a lower status job or becoming unemployed (Hemming, 1985). In terms of organizational costs: absenteeism, low productivity, low morale, employee turnover, and high costs of litigation result from the failure to address problems associated with sexual harassment in the workplace (Sandroff, 1988). In terms of education, sexual harassment can adversely affect students' learning and academic standings. Harassment can result in lower grades, changing classes, changing majors, or even quitting university altogether (Cammaert, 1985). These factors can all contribute to women's career paths and eventual employment or underemployment. The second problem with sexual harassment is how to effectively stop it. On the surface it appears that inroads have been made in addressing the problems of sexual harassment. However, in terms of justice, "the law of sexual harassment is still in its infancy, and its interpretation and enforcement remain heavily politicized" (Stark, 1992, p . 116). Attempts by women to prosecute their harassers have failed largely due to the difference in views that men and women have in regard to what is acceptable conduct and what behaviors serve to demean and demoralize (Brenneman, 1992). According to Brenneman (1992), our laws reflect a male point of view and we may have judges who are appointed to uphold and enforce laws on sexual harassment that possibly share the same views of those who harass. In terms of social change, it can also be asked whether initiation of policies against sexual harassment in the work place; laws against discrimination on the 3 basis of sex; and educational efforts made by employers and universities are effectively reducing sexual harassment. According to Riger (1991), complaints alleging sexual harassment in the United States have decreased since 1984 even though the number of women in the workforce increased during that time. This could be looked at optimistically if it were not for surveys that suggest that the rate of sexual harassment has remained relatively stable (U. S. Merit Systems Protection Board, 1981,1988). When you look at men and women and their social expectancies of work and social roles, you can see a pattern of slow change with an underlying layer of continuity. Historically, Canadian women at the end of the nineteenth century and the beginning of this century were thought to be the embodiment of purity and gentleness, a creature in need of man's protection (Prentice et al. , 1988). Home was a woman's proper sphere and looking after her husband and children was her role in life. Coupled with this was the belief that women were weak and fragile creatures susceptible to emotional disorders. Medical doctors also believed that a woman's sex drive was no where near as strong as a man's and thus it was her responsibility to keep both under control. In the Victorian era, it was simply accepted that women were weaker and less emotionally stable than men and easily led to the view that women needed to be protected and controlled (Prentice et al . , 1988). Our laws reflected these views. Women in Canada have come a long way in achieving equality with men in a relatively recent period of time. Women in Canada only got the right to vote; the right to have an education; the right to use birth control; the right to earn their own money and have control of it; the right to have child custody; the right to own property and be able to deal with it as they saw fit; and the right to divorce during the earlier decades of this century (Prentice et al. , 1988). Some 4 of these social gains only began to occur as recently as the late 1930s. Underlying each of these social wins, the primacy of woman as wife and mother remained constant and reinforced. For example, higher education for women was accepted reluctantly and with a view to reinforcing women's traditional roles, deeming the teaching of domestic science, art, music, and literature as appropriate for a woman (Prentice et al. , 1988). This in turn led to regarding only certain careers as being appropriate for a woman, i. e. , teaching, nursing, and clerical (Prentice et al. , 1988). During World War II women went out to work in the factories to replace the men who were off to war. Previous to this period of time, it was ordinarily not "socially correct" for a woman to work outside the home. The war effort briefly changed this view. However, at the end of the war, social campaigns occurred reminding women that their place was in the home and that they should not take jobs away from the male breadwinner of the family (Prentice et al. , 1988). By 1944, the federal government of Canada recognized the inadequacy of the "family wage" and instituted the family allowance. According to Prentice et al. (1988), the underlying assumption of this allowance was, except in emergency situations, married women were responsible for the children and should not be wage earners. Jumping to the 1990s, there remains some continuity to women's role in society. In terms of sexuality, there still appears to be a view that women are responsible for sexual behaviors in relationships. Women's dress and behavior are still generally looked at and examined in great detail in matters of rape and sexual harassment charges. In terms of work, by the 1960s it was widely accepted that childless married women, or women whose children were in school, could work outside the home and in a 1976 Gallup poll, 81% still thought that when children are young, a 5 mother's place is in the home (Prentice et al. , 1988). Today women are working outside of the home in unprecedented numbers. The majority of working age women participate in the labour force (Statistics Canada, 1990). By 1988, women accounted for 44% of the labour force with 59.1% of married women and 79.5% married men in the work force (Statistics Canada, 1990). Although great numbers of women are in the labour force they have generally been occupationally segregated (Peitchinis, 1989). Role differentiation, which limited women to the home and family, has largely become a remnant of the past. However, the notion that some work activities are suitable for men and some suitable for women may still remain. According to Peitchinis (1989), custom-based occupational choices within a society give the impression of being free choices and that often what appears to be freely chosen has often been dictated by underlying social forces of great strength and authority. Custom does not respond easily to change. Although the percentage of women in the workforce keeps climbing, there is some continuity of cultural beliefs about women and work that act to create an atmosphere for discrimination. Peitchinis (1989) lists four conditions that predispose women to discriminatory work practices against them. One is the predominance of women in ancillary work activities, commonly under the direction of men. Women are seen as the assistant, often even in professions where women predominate. This is a continuity of views about women from an older era—women are in need of protection and guidance. The second factor is the belief that women are supplementary family workers whose earnings are meant to supplement the family income (Peitchinis, 1989). This notion has not kept up with the reality in life. By 1983, only 16% of family incomes was provided by husbands as sole breadwinner (Prentice et al. , 1988). Once again, a continuity in beliefs—males are the primary breadwinner, therefore their 6 work is more important. The third belief is that the productivity of women is generally lower because of lower motivation of women in work outside of the home due to the conflicts of responsibilities of home, family, and work (Peitchinis, 1989). The fourth condition is the frequent discontinuity of many women in the workplace who have left work to have and raise children. This prevents women from active participation resulting in failure to gain seniority and experience, unlike those who remain in their jobs continuously without interruption (Peitchinis, 1989). This brief review points out the rapid change in women's rights and their ensuing roles in a relatively short period of time with lagging changes of attitudes. Each female worker may work beside a male or female worker who has very different views about where women should be in terms of work, in terms of family, in terms of behavior, and in terms of values. It is with this in mind that the challenge in successfully deterring sexual harassment may exist in the challenging of the continuity of beliefs about women still embedded in our society. The question is, how do you do that? B. Purpose of the Study The purpose of this study is to explore whether television can be used as a meaningful way of raising people's awareness and knowledge about a social issue through seeing by example. In particular, the purpose of this study is to explore whether there is a relationship with those who watch television programs dealing with sexual harassment from those who do not watch such programs in terms of their knowledge and attitude about sexual harassment in the workplace. Additionally, whether people perceive their own attitudes change as a result of watching television programs on sexual harassment will be explored. 7 C. Significance of the Study Sexual hararassment has been generally described as a barrier to women's career opportunities, career choices, and career advancement. Repeat surveys on sexual harassment (USMSPB, 1981, 1988) indicate that the rate of sexual harassment may be staying relatively stable despite attempts to prevent it. The challenge is to find creative solutions to a difficult problem that is exacerbated by the widely varied views of men and women in relationship to sex-roles, women and work, and sexual harassment itself. Providing education and information without attitude change may not be adequate to prevent sexual harassment. It is hoped that this study could demonstrate that television could aid in changing peoples' views as well as educating and raising awareness about sexual harassment. It is also hoped that this study will add knowledge to the existing literature on attitudes and attitude change. In terms of career counselling, this study may also add information on how attitudes and perceptions vary in men and women in regard to different aspects of sexual harassment and perhaps give some insight into what issues are problematic for women in the world of work. D. Definition of Terms There is no single definition of sexual harassment that has been totally satisfactory to those who study it. There is a wide range of behaviors that can be characterized as sexual harassment. Some of these behaviors are: repeated and unwanted sexual advances, leering, ogling, jokes, innuendoes, looks, "accidentally" brushing against a woman's body, a "friendly"pat, a pinch or squeeze (Cammaert, 1985). Other behaviors also described are: unwanted letters, phone calls or materials of a sexual natured; requests or demands for sexual favors accompanied by implied or overt threats about job or letters of 8 recommendation; pressure for unwanted dates; and physical assault or actual rape (Sandler, 1990). In terms of the law, the Equal Employment Opportunity Commission, EEOC (1980) defines sexual harassment as: Unwelcome sexual advances, requests for sexual favors, and other verbal or physical conduct of a sexual nature when submission to such conduct is made either explicitly or implicitly a term or condition of an individual's employment; submission to or rejection of such conduct by an individual is used as the basis for employment decisions affecting such individual; or such conduct has the purpose or effect of unreasonably interfering with an individual's work performance or creating an intimidating, hostile, or offensive working environment, (p. 74677) This legal definition is also followed and is reflected in the Human Rights Act of British Columbia. In addition to the EEOC guidelines (1980), five categories of behavior formulated by Till (1980) from a content analysis of a national sample of college women using an open-ended format, were used to operationalize the definition of sexual harassment. The five categories of behavior by Till are: (1) gender harassment-generalized sexist remarks and behavior which convey insulting, degrading, or sexist attitudes about women. The remarks are not necessarily meant to elicit sexual cooperation; (2) seductive behavior-inappropriate and offensive behavior that is basically sanction-free with no penalties attached for a woman's negative response; (3) sexual bribery-seeking sexual activity or sex -related behavior in exchange for reward; (4) threat-coercion of sexual activity through threats of punishment; and (5) sexual impostition-sexual imposition and assault Till's categories reflect both "quid pro quo" (i. e. , sexual cooperation coerced by promises of reward or threats of punishment) and 9 "conditions of work" (i. e., sexist and/or sexually offensive behavior or unwanted sexual behavior and attention at work or in academia that do not require sexual exchange). E. Research Questions In essence, this was an ex post facto design, where there was no experimental manipulation and subjects recalled information about events that had already occurred. Simultaneous multiple regression analysis was chosen as the primary statistical analysis to answer the following questions: (1) Is there a variable or combination of variables that might predict how knowledgable someone is about sexual harasssment in the workplace?; (2) Could this variable, or combination of variables also predict people's attitudes about sexual harassment?; (3) Is there any relationship between people's sources of knowledge about sexual harassment and their actual knowledge about sexual harassment or their attitudes about it?; (4) Specifically, is there a relationship with viewing television programs about sexual harassment with knowledge or attitudes about sexual harassment?; and (5) If so, what variables might be related to someone perceiving that their views about sexual harassment had changed? 10 Chapter II. Review of the Literature A. Introduction In this first section, a general review of literature relevant to television and other media is presented. Included in this is a discussion of how television presents symbolic representation of sex roles and the law. In addition, information about how the general public feels about televised reporting of rape trials is presented. Agenda-setting, cultivation analysis, perceived reality, knowledge gap, and message discrimination are briefly discussed. A brief review of relevant information from the sexual harassment literature follows the media review. Media Introduction Mass communication occupies a central role in our lives. "Newspapers, radio, and television are major sources of information, ideas, and images concerning events which take place beyond our immediate social milieu" (Thompson, 1988, p. 360). In this past decade there has been a renewed belief that mass media powerfully influences the way that people perceive, think, and ultimately act in their world (Roberts & Bachen, 1981). Television and other forms of media serve as an "ideological apparatus" serving to reproduce and sustain the social order (Thompson, 1988). Literature to date seems to argue that media can be a powerful social agent in both positive and negative directions (Fisher, 1989). According to Behr and Iyengar (1985), the amount of media attention given to particular issues determines the amount of public concern for these issues. Through emphasis in the media, people learn the importance of particular issues. According to Hubbard, DeFleur, and DeFleur (1975), through group activities and social exchanges, public awareness of the problem and shared definition of the social problem begin to be formulated. If a problem is given wide recognition it A -11 may become legitimized. Very little is known about the specific roles media plays in this process. In the fall of 1991, millions of television viewers in the United States and Canada watched two television spectacles—the rape trial of William Kennedy Smith and the Senate confirmation hearings of Judge Clarence Thomas where he defended accusations of his alleged sexual harassment of his former employee, Anita Hill. Since the Thomas-Hill hearings, complaints to companies and government agencies skyrocketed. In the nine months following the hearing, sexual harassment inquiries sent to the Equal Employment Opportunity Commission (EEOC, 1980) in the United States rose 150% and actual charges filed rose 23% (Sandroff, 1992). These figures suggest that sexual harassment became recognized by more than those who write about it and that extensive coverage of a social issue on television may have had an impact on some of the viewers. Listed below are studies demonstrating television and media impact. B. Review of Television Literature Media: the law associated with sex role socialization In the case of public trials media can construct powerful images of a case through selective reporting of information and structuring of perceptions (Bumiller, 1990). Criminal trials are symbolic in that the defendants and victims come to represent social roles (Bumiller, 1990). In examining both the William Kennedy Smith rape trial and the Thomas Senate confirmation hearings, sex-role socialization may have played a factor in the participants' view as well as the interpretations made by the television viewing audience in regard to whether rape or sexual harassment had occurred and who would be considered to be blameworthy. It is this "blamelessness in contributing to her own harm" that is often the focus on the female accuser. As Bumiller (1990, p. 127) states, 12 "The accuser is forced into the role of an 'angel' who must defend her heavenly qualities after her fall from grace." According to a learning/conditioning model, social sex-roles are the predominant influence upon attitudes and behavior of males and females (Terpstra & Baker, 1986). Male sex-roles encourage men to be dominant, aggressive, and forceful whereas female sex-roles teach women to be passive and acquiescent (Terpstra & Baker, 1986) . Women are taught to seek their self-worth in the evaluation of others, particularly men (Barwick as cited by Tangri, Burt, & Johnson, 1982). In keeping with these roles, women have also been socially conditioned to be responsible for controlling sexuality while men initiate it (Bularzik, 1978). Women "somehow" are at fault for their own victimization according to our social roles and when women rise above their own internalized guilt for an assumed failure to control the situation they are blamed for "leading him on." Because of the tendency to "blame the victim" men have power in situations of harassment and rape where there is no corroborative testimony for the victim. Consequently, women who report harassment may be accused of being seductive, vindictive, mistaken, or maybe thought to have misinterpreted or misrepresented the situation for their own benefit (Walker, Erickson, & Woolsey, 1985). In addtion, women who have traditional sex-role values may blame other women as well as themselves for incidents of sexual harassment (Jensen & Gutek, 1982). Sexual harassment can also be largely ignored as a problem because its aggressive aspects are unrecognized and can become confused with courting behavior (Hemming, 1985). In these terms then, the viewing audience saw our sex-roles played out in the courtroom venue of the rape trial and the drama of the Senate hearings, leaving stereotypical views of women's responsibilites in terms of sexuality and relationships virtually unchallenged. 13 Hans and Dee (1991) state that viewers use television to construct images of crime and violence as well as learning about the court system and the legal process. According to Stark (1990), the media did a poor job of explaining the problems that Anita Hill would have had in coming forward 10 years earlier. He suggests that a fair representation of the issues was not made because of the selective attention and reporting of the media. The following study is an example of how powerful a role the media may play in contributing to public opinion and beliefs. Public opinion about television coverage of rape trials was exmined using a cross-sectional random sampling probability sample of the Minneapolis-St. Paul metropolitan area (Swim & Borgida, 1987). Public beliefs about the legal and constitutional issues, the psychological effects of the media on trial participants, and the extent that effects of media coverage had on the public were studied. Descriptive results indicated that the general public disapproved of television coverage of courtroom trials. In addition, women reported that they would be less likely to report a rape knowing that other rape trials had been televised. Multiple regression analysis indicated that women's behavioral intention to report a rape were highly related to believing that television rape trials would increase a rape victim's trauma. Also significant was viewer diaspproval of televised rape trials, which was closely related to respondents' views of the symbolic constitutional issues. Of the respondents interviewed, 42.7% believed that televised rape trials would raise awareness and 35.2% believed that televising rape trials would encourage more rapes. It was noted that 78.8% of respondents had not seen the broadcast so a meaningful comparison between viewers and nonviewers could not be analysed. However, this study does infer that those polled do believe that televised viewing of a rape trial would affect their views. 14 Agenda Setting According to agenda setting hypothesis, the amount of media attention given to particular issues determines the amount of public concern for these issues, that is, mass media does not tell us what to think, but tells us what to think about (Behr & Iyengar, 1985). In a study of agenda setting, Hill (1985) analysed viewing diaries of 1204 adult subjects. Participants kept a program-by-program log for a 2 week period. For each program watched, participants were asked to report on attentiveness and attitudes toward television. Other viewing traits and general questions asked were not described. Only subjects seeing at least one news program on at least one occasion made up this group. Subjects were asked to rank order their interest in each social and political issue. These personal agendas were compared to media agenda which were ranked according to the number of stories done on each issue. A comparison of the media agenda and personal agendas showed only a weak relationship. Of the top 10 issues given most attention by media, only 3 of these were ranked in the top 10 of personal agendas. Although agenda setting was not confirmed among news viewers, Hill stated that the actual magnitude of agenda setting discovered was relatively unimportant. This rationale was justified in that agenda setting is more difficult to demonstrate in cross-sectional surveys. From multivariate analysis the researcher concluded that personal characteristics and viewing habits, particularly viewers' news knowledge and attentiveness to news programs are more important in agenda setting than quantity of exposure. Print media exposure, college education, and paying full attention to news programs were the strongest and only significant predictors of interest in the news topics studied. 15 Volgy and Schwartz (1980) studied the potential impact of television entertainment programming on socio-political attitudes. Content analysis was performed for a 4 week peried for each network. Unanimity for three themes were obtained. All programs that focused on medical doctors portrayed them as wise, helpful, intelligent, and empathetic. Most programs represented male/female relationships according to traditional values and norms. The third theme was the treatment of minorities. Blacks were often portrayed as if they were living in a society relatively free of racial and ethnic bigotry or severe economic problems. Unique problems of blacks were seldom explored in the televised programs. A paper and pencil questionnaire was mailed to a random sample of voters in a southwestern metropolitan community of the United States. There was a 66% return rate with a 90% satisfactory survey completion rate. Three sub-groups in the area lacked sufficient response rates to be included in the analysis, therefore results only represent "Anglos". Three hypotheses were developed and confirmed by this study: (a) as exposure to medical programs increased, the viewers' positive affect toward doctors increased; (b) as exposure to entertainment programs in general increased, viewers' acceptance of traditional sex roles increased; and (c) as exposure to ethnic programs increased, the viewers' concerns about racial problems in society decreased. The researcher suggested the following: (a) viewers may select programs that conform to their pre-existing attitude. However, if this is the case, television may then reinforce existing attitudes; (b) the study was not generalizable; and (c) there was no relationship between number of programs preferred and the number of programs watched. In addition, there was no relationship between the type of program preferred and the type of program watched thus indicating that self-selection was not a dominant factor in determining entertainment program choice. 16 Leff, Protess, and Brooks (1986) did an analysis of the agenda setting effects of a 5-part television investigative series of repeatedly brutal police officers. These researchers surveyed 428 members of the general public in the Chicago area on their attitudes toward various social issues, including the conduct of city police officers. Preseries measurement of public attitudes toward the problem was taken and following the series, a postseries measurement was performed. Questions about police brutality were purposely embedded among questions about other social issues to avoid sensitizing respondents to the issues. Data for the general public survey indicated that significant attitude change occurred among those exposed to the series. Persons in the exposed groups increased their views of the importance of police brutality and showed significant changes in attitude toward police brutality in the posttest survey. However, on factual questions related to the specific content of the police brutality series, change in the exposed groups was weaker than attitudes about police brutality. Leff et al. (1986) also suggested from their research that it is possible that the less information and sensitivity the public has to an issue prior to its treatment in the media, the more likely the public will be influenced by subsequent media reports. They suggested that there may be a saturation level on the public's agenda where the public may already have knowledge and have formed opinions on the issue. Wilson, Linz, Donnerstein, and Stipp (1992), conducted a field experiment to evaluate the impact of a television movie about acquaintance rape on subsequent attitudes about rape. A nationally representative sample of 1,038 American male and female adults from three age groups (18-34; 35-49; 50 years and older) were selected for the study. From this group, individuals were randomly selected to view or not to view the made-for-TV movie, "She Said No". Five hundred and twenty-six subjects viewed the movie over a closed-circuit 17 channel prior to the network broadcast of the film while 512 subjects were assigned to the control group. After showing the movie, the viewing group and the non viewing group were assessed for their acceptance of rape myths and perceptions of rape as a social problem. Results of the study indicated that a dramatic movie on television was an educational tool in altering perceptions about the social issue, date rape. Gender and age were found to be significant factors in relationship to rape myths. In this study, females were less likely than men to blame the woman in a date rape situation. They were also more likely to perceive date rape as wrongful coercion, and more likely to be concerned about date rape as a societal problem than men. Adults over 50 were significantly more likely to blame the woman in a date rape situation than the two younger age groups and were significantly less likely to perceive date rape as wrongful coercion than the youngest age group. In addition, subjects who had personal knowledge of a rape victim were less likely to blame the woman in a date rape situation. These subjects were also more concerned about the societal problems of date rape than those who did not personally know a rape victim. In terms of impact of the movie, there appeared to be a mass media agenda-setting effect. All viewers, regardless of age, gender, and personal knowledge of a rape victim had a heightened awareness of the social problems associated with acquaintance rape. Viewers perceived the problem of date rape as more serious than non viewers. Those who saw the movie also tended to more strongly agree that the legal system is biased against women in rape cases. Additionally, 80% of the viewers actually believed that the fictional story in the movie was based on a true story. The researchers of this study suggested that this believability of the movie was a contributing factor to its positive effect. 18 Although the results suggest that the movie was associated with an agenda effect, it had a somewhat limited effect on rape attitudes. It was suggested that prior attitudes mediated the impact of the movie. It was noted by Wilson et al. that previous research indicated that media campaigns may often need to combine with interpersonal and community support systems when trying to change deep-rooted attitudes and behaviors. The researchers did however, isolate a subgroup, older women, for which the movie affected attitude change after exposure. It was found that all women, regardless of their knowledge of a rape victim, reported more emotional involvement with the film than did the men and this was used to explain the attitude change for older women. It was suggested those subjects more emotionally involved with the movie would more carefully attend to its message. It was suggested that younger women would already accept the message of the movie since the control group demonstrated that the younger subgroups of women were less accepting of rape myths than the oldest group of women. This would suggest that the educational impact of the movie brought older women in Une with the perceptions of the younger women in terms of victim blame. In contrast, it was interpreted that older men, through the principle of selective perception, construed the movie differently from all the other groups. Older men seemed to interpret and process the movie in a stereotypical fashion and the antirape message had little effect for this group. The movie actually tended to reinforce certain rape myths (i. e. , the woman is responsible, the man suffers more) for the older male subgroup. The researchers suggested that this result would be consistent with previous research that has found that viewers frequently identify and empathize with same-sex rather than opposite-sex characters. Since the movie was presented primarily from the female victim's perspective, older male subjects would be 19 less emotionally involved with the victim than older female subjects and thus would be less likely to focus on the message content of the film. Cultivation Analysis According to Gerbner, Gross, Morgan, and Signorielli (1980) television is the dominant mass medium in our culture. From a cultivation analysis perspective, they suggested that the more time spent watching television, the more likely conceptions of social reality will reflect what people see on television. In addition, television viewing contributes to the cultivation of common perspectives (mainstreaming) among diverse groups of people. Heavy viewers of television tend to report perceptions of social reality that can be traced to television's consistent portrayals of life and society. In addition, particular salience to specific issues may boost the cultivation effect when the viewer sees on television, programs which are congruent with real life experience or even "perceived reality." This congruence of real-life and the television world (resonance) increases the likelihood of cultivation effects. Cultivation analysis of television and conceptions about sex roles was explored in a study of prime-time dramatic television programming and the relationship between television viewing and espousing sexist views of the roles of men and women in society (Signorielli, 1989). The design of this study consisted of two interrelated procedures: (1) examining content analysis of prime-time network dramatic programs broadcast between 1969 and 1985, and (2) cultivation analysis examining the conceptions of social reality in different groups of television viewers using data from the National Opinion Research Center (NORC) General Social Surveys of 1975,1977, 1978, 1985, and 1986. This study confirmed that women are still generally underrepresented on television. According to this study, in a 17 year period from 1969 to 1985, prime-time programs were populated by 2-3 males for every female. Only 3 out of 10 20 female characters were employed outside the home according to television, quite different from the "real world" where more than half of all married women work (Signorielli, 1989). In addition, women on television were often cast in traditional female occupations in less prestigious jobs as men. On television only 29% of the professionals were women whereas in the U. S. labor force, there are approximately equal numbers of men and women in the professions (Signorielli, 1989). In addition, women were portrayed as younger than men, less powerful, more attractive and nurturing. They were portrayed in the context of romantic interests, home, and family, and were more likely to be victimized (Signorielli, 1989). In terms of cultivation analysis, this study provided mixed support for the general hypothesis that those who watch more television will have more sexist views and the mainstreaming hypothesis, i. e. , heavy viewers would have more similar outlooks than light viewers in regard to woman's role in society. What was demonstrated was a trend between the 1970s and 1980s for fewer subjects to agree with sexist statements. In addition, more recent surveys (1983, 1985, and 1986) indicated a small, negative relationship between television viewing and sexism-interpreted as reflecting the "liberalization" of mainstreaming. Analyses also showed that attending college and sexism are negatively related. Perceived Reality Potter (1988), in a review of perceived reality effects of television, identified three dimensions that define the construct of perceived reality: magic window, social expectations of perceived reality, and identity. Magic window is the central component of perceived reality and is defined as the belief in the literal reality of television messages. Perceived utility is defined as the applicability of television conveyed lessons to a viewer's own life. This component infers that some viewers will feel that they can learn a great deal vicariously by watching 21 the role models succeed or fail. Other viewers will believe that television is fiction and that there are no real-life lessons to be learned from television. The third component "identity" is a feeling of closeness to characters on television and the development of a strong sense of realness about the characters. This can result in a strong feeling towards the characters similar to people in the viewer's own life. Potter (1988) suggested that these dimensions should be taken into consideration in guiding item design for perceived reality. According to Potter (1988), when people are given explicit cues that media material is real they are more likely to change in behaviors and attitudes in the direction of the real material versus nonfiction material. In a study of law enforcement program viewing, Slater and Elliott (1982) concluded that direct experience is an important variable in the study of perceived reality. In addition, it was suggested that viewing a particular program thought to be highly realistic and viewed over a long period of time might produce a noticeable influence on social reality independent of other viewing behaviors. Creation of awareness of social problems through may, r^priia Although there does not appear to be a particular theory of mass communication and social problems, researchers suggest that media coverage of social problems creates an awareness of these problems and a sense that something is being done about them (Hubbard, DeFleur, & DeFleur, 1975). In a study of mass media influences on social problems, Hubbard et al. (1975) analysed the amount of newsplay given a selected social problem by local newspaper and television stations over an 18 month period. An index of recorded prevalence in the news was compared to rankings of prevalence of agency records and a survey of the general public. Results indicated that amount of newsplay did not lead to accurate predictions of incidence of the 22 problem among the audience surveyed. Researchers suggested that media may possibly play a significant role in the emergent stage of social problems rather than in well known established social problems. Media's role may be in the shaping of the importance of the social problem. Lack of a relationship in incidence was posited to be related to media practices and policies in defining what is news. It was suggested that certain kinds of events would be judged as attention-getters and would have the highest probability of being included in the news (e. g. , Anita Hill and Clarence Thomas). Knowledge Gap "Knowledge Gap" is an area of research in mass communication which predicts that: "1 . Over time, acquisition of knowledge of a heavily publicized topic will proceed at a faster rate among better educated persons than among those with less education; and 2. At a given point in time, there should be a higher correlation between acquisition of knowledge and education for topics highly publicized in the media than for topics less highly publicized" (Tichenor, Donohoe, & Olien, 1970, p. 165). However, this hypothesis was in reference to print news and these authors suggested that television may be a "knowledge leveler" in some instances. Media effects on learning have found that correlations between use of mass media and levels of information are inferior to comparisons with correlations between educational attainment and information (Tichenor et al. , 1970; Wade & Schramm, 1969). However, Clarke and Kline (1974) suggested that information may be biased due to definitions used conceptualizing "knowing" and the types of questions asked to assess such knowledge. These authors suggested abandoning media use as an index of communication experience in favor of message cUscrimination. They suggested that three kinds of information are important in assessing individual knowledge from media: cognitions about the 23 personal relevance of an issue, awareness of what has been attempted to deal with a problem, and information about groups of persons trying to influence what government does about a problem. Message discrimination is measured by asking whether a person has read, seen, or heard anything about a particular issue or problem that the person has nominated as a problem. For those who recall such experiences, they are asked to describe the content and source of information. Message Discrimination Clarke and Kline (1974) in a study on message discrimination using a probability sample of 137 Ann Arbor heads of household, found that the level of education and information holding was significant but modest compared to communication behavior. In addition, it was found that people who know others who tend to agree or disagree with the person's nominated problem is a powerful predictor of information holding as level of education. As well, message cUscrimination in the broadcast media correlated as highly with information-holding as described from print media. Participants used, in order, television, newspapers, and magazines as prominant sources of information and about one out of five persons obtained information from other people. C. Review of Sexual Harassment Literature Demographic variables associated with sexual harassment In a survey of 23,000 federal employees in the United States, with a return rate of 85%, respondents were asked a series of questions regarding sexual harassment related to work in the previous two years (U. S. Merit Systems Protection Board, 1981). Results of this survey showed that sexual harassment was widespread with 42% of all female employees and 15% of all male employees reporting being sexually harassed. Results from this study indicated that sexual harassment is widely distributed among men and women of various positions 24 and backgrounds. However, there were certain personal and organizational characteristics that were associated with the more likelihood of being sexually harassed. Demographic characteristics of victims most strongly associated with harassment were age, marital status, and gender composition of the workgroup. Factors that were somewhat weaker bearing were education level, race, ethnic background, job classification, non-traditional nature of job, and sex of immediate supervisor. Conclusions from this study indicated that the typical men and women who were likely to be harassed were: young, not married, higher educated, and members of a minority, racial, or ethnic group (if male). In addition, if subjects held a trainee position; held a non-traditional job for their gender; had an immediate supervisor of the opposite sex; or had an immeditate work group predominantly composed of the opposite sex, they were more likely to experience harassment. Females were more likely to be harassed by older men and males were more likely to be harassed by younger women. It was also found that workers believed that supervisors should have higher standards of conduct than other workers regarding sexually oriented behaviors in the workplace. In addition men showed a greater tendency than women in thinking victims were somewhat responsible for bringing sexual harassment on themselves. They were also more likely to believe the issue of sexual harassment as being exaggerated. Other studies confirmed that the increased likelihood of being sexually harassed will occur if the person is young, female, and single, separated or divorced (Schneider, 1982). Additionally, women working in workplaces with a predominance of men report more incidents of sexual harassment (Gutek & Morasch, 1982; Schneider, 1982). Kanter (1977) also suggested that in skewed groups the numerically dominant type control the group and its culture. Those 25 who are numerically few are "token" and tend to stand out in the group. In these situations the salient statuses of gender, race, or ethnicity are highly noticed and these members can never be just another worker of the group. "Token" members would therefore, be more likely to be be sexually harassed. In terms of socioeconomic status, Gutek, Morasch and Cohen (1983) reported that the status of the initiator in socio-sexual interactions is more important than the sex of the initiator when people are evaluating the appropriateness of the behavior. An incident is thought to be inappropriate if it is initiated by a higher status person. According to Brewer (1982), the person in the high power position is more likely to see compliance to requests for sexual favors as motivated by personal attraction. However, the person in the low power position may see the the high status initiator as abusing their power and authority. Women are generally in lower status jobs, lower levels of authority, and generally receive less wages for their labor (Schneider, 1982) and thus would be more vulnerable to harassment. Personal experience of being sgxuajly harassed Personal experience may also be reflected in participants' awareness of sexual harassment. Mazer and Percival (1989) surveyed 210 university students to study the relationships among sexual harassment experiences, perceptions about harassment (definitions, seriousness ratings, commonness estimates), and attitudes about both harassment and sex roles. A 20 page survey asking questions about sexual harassment concerning four categories of experience was used for the study. Incidents were intended to represent a full range of experiences that would be considered to be sexual harassment. Each student was asked to indicate whether they would define the incident as sexual harassment; whether they had ever had the experience with a male source or a female source; how common they thought the experience was for a male and a female 26 student; and how serious they thought the incident was. Two attitude measures were also given: The "Macho" Scale (Villemez & Touhey, 1977), which measures individual differences in sex role stereotyping and sex discrimination and the Sexual Harassment Attitude Scale (SHAS, Mazer & Percival, 1989). High scores on the SHAS indicate more acceptance and tolerance of sexual harassment and less agreement with contemporary feminist understandings of its causes. Results indicated that women defined more incidents as sexual harassment than men. This supported previous research that indicated that women label more incidents as sexual harassment (Collins & Blodgett, 1981; Gutek et al., 1983; Konrad & Gutek, 1986; Powell, 1986). Results of this study also indicated that men are more uncertain than women about whether an incident is sexual harassment. It was also shown on the attitude scales (Macho, SHAS), which were strongly and positively related to each other, that those with more sexist and discriminatory attitudes are more likely to have more tolerant and accepting views of sexual harassment. Respondents who rated sexual harassment as serious tended to endorse fewer sexist attitudes and tended to be less tolerant and accepting of sexual harassment. In addition, those who defined more incidents as harassment had less tolerance of it and fewer sexist attitudes. Means on the Macho Scale indicated that men have more cUscriminatory and traditional sex role attitudes than women. Scores on the SHAS indicated that men also have more tolerant and accepting attitudes toward sexual harassment than women. Experience of being sexually harassed was related to perceived commonness of sexual harassment but was not related to defining more incidents as sexual harassment nor seeing incidents as more serious. This finding that experience was not related to definition and seriousness ratings was consistent with previous studies (Konrad & Gutek, 1986; Powell, 1986). 27 Perceptions and attitudes about sexual harassment Previous studies have reported that there is a tendency among men to perceive women's friendly behavior as a sign of sexual interest or availability (Abbey, 1982; Saal, Johnson, & Weber, 1989). When men (mis)interpret women's attempts at creating a friendly atmosphere at work or school, they may act on these (mis)interpretations in a way that is offensive to women. Women in turn may label men's actions as sexual harassment. Men have also been found to view male/female interactions as more sexual than women do (Abbey, 1982; Collins & Blodgett, 1981; Shotland & Craig, 1988). It has also been found that men tend to believe sexual behaviors are more normative and acceptable in the workplace (Gutek et al. , 1983) even though the ovemhelming majority of working women do not like or welcome socio-sexual behaviors in the workplace (Gutek, 1985; Schneider, 1982). Konrad and Gutek (1986) suggested that men may initiate sexual behaviors at work that lead to complaints of harassment since they are less likely to view the behaviors as unacceptable. In terms of responsibility for the occurrence of sexual harassment, sex-role beliefs were found to be related to both self-blame and blaming other women. Jensen and Gutek (1982) found that men and women differ in their assignment of responsibility in regards to sexual harassment. Men are more likely to place blame on the woman for being sexually harassed and women with traditional sex-role beliefs will be more likely to blame other women as well as themselves. In addition, women with traditional sex-role beliefs will be less likely to report their harassment to someone in authority or talk to co-workers or friends about it (Jensen & Gutek, 1982). D. Research Questions The purpose of this study is to explore whether television can be used as a meaningful way of raising peoples' awareness and knowledge about a social 28 issue, i. e. , sexual harassment. Additionally, the question is asked, whether people perceive that television changes attitudes about sexual harassment. Because this study is strictly exploratory, no research hypotheses have been made. From review of the literature on media and on sexual harassment, the following research questions have been formulated: 1. What is the relationship between the independent variables: number of television programs watched on sexual harassment, other sources of knowledge (newspapers, radio, television, magazines, friends, union, other), sex, age, ethnicity, socioeconomic status, education level, and experience of being sexually harassed with the dependent variables, knowledge and attitude about sexual harassment? 2. Is there a relationship of television viewing of sexual harassment programs with changes of view in regard to sexual harassment? 3. Does any variable or combination of variables from the independent variables of age, sex, education, ethnicity, socioeconomic status, experience of being harassed, number of television programs about sexual harassment watched, and other sources of knowledge predict level of knowledge about sexual harassment? 4. Does any variable or combination of variables from the independent variables of age, sex, education, ethnicity, socioeconomic status, experience of being sexually harassed, number of television programs about sexual harassment watched, and other sources of knowledge predict attitudes about sexual harassment? 5. Does any variable or combination of variables from the independent variables of age, sex, education, ethnicity, socioeconomic status, experience of being sexually harassed, number of television programs about sexual 29 harassment watched, and other sources of knowledge predict changes of view about sexual harassment? 30 CHAPTER III. METHODOLOGY This chapter begins with an outline of the research design and is followed by a description of the sample, procedure, instrumentation, and analysis. A. DESIGN OF THE STUDY This study was designed to explore whether there was a relationship between people who watched television programs focussed on sexual harassment with those who did not watch such programs in terms of their knowledge of and attitudes about sexual harassment. It was also designed to explore whether subjects actually perceived any changes of view about sexual harassment in relationship to watching television programs focussed on sexual harassment. The study was correlational in nature and was an ex post facto design employing a survey. Subjects were asked on the survey to recall whether they had seen television programs dealing with sexual harassment. It was expected that two groups would emerge-those who did watch and those who did not watch television programs (on sexual harassment). It was also posited that other personal characteristics such as age, gender, education level, ethnicity, and socioeconomic status might play an important role. In addition, such factors as: experience of previously being sexually harassed, number of television programs actually watched on the subject of sexual harassment, and subjects' perceived sources of knowledge about sexual harassment should be explored. Listed below are the variables of the study: Dependent Variables 1. Attitude about sexual harassment 2. Knowledge of sexual harassment 3. Perceived change of view about sexual harassment 31 Independent Variables DAge 2) Gender 3) Education level 4) Socioeconomic Status 5) Ethnicity 6) Experience of being sexually harassed 7) Number of television programs watched on sexual harassment Perceived sources of knowledge about sexual harassment: 8) Newspaper, (9) Radio, (10) Television, (11) Magazines, (12) Friends, (13)Information from union, (14) Other B. SUBJECTS The sample was drawn from a large union in British Columbia. As of October, 1993, the union had a membership of 30,839 workers comprised of 42% males and 58% females. Only members of the union who were listed by computer were selected rather than the full complement of the union membership because information on allied service workers was not readily available. A random sample of 541 subjects was generated by computer in order to achieve a final sample of 500 subjects representative of the union membership. Subjects currently on "Long Term Disability" were dropped from the list of 541 subjects and then random names of subjects were dropped to achieve the 42% male/5 8% female ratio. The union frequently updates its membership list and statistics generated for the union were compiled a month before the survey mail out. The sample was drawn from the entire membership of the union that was Usted by computer in the province of British Columbia. C. PROCEDURE A pilot study was done on a convenience sample of 39 subjects of various ages and occupations. Participants were asked to comment on the items, and clarity 32 and ease of understanding of the survey in general. On the basis of participant feedback, changes were made to the wording of some items and to some of the response categories. In addition, 2 items of the "attitude" section of the survey were deleted to improve internal consistency. Surveys to 500 government workers were sent out in the last week of November, 1993. The two follow-up mailings were sent out in the first week of January and the first week of February. The second mailing was sent out much later than the customary 2-4 week time interval because of the Christmas season. D. INSTRUMENTATION A five-page survey was designed by the researcher incorporating items from previous surveys from the sexual harassment literature for two of the dependent variables (attitudes about sexual harassment and knowledge of sexual harassment) and one of the independent variables (experience of being sexually harassed) [see survey in Appendix A] . a. Attitude scale: Section 1 This section consisted of 9 items randomly drawn from the 19-item Sexual Harassment Attitude Scale [SHAS] (Mazer & Percival, 1989) that reflects attitudes about sexual harassment. This inventory is an extension of the Tolerance for Sexual Harassment Inventory [TSHI] (Lott, Reilly, & Howard, 1982; Reilly, Lott, & Gallogly, 1986). Nine items were added to the TSHI in the development of the SHAS that most enhanced scale reliability while broadening the assessment of respondents' understanding of harassment, specifically evaluating contemporary feminist conceptions of harassment. Mazer and Percival (1989) reported that high scores on the SHAS reflected more tolerant and accepting views of sexual harassment. 33 Formulation of the Total Attitude Sum Subjects were asked to indicate the extent of agreement with each of the nine items (a to i) on a 5-point Likert scale: strongly disagree (1), disagree (2), no opinion (3), agree (4), strongly agree (5). Item lb was reverse scored. The scores on the nine items were totalled to obtain a total attitude sum with a range of 9 - 45. High total attitude sums indicate more acceptance and tolerance of sexual harassment and less agreement with contemporary feminist views. Construct Validity Construct validity of the SHAS was demonstrated by comparing the SHAS with the "Macho" Scale (Villemez & Touhey, 1977). The "Macho" Scale measures individual differences in sex role stereotyping and sex discrimination with high scores reflecting more sexist attitudes. The "Macho" Scale has demonstrated significant relationships with other sex role inventories such as the Bern Sex-role Inventory (BSRI). Masculine-typed males according to the BSRI, had higher Macho scores than either androgynous- or feminine-typed males, or masculine typed females (Andersen as cited by Mazer & Percival, 1989). Subjects who got high scores on the "Macho" Scale also got high scores on the SHAS indicating that subjects with more sexist and discriminatory attitudes were also more tolerant and accepting of sexual harassment. Intuitively, it would be expected that those who have more sexist and discriminatory attitudes would also be more tolerant and accepting of sexual harassment as indicated by the study of Mazer and Percival (1989). Reliablilitv Internal consistency of the attitude scale in the survey using Cronbach's coefficient alpha was 0.79, somewhat lower than the SHAS (0.84) but higher than the THSI (0.75). 34 b. Knowledge about sexual harassment: Section2 Formulation of the Total Knowledge Sum. Respondents were asked to indicate the extent of agreement with each of the 30 items (a to s) on a 5-point Likert scale: strongly disagree (5), disagree (4), no opinion (3), agree (4), strongly agree (5). Items 3f, g, h, i, j , k, m, o, p, q, r were subdivided into two parts asking subjects to identify whether the statement was sexual harassment if the behavior came from (a) a supervisor or (b) a co-worker. Items 2e, 21, and 2n were not examples of sexual harassment and were included as dlstractors to avoid response set. These three items were not included in the total score. A total knowledge score was summed from 27 items with a range of 27 to 135. Content Validity According to Allen and Yen (1979) content validity is established through a rational analysis of the content of a test and through logical or sampling validity the researcher carefully defines the domain of behaviors to be measured and the logical design of items to cover all the important areas of the domain. Items in the knowledge section were based on four criteria: (1) the legal definition of sexual harassment according to the Equal Employment Opportunity Guidelines (EEOC, 1980); (2) statements from Section H of the USMSPB (1981) survey on sexual harassment; (3) typologies of sexual harassment (Gruber, 1992); and (4) a classification system of sexual harassment by Till (1980). Face validity of the survey was also determined by the use of two experts in the field of sexual harassment: a representative of the Sexual Harassment Office of the University of British Columbia and the Harassment Policy Coordinator of Simon Fraser University. In conjunction with their expert opinion, changes were made to the survey to enhance the content domain. 35 Construct Validity; Total knowledge sum versus total attitude sum A moderate negative correlation of total attitude scores with total knowledge scores (r=- 0.50, p_ < .01) indicated that there was construct validity in terms of these two dependent variables. As was expected, high total attitude scores (indicating more acceptance and tolerance of sexual harassment) were associated with low total knowledge (about sexual harassment) scores. I intuitively expected this relationship because it would be logical that those persons who do not know what constitutes sexual harassment would also be more likely to not recognize that certain behaviors are unacceptable. Less knowledgeable persons might be more likely to think of some harassment behaviors as just annoying or even normal and part of the status quo of relationships between men and women and therefore would appear more accepting and tolerant of sexual harassment. Conversely, those more knowledgeable about sexual harassment would likely more readily identify harassment behaviors when they occur and would know that they do not have to tolerate them. These persons would be more inclined to know their personal as well as legal rights and therefore would score lower on the attitude scale indicating they are less accepting and tolerant of sexual harassment. Reliability of Knowledge Section Cronbach' s coefficient alpha for the knowledge section was 0.94 indicating high internal consistency, c. Experience of being sexually harassed: Section 3 Items in this section were a replication of the knowledge items in section 2, but were phrased in terms of experience. The same three distractors from the knowledge section were carried over into Section 3 to maintain consistency and to prevent subjects from recognizing that the distractors were not forms of sexual harassment in the knowledge section. 36 Formulation of the Total Experience Sum Respondents were asked to indicate the extent of their experience with sexual harassment listed on a 5-point Likert scale: never (1), once (2), sometimes (3), often (4), and very often (5). There was a total of 20 items (a to s) with three distractors (3e, 31, 3n) which were not included in the total score. Item 3r requested subjects to respond to whether they had experienced the behavior coming from (a) a supervisor and (b) a co-worker. A total experience sum was calculated for each subject from 17 items. The range for the total experience score was 5-85. d. Sources of knowledge about sexual harassment: Section 4 In the initial development of this study, a total source of knowledge score was to be calculated in this section. It was found in analysis of the data that a total source of knowledge score was not meaningful as a single independent variable. I elected to treat each source of knowledge as an independent variable in the multiple regression. "Other" was dropped as a variable because of the low response level of the subjects to this item (see Appendix C for description of this item). This left a total of 6 items associated with sources of knowledge. Formulation of the Source of Knowledge Independent Variables Respondents were asked to indicate the relative amount of information about sexual harassment that they believed they acquired from six sources of knowledge: (1) newspapers, (2) radio, (3) television, (4) magazines, (5) friends, relatives, or co-workers, (6) information from union, (7) other. In the "other" category the subject was requested to write in the source of their knowledge. Statements were answered on a 5-point Likert scale from: none at all (1), very little (2), some (3), quite a lot (4), a great deal (5). The face value on each item represented the source of knowledge score for each independent variable (i. e. , 37 newspapers, radio, television, magazines, friends, union, and other). The range for each source of knowledge was 1 to 5. e. Number of television programs (sexual harassment) watched: Section 5 The "number of television programs watched" was an independent variable of the study and its main purpose was to delineate those who watched television programs regarding sexual harassment from non-viewers of such programs. Television programs listed on the survey were programs that the researcher had viewed as well as programs viewed by a peer in the Master's program of Counselling Psychology at the University of British Columbia. With the exception of the Clarence Thomas Senate Hearings, all programs listed had been seen on television at least once in the previous year before the survey mail out. It is not known how many of the programs were actually aired in reruns, but I saw four programs that were aired twice during the year. Specifically, "Designing Women" reran its sexual harassment episode in January, 1994 during the time of the second mail out of the survey. This was a program directly referring to the Senate confirmation hearings of Clarence Thomas. During the time of the third mail out of the survey (February, 1994) the "Sexual Harassment Commercial" was viewed once again by the researcher. Description of the television programs: "Herman's Head", "Designing Women", "Nurses", "Reasonable Doubts", "Golden Palace", "Murphy Brown", and "Grace Under Fire" were 30 minute light entertainment programs. "Catwalk" was a 60 minute light entertainment program. "Sex, Power & the Workplace" was a 60 minute educational program on the Public Broadcast station. "The Young and the Restless" was a soap opera and carried the topic of sexual harassment over several episodes. The Senate confirmation hearings of Judge Clarence Thomas was carried over several days. 38 At the time of the first mail out, it had been 2 years since the senate hearings had been aired on television, but it was included because of the intense notoriety that it received at the time. It was felt by the researcher that some people might still remember the details and the feelings associated with the programs. "60 Minutes" was a 15 minute episode discussing sexual harassment in a branch of the American secret service. "Business As Usual" was a television movie. The "Sexual Harassment Commercial" was a commercial shown on late night television with an approximate length of 30 to 60 seconds. Formulation of the "Number of television programs watched" Variable A total of 14 television programs that contained an episode regarding sexual harassment was listed. Subjects were asked to indicate "... if you watched the sexual harassment episode on any of the following programs. " The subjects were asked to respond to three possible categories: yes, no, I don't know. In addition, an open ended category was included which asked the subject to write in any other television programs that they had seen dealing with sexual harassment. The "yes" category was summed (including any programs that the subject wrote in). The range of scores was from 0 - 10. f. Perceived change of view about sexual harassment: Section 6 "Perceived change of view " was the third dependent variable of the study. This section was composed of 11 items the researcher posited that might reflect possible changes of view in regard to sexual harassment associated with watching television programs on sexual harassment. Formulation of the "Total perceived change of view mean " variable Respondents were asked to indicate the extent to which they agreed to each statement (a to k), answering on a 5-point Likert scale: strongly disagree (1), disagree (2), undecided (3), agree (4), and strongly agree (5). Item k was reverse 39 scored. The mean "total perceived change of view" score was calculated from 10 items (a to j). The range of scores was 10 to 50. Reliability A unidimensional scale contains items which measure the same underlying concept. According to de Vaus (1990), as a rule of thumb, items with an item to total item correlation of less than 0.3 are dropped from a scale to enhance unidimensionality. Item k, with an item-total correlation score of .0981 (indicating this item was measuring something different from the other items) was excluded from the other 10 items and was analyzed separately. Cronbach's coefficient alpha for the resulting 10 item scale was 0.84 indicating a moderate level of internal consistency. According to de Vaus (1990), the higher the 'alpha' the more reliable the scale and as a rule of thumb alpha should be at least 0.7 before we conclude that the scale is reliable. Validity The items in this section are based on what subjects perceived as a change in view about sexual harassment. In terms of face validity, the 10 remaining items appeared to measure subjects' views associated with television viewing of sexual harassment related programs. The researcher found no available inventory or scale that could be compared with the "change of view scale" to attempt to establish construct validity, g. Demographic Variables as Independent Variables 1. Age Question 7 of the survey was an open ended statement "Your age is ". Age in years was used as the variable for each subject. 2. Gender In question 8, subjects were asked to indicate between "male" and "female" and were coded male (1) female (2). 40 3. Education In question 10, subjects were asked to indicate their highest level of education completed. Education levels were in six categories and were coded: (1) less than high school, (2) high school graduation, (3) technical training or apprenticeship, (4) some college or university, (5) university graduate, and (6) advanced degree. 4. Ethnicity In question 12, subjects were asked in an open ended format, "With what ethnic group do you most closely identify?" Responses were to be coded into meaningful categories. This variable was dropped because the majority of respondents classified themselves as Western European and other categories associated with being Caucasian. The researcher felt that there was insufficient numbers of subjects in other ethnicities to form meaningful groups (see Appendix D for descriptive breakdown of responses). 5. Socioeconomic Status In question 13, subjects were asked , "What is your occupation or job?" Using the 1981 Socioeconomic Index for Occupations in Canada (Blishen, Carroll, & Moore, 1987), a socioeconomic index was calculated for each survey subject in reference to their job or occupation response. The 1981 Socioeconomic Index for Occupations in Canada was constructed on the basis of 1981 census data of 514 occupation classifications according to the Canadian Classifications and Dictionary of Occupations (Blishen et al. , 1987). The index is used when data are limited to occupation titles and is used for describing inequities in the division of labour (Blishen et al. , 1987). The index is calculated with equal weightings on income level and education level. The income level is the pooled median income for all paid labour force participants in each 41 occupation . The education level is based on the net proportion of well-educated persons in each occupation (Blishen et al. , 1987). The 1981 Socioeconomic Index for Occupations in Canada is meant to be used with the total Canadian labour force and its purpose is to measure the occupation and not the person. The index is a continuous, interval scale measuring to two decimal places ranging from 17.81 to 101.74 with a mean of 42.74 (Blishen et al., 1987). h. Demographic variables as descriptive information of the sample 1. Marital Status Marital status was assessed in 6 categories and coded in the following manner: (1) single, (2) married, (3) remarried, (4) divorced, (5) widowed, (6) separated, and (7) common-law. 2. Household Income Question 11 assessed household income in eight categories in increments of 10,000 and was coded in the following manner: (1) less than 20,000, (2) 20,000 -29,999, (3) 30,000- 39,999x (4) 40,000- 49,999, (5) 50,000- 59,999, (6) 60,000-69,999, (7) 70,000- 79,999, (8) more than 80,000. 3. Present Work Status Question 14 assessed current work status and assessed in 5 categories and coded in the following manner: (1) unemployed, (2) welfare, (3) casual, (4) part time, (5) full time. 4. Union Component Question 15 was an open ended question asking subjects to indicate to which union component they belong. This question was designed to aid in assessing the representativeness of the sample to the entire population of government workers. 42 E. DATA ANALYSIS Survey results were coded and scored by hand. Males were dummy coded as (1) and females as (2). Data was analyzed using SPSS-X statistical program. The analysis began with descriptive statistics outlining the sample in terms of the entire group as well as broken down into description by gender. Socioeconomic status was determined for each subject using the 1981 Socioeconomic Index for Occupations in Canada (Blishen et al. , 1987). Means and standard deviations were calculated for the three dependent variables (total attitude sum, total knowledge sum, and change of view) as well as for the independent variables (age, gender, experience sum, sources of knowledge, number of television episodes watched, and socioeconomic status). Number of subjects and percent of subjects in each category for education level, marital status, household income, work status, and union component were presented. Response rate of the survey was presented as well as a comparison in percentage of the sample distribution in terms of the union components in relation to the entire union population. 1-tests for independent means were calculated for male versus female in terms of mean total attitude scores, mean total knowledge scores, and mean change of view scores. T-tests for independent means were also calculated to test for significant differences between "TV viewers" versus "no TV" in terms of total attitude scores and total knowledge scores. Hotelling's X2 was performed to determine significant differences between males and females for mean scores on each item of the following scales: attitude, knowledge, and perceived change of view. 43 Reliability of the three dependent variable scales (attitude, knowledge, and change of view) was determined using Cronbach's coefficient alpha as a measure of internal consistency. Simultaneous multiple regression and multiple correlation were chosen to explore the primary research questions: (1) What is the relationship between the independent variables: number of television programs watched on sexual harassment, other sources of knowledge (newspapers, radio, television, magazines, friends, union, other), gender, age, ethnicity, socioeconomic status, education level, and experience of being sexually harassed with the dependent variables, knowledge and attitude about sexual harassment? (2) Is there a relationship of television viewing of sexual harassment programs with changes of view in regard to sexual harassment? (3) What independent variables from those of age, gender, education, ethnicity, socioeconomic status, experience of being sexually harassed, number of television programs about sexual harassment watched, and other sources of knowledge reliably predict the dependent variables: (a) level of knowledge about sexual harassment; (b) attitudes about sexual harassment; and (c) changes of view about sexual harassment? Two separate multiple regression analyses were performed. Multiple regression for the prediction of the dependent variables, attitude and knowledge were analyzed simultaneously on a sample size of 231. The multiple regression analysis for the change of view dependent variable was run with a sample size of 161 owing to the fact that 55 people did not watch television programs dealing with sexual harassment and therefore could not answer the "change of view section". An additional 20 subjects watched sexual harassment television programs but did not answer the "change of view" section. 44 Subsequent to these first analyses, separate multiple regressions were then run separating out for gender. Driven by the analyses, the researcher felt that additional information could be gained through gender separation of the regression analysis. Limitations to these analyses by gender are presented in the results section. Rationale for Choice of Simultaneous Multiple Regression According to Pedhazur (1973), multiple regression analysis is a method of studying individual and collective contributions of two or more independent variables on the variation of the dependent variable. It is applicable to designs in which the variables are continuous, categorical, or combinations of both. According to Wampold and Freund (1987), simultaneous regression is most often used in the context of prediction and would be appropriate when there is no basis for entering any particular independent variable prior to another independent variable. According to Tabachnick and Fidell (1989), simultaneous regression is generally the method of choice. Although stepwise regression is commonly used in the behavioral sciences, there are some serious criticisms of its use. According to Cohen and Cohen (1983), there can be very serious capitalization on chance when using large numbers of independent variables in a stepwise multiple regression. In stepwise regression, the significance test of an independent variable's contribution to multiple R squared proceeds in ignorance of the large number of other tests being performed at the same time for the other competing independent variables. Because of this, Cohen and Cohen (1983) conclude that neither the statistical significance tests for each variable nor the overall tests on the multiple R squared at each step are valid. Further to this, Cohen and Cohen (1983) suggest that stepwise regression should only be used when: (1) the research goal is primarily predictive; (2) the subject to independent variable ratio is at least 40 to 1; and (3) a cross-validation of the 45 stepwise analysis in a new sample is performed, only drawing conclusions that hold for both samples. In addition, Tabachnick and Fidell (1989) report that the equation derived from a sample using stepwise regression is too close to the sample and may not generalize well to the population i. e. , it may overfit the data. Because this study was strictly exploratory, with a limited subject to independent variable ratio, and with the concerns expressed in using stepwise regression, the researcher chose simultaneous or standard multiple regression. Treatment of Missing Data There were two options for treatment of missing data. The first was to drop those cases with missing data. The second option was to estimate missing data by using prior knowledge, inserting mean values, or using regression (Cohen & Cohen, 1983; Tabachnick & Fidell, 1989). Because missing values were scattered throughout cases and variables, deletion of cases would have meant a substantial loss in data and therefore the researcher elected to estimate missing data for the regression analyses. No missing data were replaced for descriptive statistics. For complete description of missing data, see Appendix E. For the multiple regression analyses, missing data for the attitude and knowledge sections were replaced with the subject's mean score for the section. According to Tabachnick and Fidell (1989), the mean is your best guess about the value of a variable. By inserting the mean score, the mean for the distribution as a whole does not change. However, it may reduce the variance of the variable and thus may also reduce the correlation of the variable with other variables because of the reduction in variance. Missing data for the experience of being sexually harassed (section 3); the sources of knowledge (section 4); and the change of view (section 6) were left missing. Missing data for demographic variables were left missing. 46 Driven by the data, experience of being sexually harassed (section 3) and sources of knowledge (section 5) were recoded to improve normality as indicated by the frequency histograms. In section 3 (experience of being sexually harassed), the data was recoded from never (1), once (2), sometimes (3), often (4), very often (5) to missing data and never (0), once (1), sometimes (2), often (3), and very often (4). Range of experience sum became 0 to 18. Section 5 (sources of knowledge) was recoded similarily from none at all (1), very little (2), some (3), quite a lot (4), and a great deal (5) to missing data and never (0), very little (1), some (2), quite a lot (3), and a great deal (4). Range of sources of knowledge became 0 to 4 for each source. An additional complication to the study was a typing error in section 6 (perceived change of view) for the first mail out of the survey. Response categories went from strongly disagree, disagree, undecided, agree, strongly disagree. The last response category was supposed to read strongly agree. Although section 6 began on page 4 of the study (with the correct response categories) and resumed on page 5 (with the incorrect category of strongly disagree instead of strongly agree) it was elected to not assume the respondent meant to answer strongly agree. Where the subject answered in the typing error category, it was elected to insert the subject's mean score to retain the subject for the regression analysis. Twenty subjects out of 144 were affected. No score was used when the subject answered in the typing error category for the descriptive statistics. 47 CHAPTER TV. RESULTS This chapter presents the results of the statistical analyses. It is divided into three sections. The first section begins with a brief discussion of the return rate of the survey, followed by a summary of demographic data concerning the respondents. The second section presents descriptive statistics for sections 1 through 6 of the survey. This section also includes t-tests for significant differences between women and men in terms of their scores on the three dependent variables, mean total attitude sum, mean total knowledge sum, and total perceived change of view mean. J-tests for significant differences between men and women are also reported for income, age, and experience of being sexually harassed. Additionally, i-tests comparing "no TV viewers" with "TV viewers" in terms of mean total attitude sum and mean total knowledge are reported. Chi-square analysis is reported for signficant differences in education level between men and women. Hotelling's J? is used to determine if there are significant differences between the mean scores of men and women for each of the three dependent variables, mean total attitude sum, mean total knowledge sum, and mean total perceived change of view and as well, significant differences between men and women are determined for the independent variables, sources of knowledge. Finally, the fourth section contains results of the multiple correlations and multiple regression analyses presented in terms of the research questions. A. Return Rate of the Sample A total of 500 surveys were mailed to members of a large union in British Columbia. As of October, 1993, the union had a membership of 30,839 comprised of 42% males and 58% females. In keeping with the male/female ratio of the union population, 210 surveys were sent to males and 290 surveys were sent to females. 48 Two hundred and forty seven subjects (49.4%) completed the survey with an additional 7 subjects (1.4%) who sent back the survey without completing it. In addition, there were 18 returns (3.6%) because of address change. Of those 18 returns, it is not known how many of these would have been in the first mail out. However, 14 of the 18 returned were received with the second mail out and only 4 were returned with the third mail out. In analyzing the return rates, it was noted that 4 subjects (who were listed by the union as being male) identified themselves as female. These four subjects were treated according to how they responded. There was also one subject that could not be identified by gender. Adjusting for the descrepancy in gender identification, there was a 52.7% (155/294) return rate for women; a 44.7% (92/206) return rate for men; and a 0.2% (1/500) "gender unknown" return rate (see Table F-l in Appendix F for a description of return rates by mail outs). B. Summary of Descriptive Statistics Age. Of the 247 completed surveys, 239 were suitable for analysis. The group analyzed consisted of 148 (61.9%) women and 91 (38.1%) men. The mean age of the total sample was 39.3 (SD = 9.0), with females having a mean age of 38.8 (SE) = 8.7) and males having a mean age of 40.8 (SD = 9.3). This age difference between male and female respondents of the survey was found to be significantly different, £ = 1.99, df = 237, p_ = .048. See Table 1 for categorization of age groups. Education Level Respondents were generally well educated. Only 4 subjects (1.7%) had less than high school education. Twenty-eight subjects (11.8%) had high school graduation; 28 subjects (11.8%) had technical training or apprenticeship; 103 (43.5%) had some college or university; 59 (24.9%) were university graduates; and 15 (6.3%) had advanced degrees. The most common education level was some 49 college or university (43.5% of entire group; 46.6% women; 38.5% men). Men and women's education levels were not significantly different, x 2 = 5.36, df = 5,p = .37. There were two subjects that did not indicate their education level. See Table 2 for breakdown of male and female education levels. Socioeconomic Status The socioeconomic status for the respondents was based on the socioeconomic index for occupations in Canada (Blishen et al. , 1981). The entire group had a mean occupational index of 48.10 (SD = 10.31). This was higher than the mean for the socioeconomic index for occupations in Canada which was 42.74 (SD = 13.28)(Blishen et al., 1981). Male respondents had a higher mean index (mean = 50.06, SD - 11.67) than female respondents (mean = 46.89, SD - 9.21). This difference was significant, t = 2.19, df = 155.96, p_ = .03. There were two subjects who did not indicate an occuption so their socioeconomic status could not be coded. See Table 3. Work Status The majority of subjects worked full time (88.2%). One person was unemployed, one person was on Long Term Disability, and no one was on social assistance. Five subjects (3.4%) worked as a casual (auxiliary), and 22 subjects (9.2%) worked part time. One female did not indicate work status. For male and female breakdowns, see Table 4. Marital Status Slightly more than half of the respondents were married (54% of entire group; 54.1% of women; 52.7% of men) with the next most prevalent status being "single". Thirty-nine (16.3%) were single; 17 (7.1%) were remarried; 19 (7.9%) were divorced; 3 (1.3%) were widowed; 9 (3.8%) were separated; and 23 (9.6%) were common-law. See Table 5 for further breakdown for males and females. 50 Table 1 Mean Age of Survey Respondents and Number of Respondents in Age Groups of Ten Year Increments Age (in years) Mean SD Range Entire group (n = 239) Women (n = 148) Men (n = 91) 39.28 38.38 40.75 8.99 8.70 9.30 20-64 20-60 21-64 Age by category: 20 -29 30 -39 40-49 50-59 60 -69 Entire group Women Men n % n % n % 36 91 78 32 2 15.1 38.1 32.6 13.4 0.8 25 62 44 16 1 16.9 41.9 29.7 10.8 0.7 11 29 34 16 1 12.1 31.9 37.3 17.6 1.1 Table 2 Education Levels of Survey Respondents Education levels Entir n 4 28 28 103 59 15 e group % 1.7 11.8 11.8 43.5 24.9 6.3 Wo: n 2 20 13 68 34 9 men % 1.4 13.7 8.9 46.6 233 6.2 Men n % 2 2.2 8 8.8 15 16.5 35 38.5 25 27.5 6 6.6 (1) Less than high school (2) High school graduation (3) Technical training or apprenticeship (4) Some college or university (5) University graduate (6) Advanced degree (Masters, PhD) Entire group (n. = 237) Women (n = 146) Men (n = 91) Note. Most common level of education for each group is "some college or university". 51 Table 3 Socioeconomic Status of Survey Respondents Mean SD Range Entire Group (n = 237) Women (n = 148) Men (n = 90) 48.10 46.89 50.06 10.31 9.21 11.67 25.56 - 79.23 25.56 - 75.60 25.56 - 79.23 Table 4 Current Work Status of Survey Respondents Entire Group Women Men Work Status n % n % n 9 Unemployed 1 0.4 1 0.7 Welfare - - . . . Casual Part time Full time Total 238 147 91 5 22 210 2.1 9.2 88.2 5 20 121 3.4 13.6 82.3 -2 2.2 89 97.8 52 Table 5 Marital Status of Survey Respondents by Category Categories Entire Group n 39 129 17 19 3 9 23 % 16.3 54.0 7.1 7.9 1.3 3.8 9.6 Women n 23 80 4 17 1 4 18 % 15.5 54.1 2.7 11.5 0.7 2.7 12.2 Men n % 16 17.6 48 52.7 13 14.3 2 2.2 2 2.2 5 5.5 5 5.5 (1) single (2) married (3) remarried (4) divorced (5) widowed (6) separated (7) common-law entire group (n = 239); women (n = 148); men (n = 91) Table 6 Household Income of Survey Respondents Income Levels Entire Group n % Women n % Men n % (1) Less than 20,000 (2) 20,000 to 29,999 (3) 30,000 to 39,999 (4) 40,000 to 49,999 (5) 50,000 to 59,999 (6) 60,000 to 69,999 (7) 70,000 to 79,999 (8) more than 80,000 entire group (n = women (n = 140) men (n = 86) 227) Mean 4.69 4.61 4.80 SD 1.92 2.01 1.77 2 28 51 27 40 31 24 24 0.9 12.3 225 11.9 17.6 13.7 10.6 10.6 2 22 31 14 21 21 14 16 1.4 15.6 22.0 9.9 14.9 14.9 9.9 11.3 -6 20 13 19 10 10 8 -7.0 22.0 14.3 20.9 11.0 11.0 8.8 Note. Mean salary is between 40,000 to 49,999 for all three groups. 53 Household Income The mean household income was between $40,000 and $49,999. Both females and males had mean household incomes between $40,000 to $49,999. Almost aquarter (22.5%) of the respondents had a household income between 30,000 to 39,999. Those earning less than 30,000 were disproportionally represented by women with 17%, in comparison to 7% men. Household income was not significantly different for men and women, t = 0.68, df, - 225, p_» .50. There were 12 people who did not indicate their income. See Table 6 for further breakdown of incomes. Ethnicity The majority of the respondents were Caucasian and represented 69% of the total who responded to the survey. There was also a sizeable group that did not identify their ethnicity (13.3%) or who misunderstood the question (5.2%). Because of insufficient sample sizes in ethnic groups identified other than Caucasian, ethnicity was dropped from further analysis. See Appendix D for detailed list of ethnicities. C. Descriptive Statistics of the Survey Respondents by Section There were three dependent variables (total attitude sum, total knowledge sum, and perceived change of view mean) calculated from the survey. Missing values were left missing for each of these three sections to accurately reflect the responses of the survey respondents. Total Attitude Sum The mean total attitude sum for the entire group was 19.47 (SD_ = 5.49). Total attitude sums ranged from 8 to 38. The mean total attitude sum for the entire group was below the midpoint of the range (23), indicating that overall, the respondents were not tolerant and accepting of sexual harassment. The total 54 attitude sum for women (from a sample size of 148) was 17.75 (SD = 4.67). Scores ranged from 9 to 33. Men (with a sample size of 91) had a total attitude sum of 22.27 (SJ) = 5.60). Male total atttitude sums ranged from 8 to 38. Total attitude sums were significantly different for male and female respondents, i = 6.45, dj = 165.05, p. < .0005, with males significantly higher than females indicating that male respondents were more accepting and tolerant of sexual harassment than female respondents. Hotelling's J_2 was performed to determine if there was any significant difference between men and women on items a to i of the attitude section. There was a significant difference found between the mean scores of men and women for the items of the attitude section, Hotelling's I 2 - 0.28, £(9, 219) - 6.75, p_ < .0005. These results also indicated, as in the T-test described above, that women were less accepting and tolerant of sexual harassment than men. As the number of univariate tests increases, the likelihood of finding significant differences capitalizing on chance may also increase. Bonferroni adjustment was made for the univariate E-tests indicating significance for each item (.05/9 « .006). See Table 7 for mean scores of men and women for each item, a to i as well as significance levels with Bonferroni adjustment. A t-test was also performed to compare television viewers with non television viewers in terms of mean total attitude sums. It was found that there was no significant difference between the two groups, t_= 1-25, df - 76.62, p « .216 indicating that respondents who had seen television programs about sexual harassment were no more or less accepting and tolerant of sexual harassment than those who did not see any of the programs. Total Knowledge Sum A mean total knowledge sum was calculated for the entire sample of 239. The mean total knowledge sum was 119.22 (SD_ - 14.09). Cases 167 and 192 with total 55 knowledge sums of 37 and 42 respectively, had scores greater than 5 standard deviations below the group mean. From the pattern of responses to the knowledge items, it appeared that the subjects may have purposely answered incorrectly. The knowledge scores were dropped from these two extreme cases (1 female and 1 male) and a mean total knowledge sum was recalculated. From a sample size of 237, the mean total knowledge sum was 119.89 (STJ * 12.05). Total knowledge sums for the entire group ranged from 75 to 135. Overall, the respondents were very knowledgeable about sexual harassment, far exceeding the midrange of 105. The mean total knowledge sum for women (n » 147) was 124.09 with a standard deviation of 9.05. Total knowledge sums ranged from 94 to 135. Men (n = 90) had a lower mean total knowledge sum of 114.51 with a standard deviation of 13.14 (range 75 to 135). Based on total knowledge sums after deletion of the two extreme cases, men and women's total knowledge sums were significantly different, i = 4.98, df = 154.99, p_ < .0005, indicating that female respondents were more knowledgable about sexual harassment than male respondents. Hotelling's T2 indicated that there were significant differences between men and women for the 27 items of the knowledge section, Hotelling's J 2 = .37, F_(27, 192) - 2.63, p_ < .0005. Women were significantly more knowledgeable than men on the items of the knowledge scale. For each knowledge item, Univariate £-tests following Bonferroni adjustment (.05/27 » .002) are indicated in Table 8. Also see Table 8 for mean knowledge scores for men and women for each item. Mean total knowledge sums for the group were also compared for respondents who watched television programs on sexual harassment with those who did not recall seeing these programs. There was no significant difference between the means of the two groups, I = .11, df =235, E - .915, indicating that 56 Table 7 Mean Attitude Scores of Women and Men Women Men Attitude Statements Mean SI) Mean _SD a. All this concern about sexual harassment makes it harder for men and women to have normal relationships 2.46 1.12 2.84 1.22 b. I believe that sexual intimidation is a serious social problem 3.85 1.07 3.59 1.12 c. It is only natural for a man to make sexual advances to a woman he finds attractive 1.87 0.83 2.30 1.06 d. Many charges of sexual harassment are frivolous or vindictive 2.23 0.96 2.82 0.95 e. A lot of what people call sexual harassment is just normal flirtation between men and women 2.28 0.95 2.87 1.00 f. Most women who are sexually insulted by a man provoke his behavior by the way they talk, act, or dress 1.75 0.81 2.22 0.93 g. An attractive woman has to expect sexual advances and should learn how to handle them 1.80 0.94 2.56 1.13 h. One of the problems with sexual harassment is that some women can't take a joke 1.90 0.97 2.44 1.08 i. It is only natural for a woman to use her sexuality as a way of getting ahead in school or at work 1.37 0.59 2.02 0.88 Note. The higher the score, the greater the agreement with the statement. Women (n ranged from 146 to 148); men (n ranged from 89 to 91). Attitude scores for men and women are significantly different for every item, rj < .006, except for item a (p. = .031) and item b (JJ = .042) Table 8 Mean knowledge scores of women and men 57 Knowledge Statements Women Men Mean SD Mean SD a. An employee hears co-workers (of the opposite sex) talking about h im/her in a highly sexualized manner in the immediate work area. b. The presence of pornographic pictures or posters, or sexual grafitti in the workplace. c. An empoyee discovers that he / she is the object of rumors or gossip among co-workers who have made remarks about his / her sexuality or appearance. d. An employee is sexually assaulted or raped at work. f. An employee is promised a promotion in exchange for sexual favors with: a) a supervisor b) a co-worker g. An employee is threatened with loss of job for refusing to have sex with: a) a supervisor b) a co-worker h. Important information or job training is withheld from an employee because of refusal to have sex with the employee's: a) supervisor b) co-worker i. An employee fears job consequences for refusing to date: a) a supervisor b) a co-worker j . An employee receives uninvited letters, telephone calls, or materials of a sexual nature from: a) a supervisor b) a co-worker k. Important information or job training is withheld from an employee because of refusal to date: a) a supervisor b) a co-worker m. An employee receives uninvited and deliberate touching, leaning over, cornering, or pinching by: a) a supervisor b) a co-worker o. An employee is repeatedly "badgered" or pressured for a date by the: a) employee's supervisor b) co-worker 4.24 0.80 3.84 0.96 4.19 0.91 3.59 1.06 3.84 0.99 3.56 1.05 4.78 0.60 4.58 0.87 4.89 0.42 4.69 0.68 4.85 0.49 4.60 0.79 4.92 0.40 4.78 0.55 4.89 0.42 4.70 0.69 4.92 0.40 4.76 0.59 4.90 0.41 4.72 0.65 4.49 0.90 3.97 1.17 4.42 0.95 3.88 1.22 4.86 0.45 4.52 0.74 4.85 0.46 4.49 0.78 4.79 0.56 4.49 0.84 4.78 0.57 4.46 0.88 4.85 0.46 4.53 0.72 4.86 0.45 4.95 0.72 4.58 0.73 4.12 1.05 4.54 0.76 3.99 1.10 Table 8 (continued) Mean knowledge scores of women and men 58 Knowledge Statements Women Men Mean SD Mean SD p. An employee receives sexually suggestive look, stares, or gestures from a: a) supervisor b) co-worker q. An employee receives unwanted small gifts over several months by a: a) supervisor b) co-worker r. An employee is given a poor reference or poor work evaluation because of refusal to: a) date a supervisor b) have sex with a supervisor s. Workers often tease one of their peers about h is /her sex life. 4.55 0.71 4.12 0.89 4.53 0.71 4.08 0.91 3.81 1.10 3.08 1.14 3.79 1.11 3.07 1.14 4.84 0.51 4.53 0.85 4.88 0.45 4.66 0.70 3.64 1.21 3.49 1.11 Note. The higher the score, the more knowledgeable about sexual harassment. Total n for statements a to s for women ranged from 145 to 148; total n for statements a to s for men was 89 to 91 . Knowledge scores for men and women are not significantly different, ,p_> .0018, for items: c, d, f(a), f(b), g(a), g(b), h(b), and s. 59 television viewers were no more or less knowledgable than those who did not recall seeing sexual harassment programs. Total Perceived Change of View Mean The total perceived change of view mean (items a to j of the survey, section 6) was calculated for the entire group from a sample size of 165. Only subjects who indicated that they had watched television programs dealing with sexual harassment answered this section. Fifty-five people (23.0%) of the 239 respondents, indicated that they did not watch television programs. However, 1 of the 55 did answer the section. Responses of this person were included in the reporting of responses by item for the survey. There were also 20 people who watched television programs but did not respond to section 6. This gave a total of 74 people who were not included in the calculation of the perceived change of view mean. In addition, there was also a typing error in section 6 for the first mail out. The final category strongly agree was mistakenly labelled as strongly disagree. Twenty of 141 respondents answered from 1 to 5 items in the incorrectly labelled category. These responses were left out for purposes of identifying reponses on an item by item basis for the section and were also left out for the calculation of the perceived change of view means for the entire group, women, and men. See Table H-l in Appendix H for break down of item responses assuming subjects did mean to respond strongly agree for survey respondents affected by the typing error. Item k was not included in the total perceived change of view mean because of its extremely low item to total item correlation (.10). Because it was analyzed by itself, it was no longer necessary to reverse score it. From 10 items (a to j , section 6 of the survey), a total perceived change of view mean was calculated. The entire group of 165 respondents had a total perceived change of view mean 60 of 2,96 (S_p_ = .62). Range of scores was 1.00 to 4.40. Women, from a sample size of 97, had a total perceived change of view mean of 3.08 (SD = .59). Scores ranged from 1.00 to 4.40. Men, with a sample size of 68, had a total perceived change of view mean of 2.79 (SD = .64). Scores ranged from 1.00 to 4.00. There was a significant difference between men and women's total perceived change of view means, t = 2.97, df = 163, p_ = .003, with more women than men perceiving a view change. Hotelling's T2 indicated that men and women were significantly different on the items of the perceived change of view section, Hotelling's T2 = .23, F( 10,126) = 2.95, p. = .002. Women were more likely to perceive a view change than men. Significance levels for each item following Bonferroni adjustment (.05/10 = .005) of the univariate F-tests are indicated in Table 9. For item k, "I no longer consider sexual harassment as seriously as I once did", the mean score for the entire group (n. = 154) was 1.81 with a standard deviation of 0.69. Women (a = 89) had a mean of 1.74 and a standard deviation of 0.68. Men (n = 65) had a mean score of 1.89 with a standard deviation of 0.69. Men and women were not significantly different on item k, i = 1.35, df = 152, p_ = . 179, indicating that on average, both groups disagreed that television programs had negatively affected their views about sexual harassment. Total Experience Sum The total experience sum (of being sexually harassed) is reported for recoded data. Coding was changed from never (1), once (2), sometimes (3), often (4), very often (5), to never and missing values (0), once (1), sometimes(2), often (3), and very often (4). It was found that the data was clearer and more understandable in this coding form. There were 11 respondents who left one item blank and 1 subject who left two items blank in this section. For these 61 Table 9 Mean Perceived Change of View Scores for Women and Men Women Men Change of View Statements Mean SD Mean SD a. It affected my views on sexual harassment. b. I did not know what sexual harassment was before I saw this program(s) but I have a better idea now. c. It reinforced my views on sexual harassment. d. I sought out more information about sexual harassment after seeing the program(s). e. It made me realize that I have experienced sexual harassment in my job. f. I would be more likely to report sexual harassment in the future if it happens to me. g. I learned that I wasn't alone; other women (men) have also experienced this problem. h. It made me realize how easy it is to discredit the victim. i. It made me realize how differently people interpret sexual harassment. j . I became more aware of sexual harassment. 3.03 1.07 2.85 1.06 2.05 0.93 1.88 0.93 3.66 0.83 3.18*1.03 2.26 0.89 2.06 0.83 2.20 1.08 2.06 0.92 3.62 0.90 2.88*0.98 3.01 0.95 2.61*1.02 3.64 0.91 3.29 1.01 3.94 0.71 3.76 0.95 3.48 0.97 3.29 1.03 Note. The higher the score, the more agreement with the statement. N for statements a to j (women) ranged from 89 to 97; n. for statements a to j (men) ranged from 63 to 67. * Change of view scores for men and women are significantly different, £ < .005, for these items, 62 subjects, total experience sums ranged from 0 (for 3 people); 2 (for 6 people); 5 (for one person); and 8 (for 2 people). The mean total experience sum for the entire group (n = 239) was 3.49 with a standard deviation of 4.0 and a range of 0 to 18. Women (n = 148) had a total experience sum of 2.72 with a standard deviation of 3.71. Total experience sums ranged from 0 to 17. Men (n = 91) had a higher mean total experience sum than women. Their mean was 4.74 with a standard deviation of 4.15. Men's scores ranged from 0 to 18. Men and women's mean total experience sums were significandy different, t - 3.89, df = 237, & < .0005. Of the entire group, 69 (28.9%) had a total experience sum of zero, indicating no experience of sexual harassment in the last 2 years. An additional 18 respondents (7.5%) had a total experience sum of 1, indicating that the subjects experienced some form of sexual harassment once in the past 2 years. Of the 239 respondents, 10% had total experience sums greater than two standard deviations from the mean. There were 24 respondents in the top 10 percentile with scores ranging from 9 to 18. Eleven men, representing 12.1% of the males who responded, and 13 women, representing 8.8% of the women who responded, comprised the people experiencing the most sexual harassment. See Table H-1 in Appendix H for descriptive statistics of top 10 percentile. The most common items that respondents experienced at least once were: "The presence of pornographic pictures or posters, or sexual grafitti in the workplace" with 94 (39.5%) experiencing this. Ninety respondents (37.7%) experienced, "Workers teased you about your sex life." Sixty-one (25.8%) respondents experienced, "Received sexually suggestive looks, stares, or gestures from a supervisor or coworker." In addition 58 (24.5%) experienced, "Co-workers talked about you in a highly sexualized manner in your immediate work area." In addition 46 (19.4%) experienced, "Was the object of rumors or 63 gossip among co-workers who made remarks about your sexuality or appearance" and 45 (18.8%) experienced, "Received uninvited and deliberate touching, leaning over, cornering, or pinching by a supervisor or co-worker." Of the more serious types of sexual harassment, less than 1% of the respondents experienced either: "Sexually assaulted or raped at work"; "Promised a promotion in exchange for sexual favors with a supervisor or co-worker"; "Threatened with loss of job for refusal to have sex with a supervisor or co-worker"; "Feared job consequences for refusing to date a supervisor or co -worker"; or "Important information or job training was withheld because of refusal to date a supervisor or co-worker." No respondent experienced, "Important information or job training was withheld because of refusal to have sex with a co-worker or supervisor." For a complete breakdown, see Tables H-2 and H-3 in Appendix H. Although men on average were significantly more tolerant and accepting of sexual harassment, the majority of male respondents generally were in agreement with female respondents with the exception of 2 statements. Collapsing the strongly disagree and disagree categories into a disagree category and similarly collapsing the agree and strongly agree categories into an agree category demonstrated that male respondents were fairly divided on two statements taken from the SHAS: "Many charges of sexual harassment are frivolous or vindictive" and "A lot of what people call sexual harassment is just normal flirtation between men and women." See Table N-l in Appendix N for breakdowns of each statement in collapsed categories. Sources of Knowledge There were seven sources of knowledge that respondents were asked to rate in terms of how much knowledge about sexual harassment they perceived that they gained from each source. The sources of knowledge were: newspapers, 64 radio, television, magazines, friends, information from the respondents' union, and other (in which the respondent was asked to specify what "other" was). There were insufficient numbers to form meaningful categories for "other" so it was dropped from the analysis. See Appendix C for details regarding other. The remaining six sources of knowledge were all independent variables in the study. Of these six sources of knowledge, television had the highest mean score for the entire group. The television mean score for the entire group (n = 236) was 3.15 (SJQ - .98). Women had a mean score of 3.21 (SD = .99) and men had a mean score of 3.06 (SD = .97). For this variable, 6.4% perceived that they got no knowledge about sexual harassment from television. Radio had the lowest entire group mean score which was 2.61 (SD =* .92) from a sample size of 234. The mean score for women was 2.67 with a standard deviation of .91 and the mean score for men was 2.52 with a standard deviation of .93. Of the entire group, 12.8% felt they got no knowledge from radio. For newspapers, the entire group (n = 237) had a mean score of 2.97 (S_D = .90). Women had a mean score of 2.99 (SJD = .92) and men had a mean score of 2.93 (SD = .88). Only 5.5% of the entire group perceived that they got no knowledge from newspapers. For magazines, the entire group (n = 231) had a mean score of 2.87 (SD = 1.03) with women having a mean score of 3.03 (SD = 1.01) and men having a mean of 2.61 (SE) = 1.03). Of the entire group, 12.1% perceived that they got no knowledge from magazines. Friends had an entire group (n. = 228) mean score of 2.86 (SD =.94) with women having a mean score of 2.93 (SJJ = .98) and men having a mean score of 2.75 (SD = -86). For this variable, 6.6% of the entire group perceived getting no knowledge from friends. 65 Last, information from the respondents' union had an entire group (n = 228) mean score of 2.83 (SD. = 1.10) with women having a mean score of 2.76 (SD = 1.08) and men having a higher mean score of 2.93 (SD = 1.13). There were 15.8% of the entire group that perceived that they got no knowledge about sexual harassment from information from the union. For further breakdown, see Tables 1-1 and 1-2 in Appendix I. Hotelling's T^ was performed to determine if there were significant differences between men and women for the sources of knowledge independent variables. There was no overall significant difference found between mean scores for men and women on the sources of knowledge, Hotelling's J_2 = 0.06, F(6,208) = 2.08,fi = .05 7. Number of Television Programs Watched Number of television programs watched about sexual harassment was an independent variable of the study. Survey respondents were asked to indicate the number of television programs dealing with sexual harassment that they had seen. Programs subjects identified from the list of television programs and any other programs identified by the respondent were totalled together to form a total score for number of television programs watched. There were 55 respondents who identified no television programs watched. Of these 55, one subject did respond to section 6, the perceived change of view question. Responses for the change of view section for this individual were included in the analysis for section 6 in terms of item by item responses. The mean number of programs watched for the entire group (a = 239) was 2.18 programs with a standard deviation of 1.95. Total number of programs watched ranged from 0 to 10. The mean number of programs watched by women was 2.16 with a standard deviation of 1.95. Total number of programs ranged from 0 to 7. For men, the mean number of programs watched was 2.19 with a 66 standard deviation of 2.19. Total number of programs watched by men ranged from 0 to 10. See Tables J-2 and J-3 in Appendix J for a complete breakdown of number of each episode watched and total number of programs watched. Union Component Representativeness of the survey respondents to the entire union population was assessed through analysis of respondents' union component as well as the male/female ratio of the respondents. There are seven components of the union that comprise the government membership of the entire union population. Percentages of the respondents in the survey sample compared with the total union population for the government sector showed that the sample was not representative of every area. Component 5, Retail Stores & Warehouse, was not sampled at all. This area accounts for 8.5% of government workers. The survey oversampled component 9 (Environment, Resources & Operations). Survey respondents for component 9 were 18.5% compared to the union population of 10.5%. Additionally, Component 7 (Education, Scientific, Technical & Administration) was sampled with a 5.0% rate in the survey respondents. This component is no longer classified as a government component of the union and is now listed as "private sector." Although the subjects chosen for the mail out were randomly selected, errors of inclusion contributed to the subjects not being selected from all of the appropriate components. Calculations were then adjusted to reflect the components actually surveyed. See Tables K-l and K-2 for initial breakdowns for union components. See table K-3 in Appendix K for comparisons of component percentages based on adjustments for the additional and missing components. D. Multiple Regression Multiple regression analyses were run simultaneously for the dependent variables, mean total attitude sum and mean total knowledge sum (a = 231). Total 67 perceived change of view mean was analyzed by itself because of the much smaller sample size (n = 161). This was due to the fact that 55 individuals did not watch television programs dealing with sexual harassment and an additional 20 people did not respond to section 6, the change of view section. Assumptions Prior to analysis, mean total attitude sum, mean total knowledge sum, total perceived change of view mean, age, sex, education level, socioeconomic status, total experience (of being sexually harassed) sum, sources of knowledge about sexual harassment (newspapers, radio, television, magazines, friends, information from union), and number of television programs watched were examined through SPSS programs for accuracy of data entry, missing values, and the fit between their distributions and the assumptions of multivariate analysis. On examination of the data, missing values for the attitude section and the knowledge section were replaced with the subject's mean score for the section (see Appendix E for description of all missing data) in order to retain as many subjects as possible for the regression analysis. Subject's mean score for the typing error responses for section 6, the perceived change of view section were inserted. Examination of histograms for each of the variables led to the deletion of knowledge scores for cases 167 and 192 which had extreme z scores greater than 6 standard deviations below the group mean. A frequency histogram run after deletions of these scores, showed a markedly improved normality for the mean total knowledge sum variable. This variable remained negatively skewed but was within acceptable ranges of normality. In addition, after recoding responses to the experience items and the sources of knowledge items, frequency histograms demonstrated improved normality. Responses for the 68 experience items, section 3, were recoded from never (1), once (2), sometimes (3), Often (4), very often (5) to missing data and never (0), once (1), sometimes (2), often (3) and very often (4). Responses for each of the sources of knowledge were recoded from none at all (1), very little (2), some (3), quite a lot (4), a great deal (5) to missing data and none at all (0), very little (1), some (2), quite a lot (3), and a great deal (4). Frequency histograms also indicated that in terms of the two dependent variables, mean total attitude sum and mean total knowledge sum, there were three independent variables with departures from normality that had the potential to be problematic. These three variables were sex (kurtosis of - 1.78), total experience sum (kurtosis of + 2.14), and socioeconomic status (-1.01). For the dependent variable, mean total perceived change of view mean, sex (kurtosis of - 1.89) and socioeconomic status (kurtosis of - 1.02) departed from normality. According to Tabachnick and Fidell (1989), although data transformations are recommended for remedying outliers and for failures of normality, linearity, and homoscedasticity, they are not universally recommended because of the difficulties in interpreting transformed scores. Residual scatterplots (predicted scores plotted against errors of prediction in standardized form) using SPSS*, were examined for the regression analysis for each of the three dependent variables-mean total attitude sum, mean total knowledge sum, and total perceived change of view mean. Scatterplots indicated that assumptions of normality, linearity, and homoscedasticity were met, thus no transformation of the SES variable, the sex variable or the total experience sum variable were undertaken. However, from a visual examination of the scatterplots, it was noted that multivariate outliers were present. There were 9 cases identified as having at least one variable as a possible univariate outlier with scores outside a three standard deviation range from the 69 mean for a particular variable of the study. Of these 9, only 2 cases (116 & 187) had a variable that exceeded the criterion standardized score of 3.67 (2-tailed), p_ = .001 (Tabachnick & Fidell, 1989). According to Tabachnick and Fidell (1989), cases with standardized scores in excess of 3 standard deviations are potential outliers but with samples with a large N, a few of these scores can be expected. Multivariate outliers were identified by Mahalanobis distance (p_ < .001) and also by casewise list of outliers. For the dependent variables, mean total attitude sum and mean total knowledge sum, 4 cases were identified as multivariate outliers. Case 187 was totally deleted from the regression analysis. This subject was a 51 year old male with a high mean total attitude sum of 30, a low mean total knowledge sum of 92, a high number of television programs watched (10), and a low SES of 31.95. The attitude score of Case 30 was also deleted from analysis. This case was originally adjusted for the attitude score, but was later reidentified as a multivariate outlier once case 187 was deleted. The subject was a 20 year old female with an SES of 36.89 (1 standard deviation below women respondents' mean). This subject had a high mean total attitude sum of 33, a mean total knowledge sum of 116 (slightly less than the mean score for the entire group), and a high total experience sum of 17. Case 116 was a 39 year old male with a high mean total attitude sum of 38, a low mean total knowledge sum of 75, and a high SES of 79.23. The attitude score was adjusted to the next lowest score and the knowledge score was adjusted to the next highest score of the entire group. The fourth multivariate outlier was case 118 which was a 59 year old male with a high mean total attitude sum of 31, a low mean total knowledge sum of 84, and a low SES of 26.36. The knowledge score of case 118 was adjusted to the next highest score which was 85. After adjustments made for outliers in terms of the dependent variables, mean total attitude sum and mean total knowledge sum, repeat frequency 70 histograms indicated an improvement in the socioceconomic status variable to within acceptable limits of normality, but the total experience sum slightly worsened in terms of kurtosis with this variable just slightly outside acceptable limits of the 2 standard deviation range. For the dependent variable, total perceived change of view mean, case 187 was the sole multivariate outlier identified. It was totally deleted from analysis. Results of analyses for the dependent variables of mean total attitude sum and mean total knowledge sum may not generalize to older males with high mean total attitude sums and low mean total knowledge sums. Results may also not generalize to subjects with low socioeconomic status who also have high mean total attitude sums and low mean total knowledge scores. Results of analysis for the dependent variable, total perceived change of view mean, may not generalize to older males of low socioeconomic status who have more acceptance and tolerance of sexual harassment and low knowledge about sexual harassment (even though they have watched many television programs about sexual harassment). Results of multiple regression for dependent variable: Mean total attitude sum A standard multiple regression was performed between mean total attitude sum as the dependent variable and age, sex, education level, socioeconomic status, total experience (of being sexually harassed) sum, sources of knowledge about sexual harassment (newspapers, radio, television, magazines, friends, information from union) and number of television programs watched as independent variables. Analysis was performed using SPSSX Regression with an assist from SPSSX Frequencies for evaluation of assumptions. After deletion of outliers, there was a total of 8 cases missing data, N = 231. The ratio of cases to independent variables was an adequate ratio (19 to 1). According to Tabachnick and Fidell (1989), using standard multiple regression, it is preferred to have at 71 least 20 times more cases than independent variables. No problems with multicollinearity or suppressor variables were found. Table 10 displays the correlations of the variables that were significantly related to the dependent variable (mean total attitude sum), the unstandardized regression coefficients (B) and intercept, the standardized regression coefficients (beta), the semipartial correlations (sx2) and & R2,and adjusted E2-Table L-l in Appendix L presents the entire correlation matrix for all of the independent variables included in the study with the dependent variables, total attitude sum and total knowledge sum. In terms of research questions asked, there were two questions related to the dependent variable, mean total attitude sum: (1) What is the relationship between the independent variables: number of television programs watched on sexual harassment, other sources of knowledge (newspapers, radio, television, magazines, friends, union, other), sex, age, ethnicity, socioeconomic status, education level, and experience of being sexually harassed with the dependent variables, knowledge and attitude about sexual harassment? and (2) Does any variable or combination of variables from the independent variables of age, sex, education, ethnicity, socioeconomic status, experience of being sexually harassed, number of television programs about sexual harassment watched, and other sources of knowledge, predict attitudes about sexual harassment? Results of the multiple correlations indicated that there were only 4 of the 12 independent variables that were significantly related to attitudes about sexual harassment. There was a moderate significant negative relationship (£ = -.42, p. < .01) of the independent variable sex, with the dependent variable, mean total attitude sum. In conjunction with the higher mean scores of men over that of women for the total attitude sum, this negative correlation indicated that male respondents were more tolerant and accepting of sexual harassment than 72 women. There was also a small, but significant negative correlation (r = -.17, p < .05) of television viewing of sexual harassment episodes with mean total attitude sum indicating a weak association between being less tolerant and accepting of sexual harassment with watching television programs about sexual harassment. In addition, there was a small but significant negative correlation (r = - .19, p_ < .01) of information about sexual harassment from friends with mean total attitude sum indicating that those who got their information from friends were associated with being less tolerant and less accepting of sexual harassment. The fourth variable that was significantly related with mean total attitude sum was age, which had a small positive correlation (r = .19, p < .01) indicating that as age of respondents increased, subjects were weakly associated with being more tolerant and accepting of sexual harassment. However, it is possible that the significant but small correlations for three of the independent variables, age, number of television episodes on sexual harassment watched, and information from friends may have occurred simply due to the large sample size. Interpretation of these results must be viewed with caution. A moderate multiple correlation of .52 was found between the mean total attitude sum and the independent variables. In terms of prediction, there were 4 independent variables in the standard multiple regression analysis that predicted attitudes about sexual harassment: sex, number of television episodes about sexual harassment watched, age, and information from friends. Of these four variables, sex contributed the most unique variance to attitudes about sexual harassment (see Table 10) with the other 3 variables contributing much less to the variance of the dependent variable. In view of these results it appears that sex would be the best predictor of attitudes about sexual harassment and that it would be predicted that men would be more tolerant and accepting of sexual harassment than women. In addition, getting information about sexual 73 Table 10 Standard Multiple Regression: Mean Total Attitude Sum Multiple R R Square Adjusted R Square Unique variabil i ty Shared variabil i ty .52 .27 .23 .17 .10 df Regression 12 Residual 218 Variables B 95% CI Beta Corr SigT srz( unique) TVProg Sex Age Reso5 Constant -.17 -4.14 .08 -.77 26.167 -.817 .696 .006 -1.427 -.105 -5.512 .158 -.116 -.16 -.37 .13 -.14 -.17 -.42 .19 -.19 -2.55 -5.95 2.12 -2.32 .0115 .0001 .0356 .0213 .02 .12 .01 .02 Note. Reso5 represents source of knowledge from friends. *_e < -ooi. 74 harassment from friends and watching television programs about sexual harassment would be predictive of being less accepting and tolerant of sexual harassment whereas, older respondents would be predictive of being more accepting and tolerant of sexual harassment. However, these results should be viewed with caution since these four independent variables represented only 23% (adjusted E square) of the total variance of the dependent variable, mean total attitude sum. There may be other variables not analyzed in this study that may be more predictive or enhance predictions of attitudes about sexual harassment. Results of Multiple Regression for Mean Tqtal Knowledge Sum A standard multiple regression was performed between mean total knowledge sum as the dependent variable and age, sex, education level, socioeconomic status, total experience (of being sexually harassed) sum, sources of knowledge about sexual harassment (newspapers, radio, television, magazines, friends, information from union) and number of television programs watched as independent variables. Table 11 displays the correlations of the variables that were significantly related with the dependent variable (mean total knowledge sum), the unstandardized regression coefficients (B) and intercept, the standardized regrssion coefficients (beta), the semipartial correlations (&?) and E, E^, and adjusted There were two research questions related to the dependent variable, mean total knowledge sum: (1) What is the relationship between the independent variables: number of television programs watched on sexual harassment, other sources of knowledge (newspapers, radio, television, magazines, friends, union, other), sex, age, ethnicity, socioeconomic status, education level, and experience ogf being sexually harassed with the dependent variables, knowledge and 75 Table 11 Standard Multiple Regression: Mean Total Knowledge Sum Multiple E R Square Adjusted R Square Unique variability Shared variability .46* .21 .17 .13 .08 Regression 12 Residuals 218 Variables 2 95% CI Beta Corr 1 SigT sr2( unique) Sex 8.26 Reso5 1.61 Constant 105.07 5577 11.677 .156 3.073 .36 .14 39 .15 557 2.18 .0000 .0302 .11 .02 Note. Reso5 represents source of knowledge from friends. * p < .0001 76 attitude about sexual harassment? and (2) Does any variable or combination of variables from the independent variables of age, sex, education, ethnicity, socioeconomic status, experience of being sexually harassed, number of televsion programs about sexual harassment watched, and other sources of knowledge, predict knowledge about sexual harassment? Results of the multiple correlations indicated that there were only 4 of the 12 independent variables that were significantiy related to knowledge about sexual harassment. There was a significant moderate positive relationship (r = .39, p < .01) of the independent variable sex with total attitude sum. Female respondents had significantly higher total attitude scores than men, thus indicating that women were moderately associated with being more knowledgable about sexual harassment than men. There was also a small significant negative correlation (£ = -.16, p < .05) of the independent variable total experience sum with total knowledge sum indicating that respondents with more experience of being sexually harassed were weakly associated with being less knowledgable about sexual harassment. In addition, there was a small significant negative correlation (£ = - .13, p_ < .05) of the independent variable, socioeconomic status with mean total knowledge sum indicating that there was a weak association of respondents from lower socioeconomic status with being more knowledgeable about sexual harassment or conversely, respondents with higher socioeconomic status being weakly associated with being less knowledgable about sexual harassment. There was also a small significant positive relationship (r = .15, p < .05) of information from friends with the dependent variable, mean total knowledge sum indicating that getting information about sexual harassment from friends was weakly associated with more knowledgable respondents. There was also a moderate significant negative correlation (r = -.48, p < .01) of mean total attitude sum with mean total knowledge sum which indicated that as 77 respondents' knowledge about sexual harassment increased, acceptance and tolerance of sexual harassment decreased. In opposite terms then, respondents who were less knowledgable about sexual harassment were moderately associated with being more accepting and tolerant of sexual harassment. It is important to note once again, that the small significant correlations of information from friends, socioeconomic status, and total experience sum with the dependent variable, mean total knowledge sum may not be of practical significance since the significance levels could have been due to the large sample size. Standard multiple regression analysis indicated that there were two variables predictive of mean total knowledge sum. Gender of the respondent and information from friends about sexual harassment were found to be the only significant predictors. Of these two variables, gender was found to be the strongest predictor, contributing nearly all of the unique variance (.11) compared to information from friends (.02). Here to, as in the prediction of the mean total attitude sum, the independent variables accounted for a small portion of the variance of the dependent variable, mean total knowledge sum. Although the regression analysis indicated that women would be more knowledgable about sexual harassment than men and that respondents who got their information from friends would be more knowledgable about sexual harassment, all the independent variables only accounted for 17% of the variance (adjusted R square) of the mean total knowledge sum variable. Thus, there are other variables not analyzed in this study that may be more predictive of knowledge about sexual harassment. Results of Regression for Total Perceived Change of View Mean A standard multiple regression analysis was performed between total perceived change of view mean as the dependent variable and age, sex, 78 education level, socioeconomic status, total experience (of being sexually harassed) sum, sources of knowledge about sexual harassment (newspapers, radio, television, magazines, friends, information from union) and number of television programs watched as independent variables. After deletion of one outlier, there was a total of 24 missing cases for a total of H = 161. The ratio of cases to independent variables was 13 to 1. Although Tabachnick and Fidell (1989) suggest when using standard multiple regression that the subject to independent variable ratio ideally should be at least 20 to 1, they also suggest that a bare minimum requirement is to have at least 5 times more cases than IVs. Table 12 displays the correlations of the variables that were significantly related with the dependent variable (total perceived change of view mean), the unstandardized regression coefficients (B) and intercept, the standardized regression coefficients (beta), the semipartial correlations (5£^) and j ^ R ^ and adjusted R.2. Table L-2 in Appendix L presents the entire correlation matrix for all of the independent variables included in the study with the dependent variable, total perceived change of view mean. There were two research questions related to the total perceived change of view mean: (1) Is there a relationship of television viewing of sexual harassment programs with changes of view in regard to sexual harassment? and (2) Does any variable or combination of variables from the independent variables of age, sex, education, ethnicity, socioeconomic status, experience of being sexually harassed, number of television programs about sexual harassment watched, and other sources of knowledge predict changes of view about sexual harassment? Results of the multiple correlation indicated that there were five variables that were significantly associated with total perceived change of view mean. There was a small significant positive correlation (z = .23, p_ < .01) of sex with 79 Table 12 Standard Multiple Regression: Perceived Change of View Multiple R R Square Adjusted R Square Unique variability Shared variability .47* .22 .16 .05 .17 df Regression 12 Residual 148 Variables B 95% CI Beta Corr J SigT sr2(unique) Reso4 .15 .051 .252 .25 33 2.98 .0034 .05 Constant 2.69 Note. Reso4 represents source of knowledge from magazines. *E<-0005 80 total perceived change of view mean. Total perceived change of view means were significantly higher for female respondents than male respondents. This result, in conjunction with the small positive correlation of sex with the total perceived change of view mean, was interpreted that there was a weak association of women perceiving that their views had changed about sexual harassment with watching television programs about sexual harassment. In addition, there was a small significant negative correlation (i = -.20, p_ < .05) of education level with total perceived change of view mean indicating that as the more educated the respondents, the less they would associate changing their views with watching television programs about sexual harassment. There was also a small significant negative correlation (r = -.26, p. < .01) of socioeconomic status with total perceived change of view mean indicating that those respondents with lower socioeconomic status were weakly correlated with changing their views about sexual harassment associated with watching television programs about sexual harassment. In addition, there was a small significant positive correlation (r. = .18, p. < .05) of information from television and a small significant positive correlation (£= .33, p_ < .01) of information from magazines, with the dependent variable, total perceived change of view mean. These small but significant correlations indicated that more knowledgeable respondents were weakly associated with perceiving their knowledge about sexual harassment coming from television and magazines. Standard multiple regression analysis indicated that there was only one significant variable, information from magazines, that was predictive of changing views about sexual harassment associated with watching television programs about sexual harassment. The sex variable approached significance (t = 2.98, p_ = .06). Although information from magazines was significant, it only contributed a small unique variance to the dependent variable (.05) whereas the 81 independent variables as a group shared most of the variance contributed to the dependent variable (.17). Although there was a significant moderate multiple R. of .47, (F = 3.49, p < .0005) the independent variables contributed only 16% (adjusted £ square) to the variance of the dependent variable, total perceived change of view mean. Other variables not analyzed in this study may therefore be more predictive of changing views about sexual harassment or may enhance prediction. When correlations of the independent variables with the dependent variables were examined, it was found that some of the correlations were much larger or in the opposite direction depending on the gender (see Table M-l in Appendix M). Because of this, it was decided to run separate regression analyses for men and women despite the much lower ratio of subjects to independent variables. Regression analyses for women Multiple standard regression analyses were run simultaneously for the dependent variables, mean total attitude sum and mean total knowledge sum (n = 143) with a ratio of subjects to independent variables of 13 to 1. Scatterplots of standardized predicted scores against standardized residuals indicated that the variables in the regression analysis met the assumptions for multivariate analysis. Outliers for the entire group identified for the dependent variables, mean total attitude sum and mean total knowledge sum, were the same for women run separately in the regression with the exception of case 118 which was no longer identified as a multivariate outlier. The regression analyses were run with case 187 deleted; the knowledge scores of 167, 192, and 116 deleted; and the attitude score of case 30 was deleted. Regression analyses: Mean total attitude sum and mean total knowledge sum 82 There was only 1 of 11 independent variables that was significantly correlated with the mean total attitude sum. There was a small significant negative correlation (r = -.18, p_ < .05) of number of television programs watched with mean total attitude sum indicating that there was a weak relationship with women being less accepting and tolerant of sexual harassment associated with watching more television programs about sexual harassment. However, it was found that all the regression coefficients did not differ significantly from zero, (Multiple R = .30, £ = 1.19, p_ = .2990) indicating that none of the independent variables could predict women's attitudes about sexual harassment. See Table 13. There were 3 of 11 independent variables that were signicantly correlated with the dependent variable, mean total knowledge sum. There was a small positive correlation (r = .18, p. < .05) of age with mean total total attitude sum indicating that there was a weak relationship with older women being more knowledgeable about sexual harassment. There was also a small significant positive correlation (r = .17, p. < .05) of information from newspapers and a small significant negative correlation (i = -.16, p < .05) of information from the union with mean total knowledge sum. These correlations indicated that there was a weak association with women more knowledgeable about sexual harassment with getting their information from newspapers contrasting with women who were less knowledgeable about sexual harssment being weakly associated with getting their information about sexual harassment from their union. Regression analysis for the dependent variable, mean total knowledge sum indicated that there were two independent variables predictive of how knowledgable women were about sexual harassment, age (t_= 2.50, p_ * .0136) and information from the union (£ = -2.62, p_ - .01). However, adjusted R square indicated that the variables contributed less than 10% of the variance of the dependent variable. This would indicate that even though statistical 83 Table 13 Standard Multiple Regression for Females: Mean Total Attitude Sum Multiple R R Square Adjusted R Square Standard Error .30 .09 .01 4.52 Regression 11 Residuals 131 F=1.19 p = .2990 Table 14 Standard Multiple Regression for Females: Mean Total Knowledge Sum Multiple R R Square Adjusted R Square Standard Error Unique variability Shared variability .39 .15 .08 8.77 .08 .07 tf Regression 11 Residuals 131 F-2.08 p = .0261 Variables B 95% CI Beta Corr T SigT sr2( unique) Age Reso6 Intercept Note. Reso6 24 .049 .423 .22 .18 -1.79 -3.15 -.436 -.22 -.17 110.87 represents source of knowledge from the union. 6.25 6.83 .0136 .0100 .04 .04 84 significance was found, the practical significance of these results would be in question, therefore no interpretation was made for predicting women's knowledge about sexual harassment. See Table 14. Regression analysis for perceived change of view There were 4 of 11 independent variables that were significantly correlated with the dependent variable, total perceived change of view. There was a small significant negative correlation (r = -.36, p_ < .01) of socioeconomic status with perceived change of view indicating that female respondents with lower socioeconomic status were weakly associated with perceiving themselves changing their views about sexual harassment associated with watching television programs about sexual harassment. There was also a small significant positive correlation (r » .24, p_ < .05) of number of television programs watched with perceived change of view, a small significant positive correlation (r - .27, p. < .01) of information from magazines; and a small significant negative correlation (r = -.25, p_ < .05) of education level with perceived change of view mean. Findings suggest that there was a weak association with less educated women perceiving their views had changed associated with viewing television programs about sexual harassment. In addition, getting information about sexual harassment from magazines was weakly associated with women perceiving their views had changed. Finally, women perceiving that their views about sexual harassment had changed was weakly related to watching more television programs about sexual harassment. Although these variables were significantly correlated with perceived change of view, only one independent variable in the regression analysis was found to be significant as a predictor of perceived change of view. Socioeconomic status was found to be predictive of women changing their views about sexual harassment associated with watching television programs about 85 Table 15 Standard Multiple Regression for Females: Total Perceived Change of View Mean Multiple R R Square Adjusted R. Square Standard Error Unique variability Shared variability .51 .26 .16 .55 .07 .19 Regression Residuals F=2.61 D = & 11 83 .0066 Variables B 95% CI Beta Corr T Sigl sr2( unique) SES -.02 -.037 -.006 -.33 -.35 7.82 .0064 .07 Intercept 110.87 86 sexual harassment. Results indicated that women of lower socioeconomic status would be more likely to perceive their views changing about sexual harassment associated with watching television programs about it. See Table 15. Regression analyses for men Multiple standard regression analyses were run simultaneously for the dependent variables, mean total attitude sum and mean total knowledge sum (n = 87) with a ratio of subjects to independent variables of 8 to 1. Scatterplots of standardized predicted scores plotted against standardized residuals indicated that the variables in the regression analysis met the assumptions for multivariate analysis. The regression analyses were run with outliers removed for case 187 and the knowledge scores of 167, 192, and 116 deleted. Regression analysis for mean total attitude sum There was only 1 of 11 independent variables that was significantly correlated with the dependent variable, mean total attitude sum. There was a small significant negative correlation (£ = -.35, p_ < .01) of information from friends with mean total attitude sum indicating that males with less tolerance and acceptance of sexual harassment were weakly associated with perceiving that they got their information about sexual harassment from their friends. However, regression analysis results (Table 16) indicated that there were 3 variables that were predictive of male mean total attitude sum. Although information from friends was the only significant correlation with mean total attitude sum, it contributed less unique variance (5%) to the dependent variable in comparison to the other significant predictors, age and total experience sum, with each contributing 8% of unique variance to mean total attitude sum. Results indicated that older males would be more tolerant and accepting of sexual harassment. In addition, men experiencing more sexual harassment would be more tolerant and accepting of sexual harassment. On the other hand, 87 Table 16 Standard Multiple Regression for Males: MeanTotal Attitude Sum Multiple R R Square Adjusted R Square Standard Error Unique variability Shared variability .55 .30 .20 4.75 .21 .09 Regression Residuals F=2.98 D = df 11 75 .0025 Variables B 95% CI Beta Corr J SigT sr2( unique) Age Reso5 Expsum Intercept .18 .054 .303 -1.48 -2.74 -.215 .43 .148 .718 14.08 .31 .22 2.87 .0054 .08 -.25 -.33 -2.33 .0225 .05 .34 .22 3.02 .0034 .08 Note. Reso5 represents source of knowledge from friends. Table 17 Standard Multiple Regression for Males: Mean Total Knowledge Sum Multiple R R Square Adjusted R Square Standard Error .42 .17 .05 12.06 tf Regression 11 Residuals 75 F=1.44 p = .1740 88 male respondents getting their information about sexual harassment from their friends was predictive of men who were less tolerant and accepting of sexual harassment. Regression analysis for mean total knowledge sum There were no independent variables that were significantly correlated with mean total knowledge sum for male respondents. In addition, regression analysis indicated that there were no variables that were predictive of male respondents' knowledge scores at the significance level of .05 for alpha. See Table 17. Regression analysis for total perceived change of view mean Multiple standard regression analysis was run separately for males (n = 66) for perceived change of view mean with a 6 to 1 ratio of subjects to independent variables. Scatterplots of standardized predicted scores plotted against standardized residuals indicated that the variables in the regression analysis met the assumptions for multivariate analysis. The regression analysis was run with one multivariate outlier removed, case 187 (previously described). Significant correlation coefficients were found for 2 of 11 independent variables. There was a small significant positive correlation (r = .28, p < .05) of information from the union and a slightly stronger significant positive correlation (i = .34, p < .01) of information from magazines with the dependent variable, total perceived change of view mean. Correlations indicated that there was a weak association of both information from the union and information from magazines with male respondents perceiving that their views had changed about sexual harassment associated with watching television programs about sexual harassment. In terms of prediction however (see Table 18), only one variable, information from magazines was found to be predictive of male respondents' perceived change of view. Information from magazines was a 89 Table 18 Multiple Standard Regression for Males: Total Perceived Change of View Mean Multiple E R Square Adjusted R Square Standard Error Unique variability Shared variability .57 .32 .18 .58 .13 .19 Regression Residuals F=2.33 p_ = & 11 54 .0198 Variables B 95% CI Beta Corr T SigT sr2( unique) Reso4 Intercept .28 1.73 .106 .451 .43 3 4 3.24 .0021 .13 Note. Reso4 represents source of knowledge from magazines. 90 strong predictor, contributing 13% of unique variance to the dependent variable. Results indicated that information from magazines about sexual harassment was predictive of male respondents who perceived that their views had changed about sexual harassment associated with watching television programs about sexual harassment. E. Summary of Results Survey respondents were generally full time employees, well educated, and Caucasian. Both males and females had on average, household incomes between $40,000 to $49,999. Male and female respondents were significantly different on most variables of the study with the exception of education and household income. Male respondents were found to be more tolerant and accepting of sexual harassment than female respondents. In conjunction with this finding, male respondents were also found to be less knowledgeable about sexual harasssment than female respondents. In terms of the third dependent variable, perceived change of view mean, women tended to agree more often with perceiving that some of their views about sexual harassment had changed associated with watching television programs about sexual harassment. Unexpectedly, in terms of experience of being sexually harassed, male respondents had higher total experience scores than women. Attitudes About Sexual Harassment For the research question, "What is the relationship between the independent variables: number of television programs watched on sexual harassment, other sources of knowledge (newspapers, radio, television, magazines, friends, union, other), sex, age, ethnicity, socioeconomic status, education level, and experience of being sexually harassed with the dependent variable, attitude about sexual harassment?", it was found that there were weak relationships for 4 of the independent variables with the dependent variable, 91 attitude about sexual harassment. These weak, but significant independent variables were sex (gender), age, number of television programs watched about sexual harassment and getting information from friends about sexual harassment. For the entire group, multiple correlations indicated that males were moderately associated with being more accepting and tolerant of sexual harassment than women. There were also weak correlations of being less tolerant and accepting of sexual harassment with watching television programs about sexual harassment and with getting information about sexual harassment from friends. In contrast, there was a weak correlation with being more tolerant and accepting of sexual harassment with increasing age of the respondents. For the research question, "Does any variable or combination of variables from the independent variables of age, sex, education, ethnicity, socioeconomic status, experience of being sexually harassed, number of television programs about sexual harassment watched, and other sources of knowledge predict attitudes about sexual harassment?", there were 4 variables identified by standard multiple regression analyses that were predictive for the entire group of survey respondents' attitudes about sexual harassment. These predictor variables were sex (gender), number of television episodes about sexual harassment watched, age, and information from friends. Of these four variables, gender contributed the most unique variance to the dependent variable. Interpreting these findings, it is suggestive that men would be more tolerant and accepting of sexual harassment than female respondents; older respondents would be more accepting and tolerant of sexual harassment; respondents who watched television programs about sexual harassment would be less accepting and tolerant of sexual harassment; and repondents who got information about 92 sexual harassment from their friends would be less accepting and tolerant of sexual harassment. Male and female respondents were also analysed separately for predicting attitudes about sexual harassment. Although women who were less tolerant and accepting of sexual harassment were weakly correlated with watching television programs about sexual harassment, there was no variable in the regression analysis found to be predictive of women's attitudes about sexual harassment. In contrast, there were three variables predictive of male respondents' attitudes about sexual harassment. Regression analysis indicated that older males and males who experienced higher levels of sexual harassment were predictive of male respondents being more accepting and tolerant of sexual harassment. In contrast, male respondents who perceived that they got information about sexual harassment from their friends were less tolerant and accepting of sexual harassment. Knowledge of Sexual Harassment For the research question, "What is the relationship between the independent variables: number of television programs watched on sexual harassment, other sources of knowledge (newspapers, radio, television, magazines, friends, union, other), sex, age, ethnicity, socioeconomic status, education level, and experience of being sexually harassed with the dependent variable, knowedge about sexual harassment?", there were 4 independent variables weakly associated with knowledge about sexual harassment. These variables were sex (gender), socioeconomic status, information from friends, and experience of being sexually harassed. Multiple correlations indicated that women were moderately associated with being more knowledgeable about sexual harassment than men. In addition, information from friends and respondents with lower socioeconomic status 93 were weakly associated with being more knowledgeable about sexual harassment. Conversly, respondents with more experience of being sexually harassed were weakly associated with being less knowledgeable about sexual harassment. For the research question, "Does any variable or combination of variables from the independent variables of age, sex, education, ethnicity, socioececonomic status, experience of being sexually harassed, number of television programs about sexual harassment watched, and other sources of knowledge predict knowledge about sexual harassment?", there were 2 independent variables found to be predictive of knowledge about sexual harassment for the entire group of survey respondents. Standard multiple regression analysis for male and female respondents combined, indicated that there were two variables predictive of knowledge about sexual harassment. Gender was the strongest predictor contributing nearly all of the unique variance to the dependent variable. The regression analysis indicated that women would be more knowledgeable about sexual harassment than men. Information from friends about sexual harassment was also predictive of being more knowledgeable about sexual harassment but was a much weaker predictor, contributing much less to the variance of the knowledge variable. Separate regression analyses for male and female respondents indicated that there were no variables significantly correlated with male respondents' knowledge about sexual harassment and no variables were predictive of their knowledge about sexual harassment. For women, there were weak associations of women being more knowledgeable about sexual harassment with being older and with getting information about sexual harassment from newspapers. In contrast, there was a weak association of women less knowledgable about sexual harassment with getting their information from their union. Of the three 94 variables significantly correlated with women's knowledge about sexual harassment, only age and information from the union were found to be predictive of women's knowledge about sexual harassment. However, these two variables, although statistically significant, the practical significance is in question since the independent variables contributed less than 10% of the variance to the knowledge variable. Perceived Change of View For the research question, "Is there a relationship of television viewing of sexual harassment programs with changes of view in regard to sexual harassment?", it was found that there were 5 independent variables weakly correlated with perceiving a change of view. Multiple correlational analysis for male and female respondents combined, indicated that the 5 significant independent variables were sex, education, socioeconomic status, information from television, and information from magazines. There was a weak association of women perceiving that their views had changed associated with watching television programs about sexual harassment. As well, respondents with lower socioeconomic status were weakly associated with a perceived change of view. Information from television and information from magazines was also weakly associated with perceived change of view. In contrast, the more educated the respondents, the less respondents perceived their views about sexual harassment had changed associated with watching television programs about sexual harassment. For the research question, "Does any variable or combination of variables from the independent variables of age, sex, education, ethnicity, socioeconomic status, experience of being sexually harassed, number of television programs about sexual harassment watched, and other sources of knowledge predict changes of view about sexual harassment?", it was found that only 1 95 independent variable, information from magazines, was predictive of perceived change of view. Multiple correlational analysis done separately by gender indicated that for female respondents, lesser educated women; women with lower socioeconomic status; women who got information about sexual harassment from magazines; and women who watched more television programs about sexual harassment were weakly correlated with perceiving their views had changed about sexual harassment associated with watching television programs about sexual harassment. However, only socioeconomic status was predictive of perceived change of view. Results indicated that women of lower socioeconomic status would be more likely to perceive their views changing about sexual harassment associated with watching television programs about sexual harassment. In contrast, male respondents who perceived their views had changed were weakly correlated with getting information from their union and information from magazines. However, getting information from magazines was the only variable predictive of male respondents perceiving their views had changed associated with watching television programs about sexual harassment. In conclusion, there were weak relationships of some of the independent variables with the dependent variables of the study. Some of the independent variables could predict attitudes and knowledge about sexual harassment and could also predict perceiving a view change associated with watching television programs about sexual harassment. For each reqression analysis, a relatively small amount of the variance was contributed by the predictor variables to the dependent variables. It is possible that there are other variables not examined in this study that may enhance predictions of attitudes, knowledge, and perceived change of view about sexual harassment. % CHAPTER V. DISCUSSION This chapter begins with a reiteration of the purpose of the study and of the research questions. Following this is a discussion of the return rate of the survey and the representativeness of the sample. Findings of the study are then integrated with current literature from media and sexual harassment. Limitations of the study, impUcations for future research, and implications for counselling then follow. Conclusions from the study end the chapter. Purpose of the Study The purpose of the study was to explore whether television could be used as a meaningful way of raising peoples' awareness and knowledge about a social issue, namely, sexual harassment. Additionally, whether people perceive that television programs contribute to attitude changes about sexual harassment was also explored. Restatement of the Research Questions 1. What is the relationship between the independent variables: number of television programs watched on sexual harassment, other sources of knowledge (newspapers, radio, television, magazines, friends, union, other), sex, age, ethnicity, socioeconomic status, education level, and experience of being sexually harassed with the dependent variables, knowledge and attitude about sexual harassment? 2. Is there a relationship of television viewing of sexual harassment programs with changes of view in regard to sexual harassment? 3. Does any variable or combination of variables from the independent variables of age, sex, education, ethnicity, socioeconomic status, experience of being harassed, number of television programs about sexual harassment watched, and other sources of knowledge predict level of knowledge about sexual harassment? 97 4. Does any variable or combination of variables from the independent variables of age, sex, education, ethnicity, socioeconomic status, experience of being sexually harassed, number of television programs about sexual harassment watched, and other sources of knowledge predict attitudes about sexual harassment? 5. Does any variable or combination of variables from the independent variables of age, sex, education, ethnicity, socioeconomic status, experience of being sexually harassed, number of television programs about sexual harassment watched, and other sources of knowledge predict changes of view about sexual harassment? Discussion of the Survey Return Rate According to Babbie (1973), there is a wide range of response rates reported from survey literature. This author considers a response rate of at least 50% is adequate for analysis and reporting; a 60% response rate is good; and a 70% response rate is very good. These percentages are based on subtracting surveys that could not be delivered because of change of address and also subtracting surveys that were returned in refusal to participate. The survey response rate for this study on sexual harassment in the workplace is adequate for analysis. There were 7 surveys sent back uncompleted and 247 completed surveys for a total of 254 or a 50.8% return rate. Surveys sent back because of change of address were not subtracted because information about returns for the first mailing was not available. However, more recently, guidelines for response rates have been suggested after reviewing survey studies from the 1980s, published in the Journal of Counseling Psychology (Weathers, Furlong, & Solorzano, 1993). Median response rates of the 34 studies responding to this enquiry were 63.7% with a range of 27.0% to 91.7%. These authors suggest that response rates should be calculated in 98 terms of useable surveys. In these terms, this study had a useable return rate of 239 surveys or 47.8%. Weathers et al. (1993) consider "useable survey" response rates of less than 50% as being not adequate although 43.5% of surveys they studied had return rates of less than 50%. These authors noted that there have been studies with less than 40% return rates which indicated that there is some variability in the criteria used by reviewers to evaluate mail survey studies. According to Babbie (1973), overall response rate is one guide to the representativeness of the sample respondents. With high response rates, there is less likelihood of a significant response bias occurring than with low response rates from a random sampling. According to de Vaus (1990), non-responders tend to be different from responders. They tend to be older, less educated, and tend to have non-English-speaking backgrounds. Thus low response rates introduce bias into the sample. Although the response rate of this study was just adequate for analysis, a much higher rate would have been better to ensure less bias in the sample. Representativeness of the Sample Representativeness of the respondent sample was assessed by two criteria: the male/female ratio of the sample compared to the union population and the percentage of respondents in the union components to those of the union population. As of October, 1993, the union population had a ratio of 42% males to 58% females. Return rates for the respondent sample was 43.8% males to 53.4% females. Adjusting for the descrepancy in finding four subjects identifying themselves as females (when they were identified by the union as males) gave an adjusted ratio of 44.7% males to 52.7% females with one subject not identified for gender. These results indicated that the respondent sample was fairly representative of the union population in terms of gender. 99 In terms of union component, it was found that there was an inclusion error in sampling. One union component was inadvertently included and one component was excluded from sampling (see Table K-2 in Appendix K). Adjusting for this error (see Table K-3 in Appendix K), components 2, 7, and 9 were not consistent with the union population percentages. However, components 1, 6, 11, and 12 were consistent with the union population figures. The respondent sample had a 73.9% total proportion compared to a 70% total proportion of the union population for these four components indicating that overall, the respondent sample was fairly representative of the union population as a whole. Experience of Being Sexually Harassed Unexpectedly, male respondents reported higher mean total experience scores of being sexually harassed, (mean = 4.74, SD = 4.15), than female respondents (mean = 2.72,S£) = 3.71). In addition, of the 91 male survey respondents, only 11% had never experienced in the last two years some type of harassment in comparison to 39.9 % of the 148 female respondents. On reviewing the literature, women have been overwhelmingly identified as being more sexually harassed than men in terms of work (Konrad & Gutek, 1986; Pryor, 1987; Sneider, 1982; USMSPB, 1981, 1988) and in academia (Fitzgerald et al., 1988; Lott, Reilly, & Howard, 1982; Reilly, Lott, & Gallogly, 1986). It is not clear why males in this survey reported higher levels of sexual harassment than women, a result which was a major departure from the sexual harassment literature. However, Gruber (1990) suggests that straightforward comparisons of harassment proportions cannot be made because of substantial differences in research methodology such as sample size, survey response rate, sample diversity, harassment frame of reference, number of harassment categories, and types of words or phrases used to eliclit responses. In analyzing 100 the methodologies and results of 18 harassment surveys, Gruber (1990) suggested that surveys with a low response rate may inflate the report of harassment. He also reported that studies with more categories of sexual harassment had higher incidence of harassment than those which either presented few explicit categories or else relied on answers to open-ended questions. In terms of these methodological aspects, it is possible that this survey may have overrepresented the incidence of sexual harassment (71.1 % had one experience or more in the past 2 years related to work) because it did have a large number of categories and it had a low reponse rate (useable surveys was 47.8%). However, it is not clear why male respondents had a higher incidence of sexual harassment than female respondents, but a review of a selected body of the sexual harassment literature does shed some light on this departure. According to a survey by Gutek, Nakamura, Gahart, Handschumacher, and Russell (1980), a majority of men and women reported some kind of social-sexual encounter with the opposite sex, ranging from complimentary sexual comments to sexual favors as a condition of work. In addition, the majority of men and women reported receiving sexual comments, looks, gestures, or touching on the job (Gutek, Morasch, & Cohen, 1983). As well, in general, men did not report significantly fewer complimentary social-sexual encounters initiated by the opposite sex in comparison to women (Gutek et al. , 1983). In general, these researchers found that males and females reported similar frequencies of social-sexual experiences in their surveys. What they did find however, was that women were more likely to label a particular behavior as sexual harassment than men. Gutek et al. (1980) also found that women were more likely than men to consider sexual comments, looks, or gestures meant to be complimentary, as sexual harassment. Additionally, in discussing social-sexual encounters 101 experienced at work, men were more likely than women to stress the mutuality, as well as positive aspects of the experience. Gutek et al. (1983) found that their study suggested that men view ambiguous, but potentially sexual behaviors initiated by women, as positive experiences but women viewed similar behaviors from men less positively. Collins and Blodgett (1981) also found that when men are faced with stringent sexual harassment guidelines, they may wonder what is all the fuss about. These results are interesting in light of the finding that both sexes agree that sexual behaviors are much more likely to happen to women (Gutek et al. , 1983) yet men report as many social-sexual experiences at work as women and actually report more sexual touching (Gutek et al. , 1980). These results were explained by Gutek et al. (1983) in their findings that men attached a sexual connotation to some interactions with women at work in which women did not label these interactions as sexual. In keeping with this, men have been found to interpret women's friendly behavior as a sign of sexual interest or availabilty (Abbey, 1982; Saal, Johnson, & Weber, 1989) and have interpreted interactions as more sexual than women (Abbey, 1982; Collins & Blodgett, 1981; Shotland & Craig, 1988). Gutek et al. (1983), also found that both sexes agreed that women are less likely to be initiators and more likely to be targets; that behaviors initiated by a woman was more appropriate, or keeping within the bounds of the work role; and that the status of the initiator is more important than the gender of the initiator when it comes to evaluating the appropriateness of social-sexual behaviors at work. It was found that an incident was inappropriate if it was initiated by a higher status person regardless of gender but, if the initiator was equal or lower in status than the target, the incident was believed to be inappropriate if it was a male. 102 In consideration of these findings, the results of this survey seem to reflect data suggesting that both men and women experience social-sexual behaviors at work but experience kinds of behaviors in different proportions depending on gender. Subjects were not asked if they had experienced a behavior that they found to be uncomfortable, unwelcomed, nor were they explicitly asked if the behavior was sexually harasssing. If this had been the case, it is surmised that female respondents may have reported higher incidences than men of situations that they felt uncomfortable with and perhaps labelled as sexually harassing. What is important to note in the results of this survey is that for the statements that one might consider the more serious and are not open to interpretation, that is, being sexually assaulted or raped at work; being promised a promotion in exchange for sexual favors; being threatened with loss of job for refusal to have sex with a supervisor or co-worker; and having important information or job training withheld because of refusal to date a supervisor or co-worker, only female respondents experienced these, in very small percentages however (0.7 to 1.4% for each item). See comparisons in Table H-3 in Appendix H. There is some concern whether more serious types of behaviors such as being sexually assaulted or raped, should have had greater weightings in scoring for the total experience sum. It was elected not to do this because the incidence was extremely low (0.8%). Apart from difficulties in determining how much extra weighting should be added to a particular item, de Vaus (1990) reports that researchers find little merit in extra weighting and thus treat all items in a Likert scale as equally important. Male respondents had higher levels of experience than female respondents in the following statements but it is not known what percentage of males and 103 females found these experiences unwanted or unwelcomed: "Co-workers talked about you in a highly sexualized manner in your immediate work area;" "Workers teased you about your sex life;" and "Was the object of rumors or gossip among co-workers who made remarks about your sexuality or appearance." As Gutek et al. (1983) suggest, men interpret more situations as sexual. This might be relevant to these three statements but it is not in keeping with the fact that male and female respondents had an equal percentage of never experiencing "sexually suggestive looks, stares, or gestures from a supervisor or co-worker" which might be more easily misinterpreted. Interpretation of the survey results for the total experience sum is challenging in the sense that what one person (male or female) might consider a serious problem, another might not, in terms of sexual harassment. However, any score greater than zero indicates some harassment and as the score increases, there is more harassment occurring. The top 10 percentile of subjects who had total experience scores of 9 to 18 were described because their scores were above 2 standard deviations from the mean total experience score for the group. There is no doubt that these subjects would be considered to be victims of sexual harassment. However, subjectively, because sexual harassment depends so much on interpretation and the context of the situation, it is difficult to arrive at a meaningful cutoff point other than a score of zero total experience. It is not possible to totally assess the actual severity of sexual harassment of the survey respondents by total experience sum, although high total scores are indicative of severity. Gruber (1990), suggests that there are five contextual factors affecting severity which are: the source of the act, the duration of the behaviors, the directness of the act, the degree of aversiveness, and the degree of threat. 104 A given act is more apt to be defined as harassing when it is done by someone in a position of authority. The duration or frequency must be sufficiently pervasive and enduring to be legally regarded as harassment. Physical contact increases the likelihood that a person would interpret a verbal statement as harassing and direct physical contact is more intrusive than invasions of personal space or intimate privacy. Seeing offensive acts directed at others or at no one in particular (e. g. , pornographic posters displayed in the workplace) are less directly offensive. Categorical remarks and sexual materials are felt to be low-severity forms of sexual harassment. Relational advances are viewed with lesser disapproval than sexual advances and sexual touching is more harassing than sexual posturing. Acts such as assault or bribery are extremely aversive and threatening. Harassing acts may be more or less threatening depending on the context and are exacerbated by the anger or rage behind the behavior, the number of people involved in the harassment, and whether the victim can avoid the harasser. In summary, this current study may overrepresent the number experiencing sexual harassment because of the large range of questions asked and because of its low useable return rate. Although Gruber (1990) suggests that it is problematic to compare survey results because of methodological differences, the present study found an incidence of 60.1% female respondents experiencing at least one potentially harassing situation in the past two years. This compares to an average of 44% (with a range of 28 to 75%) for the 18 studies analyzed by Gruber (1990). Attitudes About Sexual Harassment Multiple regression analysis was performed to explore what factors if any were related to or could predict, attitudes about sexual harassment. The research questions were: (1) What is the relationship between the independent variables: 105 number of television programs watched on sexual harassment, other sources of knowledge (newspapers, radio, television, magazines, friends, union, other), sex, age, ethnicity, socioeconomic status, education level, and experience of being sexually harassed with the dependent variables, knowledge and attitude about sexual harassment?; and (2) Does any variable or combination of variables from the independent variables of age, sex, education, ethnicity, socioeconomic status, experience of being sexually harassed, number of television programs about sexual harassment watched, and other sources of knowledge, predict attitudes about sexual harassment? In the multiple correlations of the regression analysis, male respondents were moderately correlated with being more accepting and tolerant of sexual harassment than women. Hotelling's j 2 also indicated that men and women were significantly different overall in their attitudes about sexual harassment. The gender variable was also found to be the strongest predictor in the regression analysis of attitudes about sexual harassment contributing the most variance to the dependent variable. Male survey respondents on average were found to be significantly more accepting and tolerant of sexual harassment than female respondents. This result is consistent with a study of university students (Reilly, Lott, & Gallogly, 1986) and a similar study of university students, faculty, and administrative staff (Lott, Reilly, & Howard, 1982) both using the Tolerance for Sexual Harassment Inventory (TSHI). Similar results were also found in a study of university students using the Sexual Harassment Attitude Scale [SHAS] (Mazer & Percival, 1989), an extension of the TSHI. This latter study found that males were more accepting and tolerant of sexual harassment but also that men had more discriminatory and traditional sex role attitudes than women. In addition, those subjects with more sexist and discriminatory attitudes were likely to have more 106 tolerant and accepting views of sexual harassment. It was also found that respondents who rated sexual harassment as serious, tended to endorse fewer sexist attitudes and were less tolerant and accepting of sexual harassment and those who defined more incidents as harassment had less tolerance for it and fewer sexist attitudes. Lott et al. (1982) found that male survey respondents considered sexually related behavior on the job and at school as more natural, more to be expected, and less problematic and serious than their female respondents. This current study also supported these findings with the exception of the seriousness rating of sexual harassment. Men, in this study, agreed more than women with statements taken from the TSHI such as "It is only natural for a man to make sexual advances to a women he finds attractive"; "Most women who are sexually insulted by a man provoke his behavior by the way they talk, act, or dress"; and "It is only natural for a woman to use her sexuality as a way of getting ahead in school or at work." More women than men, on average, agreed with "I believe that sexual intimidation is a serious social problem". This was consistent with results in the study of Lott et al. (1982) and with the management study of Collins and Blodgett (1981) where male managers wondered what all the fuss was about in reference to sexual harassment. However, the majority of both male (67%) and female (78.9%) respondents agreed that sexual intimidation is a serious social problem. However, there was also a sizeable proportion of females (27.9%) and males (41.1%) who felt that concern about sexual harassment makes it harder for men and women to have normal relationships. In the multiple correlations there was also a weak positive but significant correlation of attitudes about sexual harassment with age of the respondents. This variable was also found to be a significant predictor variable of attitudes 107 about sexual harassment in the regression analysis with older respondents being more likely to be more accepting and tolerant of sexual harassment. In a previous study of university students, Reilly et al. (1986), found the opposite effect in terms of age. They found that younger students had a greater acceptance of sexual harassment than older students. Mean ages for their survey respondents were 26.3 for women and 25.2 for men with 56% of the respondents below the age of 23. This contrasts to the current study of working men and women whose mean ages were 38.8 for women and 40.8 for men. It is not known whether these results were due to or characteristic of, different times of life, age cohorts, or sampling bias of the surveys [the study by Reilly et al. [1986] had an even smaller return rate, 33.2%, than the current study]. However the results do compare in the sense that men in both studies were found to be more accepting and tolerant of sexual harassment than women. The main difference between the two studies is that men were younger in the first study, but older than women in this study and this may confound the age variable. When regression analyses were done separately by gender for this study, there were no variables predictive of women's attitudes about sexual harassment. However, regression analysis for men, found that being older and male was predictive of being more accepting and tolerant of sexual harassment. This result would seem to be consistent with studies that have found age differences in attitudes about rape. Severe forms of sexual harassment have been previously compared to rape and previous studies have borrowed theories and hypotheses from the rape literature (Jensen & Gutek, 1982; Pryor, 1987). Research suggests that males over 50 are most likely to endorse rape myth attitudes (Burt, 1980; Yankelovich Clancy Shulman, 1991a) and through selective perception, interpret antirape messages in the media in a stereotypical manner 108 resulting in reinforcement of sexist attitudes (Wilson et al. , 1992). This would be consistent with Renn and Calvert (1993), who found that people who had stereotyped beliefs tended to remember information that confirmed their schemas and dismissed information that might disconfirm them. These researchers found that there appeared to be ongoing changes in contemporary gender roles with college men and women being equally distributed in classifications for being feminine, masculine, undifferentiated, and androgynous. Therefore it would not be surprising to find differences in sexual attitudes in older cohorts when being compared to well educated younger adults. It would be more likely that older individuals would appear to be more traditional in their views about sexual harassment and rape than younger viewers. Research in this related area of rape, infers that sexual attitudes of older males may be more stereotyped and entrenched, thus difficult to change. Getting information from friends about sexual harassment was the third variable significantly related to attitudes about sexual harassment for the entire survey group of respondents. Information from friends was weakly negatively correlated with attitudes about sexual harassment. Getting information about sexual harassment from friends was predictive of survey respondents being less accepting and tolerant of sexual harassment. However, when separate regression analysis by gender was performed, getting information from friends was actually predictive of male respondents being less accepting and tolerant of sexual harassment and was not predictive for female respondents. According to Erbring, Goldenberg, and Miller (1980), informal communication with others is essential in helping people understand news media content. These researchers found that informal communication with others plays a critical role in shaping public perceptions of the salience of issues. It is a social process that generates a collective definition of a situation 109 and it may strengthen or block the impact of the content of news media depending on the homogeneity of the social environment. This alternative channel of information reinforces or dissipates the impact of news media content depending on the similarity of the messages from these sources. Erbring et al. (1980) found that in terms of politics, people who got their information from social interaction did not depend on formal channels of communication nor on local real-world conditions to find out what was salient. They suggested that by talking to others, respondents may broaden their perspectives beyond their real-world experiences and with the passage of time, information from others increasingly dominates the outcome and erases any remaining effects of differential media exposure. Conversely, they found that people who did not talk to others about political issues depended heavily on media content and local experience to determine issue salience and that impact from media was then dramatically amplified. People who do not talk to peers are slower to get the message and continue to show signs of whatever media they are exposed to. The fourth variable that was significantly correlated with attitudes about sexual harassment was the number of television programs watched about sexual harassment. This variable was weakly negatively correlated with attitudes about sexual harassment and it was found that watching television programs about sexual harassment was predictive of being less accepting and tolerant of sexual harassment. It has been found in advertising that repetition of the message and pervasiveness of the message in many modes and media help penetrate our consciousness and channels our modes of thinking (Pollay, 1986). Thus it would seem, the more television programs watched about sexual harassment, the more likely prosocial messages would be effective. However, hours of television 110 watched, message units recalled, or number of programs watched are empirical ways of measuring the abstract construct of television exposure. This term is a surrogate for a more abstract concept such as attentiveness, involvement, or comprehension (Allen & Taylor, 1985). As previously discussed, Wilson et al. (1992) found that in viewing a prosocial movie about rape, the greatest attitude change occurred for older women who became more emotionally involved with the victim and thus paid more attention to the message of the movie rather than processing the movie content with superficial, shortcut strategies based on stereotypes. Those women who already had antirape views, were less affected by the movie because they were already knowledgeable. Complementary to these results was the finding that for political attitudes there was a positive association of television news exposure and attitude consistency. At lower levels of ideological utility, television exposure was associated with higher levels of attitude holding than for subjects with higher levels of ideological utility (Reese & Miller, 1981). It was suggested by these researchers that television condenses, simplifies, and summarizes information, thus making it easier for the unacquainted viewer to become aware of issues and learn "what goes with what." Television news may provide the basis for political attitudes and contribute to a more structured, well-stocked portfolio of political attitudes. Thus people lacking in knowledge about an issue or people having not formed opinions about an issue, could be the group most likely affected by television exposure. It was also suggested that even people who were uninterested in politics actually learned incidentally about issues. It has also been suggested that persuasion occurs via two routes. Under low involvement, persons respond to superficial cues in the persuasion context such as source likabiltiy or credibility. Persuasion occurring in persons with high I l l involvement is a function of the substance of the arguments presented in the message and is characterized by careful thought and consideration (Petty & Cacioppo, 1984). Taking these various factors into account, it is not surprising that the number of television programs watched about sexual harassment was only weakly correlated with being less accepting and tolerant of sexual harassment. It would seem that tracking mere exposure to the programs cannot adequately contribute full explanatory power for all groups of viewers to the relationship of television exposure and attitudes. These results are suggestive that watching television programs about sexual harassment may be effective for those persons who are lacking knowledge about sexual harassment and that if a particular program is presented in such a way as to raise emotional involvement, and thus attentiveness, the viewer may more closely attend to the prosocial message. This may result in raising awareness about the issue, but perhaps not result in changing attitudes unless the viewer has other sources of influence such as friends and credible information from other sources that combine to contribute to attitude change. However, it is also possible, that if the television programs present problems associated with sexual harassment from only the female-as-victim perspective, attitudes about sexual harassment of male viewers, particularly, males older than 50, may not be positively affected by the prosocial message. Presenting males as victims might result in programs seeming to be more personally relevant to the male audience thus more involving and potentially more effective. These results may also suggest that people who are already fairly knowledgeable about sexual harassment may not benefit from watching programs about sexual harassment. This would be consistent with Leff at al. (1986), who found that the less information and sensitivity the public has to an 112 issue prior to its treatment in the media, the more likely the public will be influenced by subsequent media reports. They suggested that there may be a saturation point at which the public already has knowledge and has formed opinions on an issue. Thus for those individuals already knowledgeable about sexual harassment, television programs about it may simply reinforce their views and not change their attitudes. Knowledge of Sexual Harassment Multiple regression analysis was performed to explore what factors if any were related to or could predict, knowledge about sexual harassment. The research questions were: (1) What is the relationship between the independent variables: number of television programs watched on sexual harassment, other sources of knowledge (newspapers, radio, television, magazines, friends, union, other), sex, age, ethnicity, socioeconomic status, education level, and experience of being sexually harassed with the dependent vaiables, knowledge and attitude about sexual harassment? and (2) Does any variable or combination of variables from the independent variables of age, sex, education, ethnicity, socioeconomic status, experience of being sexually harassed, number of television programs about sexual harassment watched, and other sources of knowledge, predict knowledge about sexual harassment? Multiple correlations indicated that women were moderatly associated with being more knowledgeable about sexual harassment than men. Gender was also the strongest predictor of knowledge, contributing nearly all of the unique variance to the knowledge variable. The finding that being female was predictive of being more knowledgable about sexual harassment than men is intuitively consistent with females being less accepting and tolerant of sexual harassment than male respondents. Additionally, as in the predictions of attitudes about sexual harassment, information from friends was also found to be 113 predictive of knowledge about sexual harassment for the entire survey group. Respondents with lower socioeconomic status were weakly associated with being more knowledgeable about sexual harassment but socioeconomic status was not predictive of knowledge. Additionally, respondents with more experience of being sexually harassed were weakly associated with being less knowledgeable about sexual harassment. This variable, however, was also not predictive of knowledge for the entire respondent group. The differences in knowledge between men and women may be partially explained by factors discussed in relation to 'Knowledge Gap' (Ettema & Kline, 1977). These researchers, in reformulating the knowledge gap hypothesis, suggested that gaps in knowledge among different groups may vary as the infusion of mass media information into a social system increases. Those segments of the population motivated to acquire information and/or for which the information is functional tend to acquire the information at a faster rate than those who are not motivated or for which the information is not functional. Gaps in knowledge between these groups may tend to increase under these circumstances. This would tend to fit the general differences experienced by men and women in regards to sexual harassment. It would be functional for women to know about sexual harassment because they experience certain socio-sexual behaviors at work that are generally unwanted, cause discomfort, and are interpreted negatively. This would also be consistent with women in lower socioeconomic status, where women would tend to be in low paying, low status positions and more subject to harassment. There is a motive to acquire information about sexual harassment for women in general, as well as for women with low socioeconomic status, and to become knowedgeable about sexual harassment and their personal rights in the workplace. In contrast, men 114 interpret socio-sexual behaviors differently than women and may feel socio-sexual interactions as being mutually enjoyable, more normative, and more acceptable in the workplace. They are less likely to view some behaviors as unacceptable (Gutek et al., 1983). Fitzgerald et al. (1988) also cite that men may concentrate on whether the situation is a serious one that may involve negative consequences for either the recipient or the initiator or whether the behavior could be considered relatively trivial. These differences in perceptions of men from women may make knowledge about sexual harassment seem personally not relevant and meaningful, and therefore not functional for men. Since there generally has been an increase in awareness about sexual harassment in the last two decades and there has been movement to reduce harassment, there now may be more motive for males to become more knowledgeable about harassment. It is now functional to be knowledgeable in order to avoid potential charges of harassment and therefore knowledge about it is more important than in the past. It is possible that the knowledge gap in this area may narrow if sexual harassment continues to be focussed upon. In addition, interpersonal communication during social contact is an intervening variable Unking motivation and function to knowledge acquisition (Ettema & Kline, 1977). Interpersonal communication may or may not cut across social strata, thus certain groups may not discuss sexual harassment whereas others do. This could be the relevant issue for male respondents in terms of attitudes about and knowledge of sexual harassment. The males of the survey who were less accepting of sexual harassment were associated with getting information from friends. This group may have felt more motivated to learn about sexual harassment because they learned from their friends that it was functional for them to know about it and thus may have been more likely to 115 attend to programs dealing with sexual harassment or even to have sought out such programs in order to gain knowledge. It was also found that there was a weak negative correlation of being sexually harassed with knowledge about sexual harassment. This fits the data for this study in the sense that male respondents had higher total scores than women for experiencing harassing behaviors but at the same time were also less knowledgeable about sexual harassment than women. This result can also be explained in terms of perception. Men would be less likely to place negative connotations to socio-sexual interactions at work and thus would be less likely to interpret information about sexual harassment as personally relevant. In terms of media, however, it would be expected that people with more experience of being sexually harassed might attend more to the message of a program about sexual harassment because it would be personally relevant. Information from health literature and public service announcements have reported the importance of viewer self-interest in relation to the effects of public service announcements. The viewer must define the message in the initial stages as "for me". Personal experiences affect viewer decisions about content relevance and thereby affect interactions among processes of attention, comprehension, and motivation (Calvert, Cocking, & Smreck, 1991). Additionally, prior knowledge about a topic affects the comprehensibiltiy of the message and thus will influence whether or not the viewer decides to pay attention to the message (Calvert et al., 1991). It would be expected then, that males who do not see sexual harassment as personally relevant and who do not have much knowledge about sexual harassment, would be less likely to pay attention to a prosocial message about sexual harassment on a television program. 116 Personal experience of being sexually harassed, although weakly negatively correlated with knowledge, was not found to be a predictor variable of knowledge because it may have been confounded with differences of male and female perceptions of their actual experiences of harassment. Information about sexual harassment may not have seemed relevant to male survey respondents with lack of knowledge about sexual harassment perhaps playing a factor in this as well. Socioeconomic status was also correlated with knowledge but was not found to be predictive of knowledge. There was a small significant negative correlation of socioeconomic status with knowledge indicating that respondents from lower socioeconomic status were weakly associated with being more knowledgeable about sexual harassment or conversely, repondents with higher socioeconomic status were weakly associated with being less knowledgeable about sexual harassment. There was no discemable pattern to the relationship when plotting subjects socieconomic status against knowledge scores so it is unclear what the actual relationship is. Both interpretations do fit the data. Female respondents, overall, had lower socioeconomic status than male survey respondents and females on average, were more knowledgeable about sexual harassment than males. Conversely, males had higher socioeconomic status than females and were less knowledgeable than females. According to 'Knowledge Gap' hypothesis, the higher socioeconomic segments of a population acquire information from the mass media faster than those from lower socioeconomic segments, thus increasing the difference in the amount of knowledge held by the two segments (Tichenor et al. , 1970). Just as communication researchers have found that persons of higher SES may be better informed about public knowledge, research in medical sociology and on the diffusion of innovations have cited persons of higher SES as being better 117 informed (LeBlanc, A. , 1993). Logically then, male survey respondents with higher socioeconomic status should have been correlated with greater knowledge rather than lesser knowledge. What these results suggest is that a motivational difference or difference in interest in the issue may be relevant in explaining this weak, but significant correlation. In summary, there was no variable predictive of male survey respondents' knowledge about sexual harassment. There were two variables, age and information from the union, that were statistically significant in predicting knowledge levels of female respondents, but these variables contributed less than 10% of the variance to the knowledge variable and therefore are not considered to be of practical significance. For the entire group of survey respondents, gender predicted knowledge, with males being predictive of having less knowledge about sexual harassment than female respondents. In addition, information from friends was also predictive of being more knowledgeable about sexual harassment. As discussed above, these two predictor variables only accounted for 17% of the variance of the knowledge variable. It is likely that there are unspecified variables that may add to the prediction of knowledge about sexual harassment. Additionally, the small sample size available for the regression analysis for male respondents may have been a factor in finding no predictor variables. However, these results may also indicate the broad range of factors involved in such a complicated issue as sexual harassment with its broad range of individual interpretation as well as the complexity of studying and assessing the effects associated with mass media. Perceived Change of View About Sexual Harassment; There were two research questions related to people associating their views changing about sexual harassment associated with watching television programs about sexual harassment: (1) Is there a relationship of television 118 viewing of sexual harassment programs with changes of view in regard to sexual harassment? and (2) Does any variable or combination of variables from the independent variables of age, sex, education, ethnicity, socioeconomic status, experience of being sexually harassed, number of television programs about sexual harassment watched, and other sources of knowledge predict changes of view about sexual harassment? Results of the multiple correlations indicated that there were five variables significantly associated with perceiving that one's views about sexual harassment had changed associated with watching television programs about sexual harassment. Education was negatively correlated with perceived change of view indicating that the more educated the respondents, the less likely they would associate changing their views with watching television programs about sexual harassment. This result may reflect a posited ceiling affect (Ettema & Kline, 1977). These researchers suggest that an audience-imposed ceiling may occur where persons more knowledgable about a topic may lack motivation to acquire any more knowledge, whereas persons lacking in knowledge might use electronic media to " catch up" on information. In addition, it has been found that persons with higher education levels tend to use the print media for information and are less likely to use television and other electronic media as a source of knowledge (LeBlanc, 1993; Samuelson, Carter, & Ruggels, 1963; Wade & Schramm, 1969). Therefore, it is less likely that higher educated individuals would perceive their views changing associated with information from television since they generally have a much broader base for sources of information. Gender was weakly associated with change of view and was interpreted from the results that women were weakly associated with changing their views about sexual harassment. In addition, respondents with lower socioeconomic status 119 were weakly correlated with changes of view about sexual harassment. Respondents who perceived they got their information from television or from magazines were also positively correlated with change of view. However, for the entire group of survey respondents, only information from magazines was predictive of respondents perceiving their views changed about sexual harassment associated with television programs dealing with sexual harassment. Regression analyses done separately by gender indicated varying results depending on gender. For female respondents, lower socioeconomic status and lower levels of education were weakly correlated with perceived change of view. Additionally, number of television programs watched and information from magazines were weakly correlated with perceived change of view. Of these four variables, regression analysis indicated that only socioeconomic status was a predictor variable indicating that women with lower socioeconomic status were more likely to perceive their views about sexual harassment had changed associated with viewing television programs about sexual harassment. For the regression analysis for the male respondents, getting information from the union and getting information from magazines were weakly correlated with perceived change of view. Only information from magazines was found to be a predictor variable of male respondents perceiving that then-views had changed associated with watching television programs about sexual harassment. Previous research has found that when people were asked to say from which source they got the most information about public affairs information, television was more likely to be the major source of information for people with little education, for females, nonwhites, farm workers, and blue-collar workers. In contrast, the print media was more likely to be the major source of 120 information for highly educated groups, whites, males, professional, managerial, white-collar workers, and high income groups (Wade & Schramm, 1969). More recent studies in health communication have also generally found that persons of higher social status and educational attainment seem to have greater access to both printed and broadcast media (LeBlanc, 1993). In a review of mass communication effects, Roberts and Bachen (1981) reported that adult women view more TV than men and the amount of viewing is inversely related to education, income, and occupational status. However, studies reviewed were from the 1970s and at that time these relationships were reported as weaker than earlier years. It is not known if these relationships would be even weaker today with the increased numbers of women working outside the home than in the 1970s. However, these relationships do seem to fit the results of this study with female respondents with less education and lower socioeconomic status perceiving their views changed associated with television programs about sexual harassment. The weak correlation of education with perceived change of view may be falsely weak because of the relatively high levels of education for both males and females of the study. There were only 4 subjects (1.7%) who had less than high school education and only 28 subjects (11.8%) who had high school education. Thus the respondent sample was quite well educated and is not comparable to older studies where subjects tended to have lower levels of education. Hence, the socioeconomic index used in this study may be a better indicator for female respondents because it is a portrayal of the occupational structure as a hierarchy based on a set of intercorrelated variables chiefly composed of prestige, income, and education (Blishen et al., 1981). As discussed above, learning about sexual harassment may be more functional for some groups than others. As LeBlanc (1993) suggests, there is a 121 challenge in gaining an understanding of how SES's effect on knowledge is altered by the cultural prism through which it is directed. Women are more likely to be harassed when they are employed in low paying, low status jobs (USMBP, 1981). Because there is a tendency of lesser educated women to use television as a source of information and a greater likelihood of women to have a lower socioeconomic status than men, it would seem to fit the data that women with lower socioeconomic status would perceive that their views have changed associated with television programs about sexual harassment. This group would be more likely to see television as their source of information thus could perceive that they learned information associated with television. In contrast, those subjects who use a wide assortment of sources of information, or who tend more to use print media, would be less likely to attribute an association with television and learning. Information from magazines was predictive of both the entire group and for the male respondents of the survey who perceived a view change about sexual harassment associated with television programs about sexual harassment. It is not clear how information from magazines interacts with perceiving a view change associated with television viewing. It is also unfortunate that it is unknown what type of magazines were read by subjects who perceived then-views had changed. It has been said that magazine advertisements often mirror contemporary society and often reflect the prevailing attitudes and stereotypes held by society. Examination of women's roles through content analysis of two magazines, "Ladies Home Journal" and "Good Housekeeping" over 30 years from the early 1950s to the early 1980s revealed a gradual decline in themes of women as wives, mothers, and homemakers and an increase in articles with political, social, and economic (i. e. , career) themes. However, traditional sex role modes 122 still dominated the pages of most women's magazines (Demarest & Garner, 1992). Similarly, Sullivan and O'Connor (1988) examined women's role portrayals in the advertising of 8 general interest magazines from 1958 to 1983 and concluded that advertisements in some ways more accurately reflected the true diversity of women's social and occupational roles than of those of earlier time periods where women's place was considered to be in the home. They found that there was an increase of portraying women as employed rather than working in the home; there was an increase in portraying women as business executives, professional, salespersons, and midlevel managers requiring meaningful decision making; and an increase of showing women as independent rather than dependent. In addition men and women tended to be portrayed as equals involved in sharing lifestyles. Correlated with women's entrance into the work place, publications of new periodicals have occurred which portray women working outside the home (and often in non traditional occupations) (Belknap & Leonard, 1991). In these newer magazines, women have also been cast in a more positive light, i. e., having roles with power, influence, and responsibility. The core message of new magazines is that working for pay is an integral part of a woman's self-concept (Ruggiero & Weston, 1985). In a study of relationships of men and women in magazine advertisements, using frame analysis, Klassen, Jasper, and Schwartz (1993) noted that although there was a disproportionate number of advertisements portraying women in "traditional" poses relative to advertisements showing women and men as equals, traditional depictions of women have been decreasing since the early 1980s and "equality portrayals" of men and women have been on the rise. What was surprising to these researchers was that "Playboy," a bastion of traditional male roles, showed a decrease in traditional portrayals of women and an 123 increase of portraying men and women as equals in their ads. Complementary to this was a study of social responsibilty themes by magazine advertisers (Lill, Gross, & Peterson, 1986). During the interval from 1977 to 1984 there was a major increase in social responsibilty themes in magazine advertisements associated with consumerism, race, and equal opportunity for women. Magazines that had more social responsibility ads than others were women's, news, general interest, and men's publications. Trade and business magazines tended toward the not-frequently-mentioned end of the scale. Unfortunately, there is very little information available on changes of male roles in magazines in recent years. There has been a study that shows men have been portrayed increasingly in decorative roles and less frequently in traditional "manly" activities and in more recent years, men have been portrayed in nontraditional roles or roles in which men and women are treated as equals (Skelly & Lundstrom, 1981). In British magazines, although they may not be comparable with North American publications, the sex appeal image is the most frequent portrayal of men in contemporary men's magazines (Lysonski, 1985). Sports/outdoors-oriented magazines commonly portray men in activities outside the home and in career-oriented portrayal. Almost 70% of advertisements in professional men's magazines have career-oriented portrayals for men (Lysonski, 1985). The overall conclusion that is generally made by researchers is that some inroads have been made in magazines in depicting more accurate portrayals of women in relationship to contemporary reality, but traditional roles continue to abound. Depending on the type of magazine a person chooses to read, the message may be varied as to women's role in society, at work, and in relation to men. In turn, men seem to also be presented in less traditional ways. The 124 conclusion seems to be that women's magazines are sensitive to change in society, but they are slow to change and that there is a cultural lag. As Demarest and Garner (1992) suggest, although there is greater coverage to nontraditional issues and lifestyles, women's magazines devote a lot of space to women's physical appearance rather than social equity with individual adaptation stressed rather than social action and social change. Glazer (1980) also suggests that contemporary magazines reflect the ideology of women and the "double day." Women are solely responsible for organizing their days of work with their domestic duties and their resulting problems associated with this double responsibility is a personal problem rather than a societal problem. She suggests that the double day contributes to women taking paid jobs that are compatible with being a wife and mother first and foremost. Thus, we still have themes of traditional sex roles in our printed media as well as the broadcast media, thus a continuity remains in our society with earlier decades but yet a change has occurred. How this relates to interpretation of magazines being predictive for the entire group of survey respondents and for male survey respondents in relation to perceived change of view can only be surmised. As discussed above, better educated persons have greater access to all types of media and tend to use print media primarily as a source of information. Because the survey respondents are generally quite well educated with the most common education level being some college or university, it is likely that respondents have become knowledgeable about sexual harassment from reading about it. Since information from the union was weakly correlated with men perceiving a view change associated with television, it could be surmised that information from the union combined with information gained from magazines could have promoted interest in the issue of sexual harassment and perhaps even feelings of conflict. It is suggested 125 in the study of knowledge gap that mass media information is more likely to narrow knowledge gaps when the knowledge domain being studied is salient and is conflict-ridden (Ettema & Kline, 1977). There is a great deal of conflict in terms of how people feel about sexual harassment, thus motivation to acquire information about it should likely increase as a result of interpersonal discussion which information from the union could generate. Prior knowledge then could predispose some male respondents to attend to television messages about sexual harassment. Because they may have sought out such programs to learn more about sexual harassment, it is possible that they would perceive television as being associated with a change of view. In terms of the whole group, magazine reading may be predictive of not only males but perhaps those females who read the newer types of magazines that project women in less traditional roles and that present social problems associated with women and paid work. Perhaps information from magazines is a consciousness raising experience for some women and men that promotes interest in other sources of information such as television. Becoming aware of the issue may raise some individuals' recognition that the issue was presented on a television program. Once again, it must be reiterated that the predictor variables found in the regression only contributed a small portion of the variance of the change of view variable. There may be other unspecified variables that may enhance our understanding of what contributes to changes of view about sexual harassment in relation to television viewing. Limitations of the Study There are several limitations to the study, one of which is connected to the respondent sample. The first problem with the study is the low survey response 126 rate (50.8%) and thus possible sampling bias. Although the survey respondents appeared to be fairly representative of the union population in terms of male/female ratio and in terms of the percentage of respondents in the union components, there was a large percentage of subjects sampled who did not complete the survey. According to de Vaus (1990), nonresponders tend to be different from responders. Nonresponders tend to be older, less educated, and tend to have non-English-speaking backgrounds. The majority of respondents in this study were well educated, middle class Caucasians with a small number of minority persons being represented. It is difficult to know whether the low numbers of minority respondents and the relatively homogeneous education levels of respondents reflected sampling bias or the true nature of the union population. It is quite possible that those subjects who did not respond to the survey may have differed significantly from those who did participate. Therefore results of this survey may at the least, represent only the survey respondents and at the most, may reflect the union population. The results do not generalize to society as a whole. In addition the ethnicity variable was not assessed in the regression analyses because of lack of numbers. This also restricts the generalizability of the study results. In terms of attitude and knowledge about sexual harassment, as previously mentioned, deletion of outliers for the regression analyses may have resulted in the findings not generalizing to older males less knowledgeable about sexual harassment and more tolerant and accepting of sexual harassment. In addition, respondents with low socioeconomic status who were less knowledgeable about sexual harassment and who were more accepting and tolerant of sexual harassment may also not be represented by the regression analyses in terms of attitude and knowledge. In terms of perceived change of view, results of the regression analysis may not generalize to older males of low socioeconomic 127 status with lower levels of knowledge and who are more accepting and tolerant of sexual harassment. Missing data may have also compromised the study results. It is possible that refusal to answer questions about attitude, knowledge, experience, etc. may have affected the relationships of the variables. Because missing data was handled by mean replacement, the variance of the variables may have been reduced and thus resulted in lower correlations with other variables (Tabachnick & Fidell, 1989). Mean replacement was also used for the typing error made in the perceived change of view section on the first mailing. This may have caused lower correlations as well. A further limitation of the study was the weak relationships of the independent variables to the dependent variables and the relatively small amount of variance contributed to the dependent variables by the predictor variables. These weak but significant relationships could suggest that there may have been other variables that could have been explored that may have enhanced predictions of the dependent variables of the study. Specification error in its two most common forms, leaving out an important causal variable or secondly, including an irrelevant variable in the regression equation could be both present in this exploratory study. The first problem results in biasing tests of statistical significance in a conservative direction (i. e. , it is harder to reject the null hypothesis) while including an irrelevant variable creates inefficiency in estimates of the regression coefficients and the intercept (Berk, 1983). Weak relationships found in the study may also reflect measurement error. Both areas that this study encompasses, sexual hararassment and communications, are frought with a wide variety of problems in terms of accurate measurement. For example, sexual harassment is difficult to study 128 because of the wide variety of individual differences in interpretation of socio-sexual behaviors and is heavily impacted by context of the situation which a survey cannot adequately capture. Traditional sex role schemas versus non traditional views further complicate this area. In terms of media: education level, attention, personal experience, motivation, and prior knowledge are but a few of many factors affecting how individuals interpret and attend to media messages. Further complicating this type of research is the difficulty in trying to separate out what we learn from media and from which type of media, if that is even truly possible in this age of multimedia use and exposure. How does one really separate out and accurately measure relationships of different media combined with our interpersonal communications with friends, colleagues from work and school? Thus results of this study may be quite in keeping with the "nature of the beast" i. e. , the inherent problems in studying both fields, and particularly, in combining the two areas together in one study. It was also noted by the researcher that some of the remarks made by survey participants indicated that the survey questions did not fully capture the breadth of personal attitudes and knowledge about sexual harassment. A survey, by its very nature, is restrictive in this regard and therefore may contribute to measurement error. Additionally, in terms of measurement problems, cases-to-IV ratio in terms of regression analyses done for the male survey respondents was at the bare minimum requirement (5 to 1). According to Tabachnick and Fidell (1989), power may be unacceptably low if you have fewer than 100 cases in the regression analysis. This problem may be relevant to the lack of any significant predictor variables found for the knowledge variable for the male respondents 129 and possibly for the attitude variable being statistically significant but not practically significant for the female respondents. A further challenge to the study is personal recall and the limits of retrospective questions. According to Pearson, Ross, and Dawes (1992), peoples' responses to retrospective questions are influenced by their current psychological and environmental state as well as by their explicit or implicit theories which consist of schemata, narratives, and scripts they hold about themselves and society. In terms of personal attributes people may notice the current status on the attribute in question (e. g. , relevant to this study, perceived attitude change) and use this as a benchmark. The question may be asked: Is there any reason to believe that I felt differently then than I do now? According to Pearson et al. (1992), people often invoke an implicit theory of stability or change to guide their construction of past attributes. Recall is mediated depending on whether individuals see their attitudes as consistent over time or changing over time. People possess implicit theories of change and ideas about what things or conditions that may foster alterations in themselves and others. People tend to exaggerate the consistency between their present (new) attitudes and their past opinions presuming that they have always felt the way they do now (Pearson et al. , 1992). Thus some survey respondents would expect no attitude change where others would expect it and may respond accordingly. Depending on one's personal theory of change, people may overestimate or underestimate a change. In conjunction with this, memories consistent with peoples' beliefs are often more accessible than memories that are inconsistent with their beliefs (Pearson et al. , 1992). Thus subjects who are currently less accepting and tolerant of sexual harassment could be more likely to remember programs about sexual harassment and conversely, subjects more tolerant and accepting of sexual 130 harassment may possibly be less likely to remember such programs. Further to this, as the recall period increases, respondents become more dependent on using their present status as a benchmark to interpret change. Additionally, some individuals may exaggerate an improvement in a direction consistent with their personal theories as to what ought to happen when taking a course meant to enhance skills (Pearson et al. , 1992). Thus if subjects believe that they learn from television, they might be more apt to perceive a change of view associated with programs about sexual harassment. Recall has also been found to be a function of time and of the saliency of the event or behavior being measured. The more distant the event, the more likely it is to be forgotten or incorrectly moved forward. Salient events are more likely to be recalled than nonsalient ones, where saliency is a function of the unusualness of the event (e. g. , Anita Hill and the senate hearings), its economic and social costs and benefits, and its continuing consequences (Pearson et al., 1992). These findings on recall raise important issues for the present study, particularly in terms of perceived attitude change. There is a question as to how well respondents could actually recall seeing television programs about sexual harassment and since some of the television programs would have been seen several months prior to the survey, could subjects accurately recall what they had seen or felt? Of particular note was the Clarence Thomas Senate Confirmation Hearings which occurred 2 years previous to the survey. This was included because of the extreme notoriety which was posited would counteract the recency in viewing problem. Because of the concern about recall, 4 subjects who participated in the pilot study were briefly interviewed in March during the time of the third mail out of the study. These four subjects were asked details about the sexual harassment 131 programs that they said they had watched. Of the four, 2 subjects could remember a lot of detail about the programs. The other two subjects could not remember details about the programs but did remember what they felt about them at the time of viewing. Three of the four subjects, referred to remembering how they felt when watching the Senate Confirmation Hearings. These findings, although very limited, suggest that time between viewing the television programs and doing the survey may not be as serious a problem as other factors. It could be surmised that people who did perceive that their views changed about sexual harassment were those people who watched a program that was salient or relevant to them and may have had internal theories of change rather than of consistency. Another aspect associated with recall is the relation between gender schemas and adult recall of televised information. On the one hand this information is an additional limitation to the study in terms of the perceived change of view variable but at the same time may be useful in terms of explanation. According to Renn and Calvert (1993), gender aschematic (i. e. , androgynous and undifferentiated) adults recall more counterstereotypical information than gender schematic adults. Gender aschematic adults are people who do not organize their self-concepts around gender stereotypes or attributes (Bern, 1983). In measuring recall of stereotypical and counterstereotypical information from television programs, Renn and Calvert (1993) found that all subjects recalled equally well traditional stereotyped information on television, but those subjects who were gender aschematic recalled significantly more counterstereotypical information than did people who were gender-schematic. In addition, they found that nontraditional men remembered counterstereotypical information better than traditional women. Additionally, they suggested that their study indicated that people who hold stereotypical 132 beliefs tended to remember information that confirmed their schemas and dismissed information that might disconfirm them. This would be consistent with the findings of Wilson et al. (1992) who found similar results in older males presented with a prosocial message in a movie on rape. It is possible that those people in this study who perceived that their views had changed in association with watching television programs about sexual harassment, may be those people who tend to be towards the gender aschematic end of the spectrum. Although this study demonstrated a moderate positive relationship between attitudes about sexual harassment and knowledge of sexual harassment, it did not assess whether behavior was also correlated. This study suggests that there may be certain groups of people who can gain knowledge about sexual harassment and perceive an attitude change associated with watching television programs about sexual harassment, but it did not assess whether behavior change would be correlated with either knowledge gain or attitude change or both in combination. In addition, because of the correlational nature of the study, conclusions about causation cannot be made. Implications for Further Research Results from this survey seem to suggest that further research could be done comparing effectiveness of different methods of teaching people about sexual harassment. Although the relationship was weak, information from the union was negatively correlated with female respondents' knowledge about sexual harassment and information from the union was weakly positively correlated with male respondents perceiving a change of view associated with television. It would seem that reading materials alone may not be adequate or may be effective for only certain groups. Results of the study would suggest that television may be helpful for learning about sexual harassment and perhaps be 133 associated with attitude change. It would seem likely that a combination of giving reading material about sexual harassment, showing a short film or video about sexual harassment and then having a discussion following the movie or video would be likely to be the most effective. A study comparing different combinations of the above might prove to be quite helpful. The findings that getting information about sexual harassment from friends was associated with being less tolerant and accepting of sexual harassment could be pursued in a study of how to decrease sexual harassment. It would seem that in problematic work environments, teaching the most popular and socially active persons at a work site or teaching highly regarded individuals at work about sexual harassment could possibly generate improvement in the work environment through interpersonal interactions and discussions by such persons with their fellow workers. Pursuant to this, identifying those who harass and placing them in work situations with same sex persons knowledgeable about sexual harassment, who additionally may be well liked and highly regarded by work peers, may be useful in reducing problematic behaviors of the harasser. I also tentatively suggest that comparing men and women in terms of sexual harassment may not be the most productive avenue in terms of eliminating sexual harassment. Perhaps a qualitative study could be done asking male and female workers what changes need to be made at their work site as well as what behaviors from co-workers are needed and what behaviors should be eliminated to enhance a feeling of comfort with peers and supervisors. A dialogue among employees in their respective work areas could be employed to help workers form a consensus of what behaviors are not wanted and not acceptable in the work environment. 134 It must be also mentioned that same sex harassment was not explored in this study and should be done in the future. Additionally, it was commented by one of the survey respondents that members of the general public at large who come into the workplace sometimes sexually harass employees. This was not explored in this study at all and perhaps would be of great benefit to learn how often this occurs and to also find ways of coping with this problem. Implications for Counselling As Renn and Calvert (1993) suggest, it would appear that there are ongoing changes occurring in contemporary gender roles. There are others who suggest that a backlash is occurring against women as they stride towards gaining their equal place in the world of work and in society as a whole (Faludi, 1991) or that women's concentration on social equity is being systematically subverted and detoured by the indoctrination of the belief that a beautiful physical appearance is a primary role for women (Wolf, 1990). As women continue to work outside the home and as segments of our society press for men to change while other segments want traditional values to remain or be reestablished, it would seem that counsellors may need to learn more and more about attitude change as well as behavioral change in relation to social issues such as sexual harassment. In an era where apparent increasing violence and aggression is occurring, ways of supporting or creating social change may become ever more important. Studying ways of using media or altering current media presentations and advertisements may be a successful avenue for this purpose. In terms of work environment, from a few of the responses to the survey, it seems that some respondents (particularly men) are unclear and concerned about how they should behave with the opposite gender at work. In addition, some respondents, both men and women, seem to have missed the reality that 135 sexual harassment really is a problem for some individuals and see the issue as being overstated. The role of counselling in this vein may be as an educative as well as supportive one for helping those victims deal with their harasser at work as well as deal with co-workers who may seem very unsympathetic to their problems at work. Conclusion The current study explored relationships associated with attitudes and knowledge about sexual harassment in association with television. It also explored whether television viewers associated television programs about sexual harassment with changing their views about sexual harassment. While there are limitations to this study, the data presented a small window into the complex interactions that may be occurring between media, societal change, and societal continuity. What this study demonstrated was that even though female respondents on average were more knowledgeable about sexual harassment and were on average, less accepting and tolerant of sexual harassment than male respondents, male and female participants in this study agreed more than they disagreed in terms of attitudes about sexual harassment. It is concluded that it would be unhelpful to assume that all male respondents or all female respondents interpret, experience, or evaluate sexual harassment similarly to other members of their corresponding gender. It may also be likely that there are small groups of both men and women who still have very traditional views about sex-roles and the role of women and work. It would seem then, very important to not assume all men are more accepting and tolerant of sexual harassment just as not all women are less accepting and tolerant of sexual harassment. 136 It is concluded that there may be many factors involved in predicting attitudes and knowledge about sexual harassment as well as attitude change. And just as there are many factors probably involved in prediction of these variables, it is likely that there may be many factors that affect how to change attitudes about sexual harassment as well as change unwanted socio-sexual behaviors at work. We also learned from this study that we could predict to a small degree subjects' attitudes and knowledge about sexual harassment and to a small degree, who might associate a perceived change in view about sexual harassment associated with television programs about sexual harassment. We learned that there might be some potential for using television as a source of attitude change. We did not, however, learn whether this might also promote behavioral change. 137 References Abbey, A. (1982). Sex differences in attributions for friendly behavior: Do males misperceive females' friendliness? Journal of Personality and Social Psychology. 42. 830-838. Allen, R. , & Taylor, B. (1985). Media public affairs exposure: Issues and alternative strategies. Communication Monographs. 52. 186-201. Allen, M., & Yen, W. (1979). Introduction to measurement theory. Monterey: Brooks/Cole. Babbie, E. (1973). Survey research methods. Belmont, CA: Wadsworth Behr, R. , & Iyengar, S. (1985). Television news, real-world cues, and changes in the public agenda. Public Opinion Quarterly. 49. 38-57. Belknap, P. , & Leonard, W. II. (1991). A conceptual replication and extension of Erving Goffman's study of gender advertisements. Sex Roles. 25. 103-118. Bern, S. (1983). Gender schema theory and its implications for child development: Raising gender aschematic children in a gender schematic society. Signs: Journal of Women in Culture and Society. 8_, 598-616. Berk, R. (1983). Applications of the general linear model to survey data. In P. Rossi, J. Wright, & A. Anderson (Eds.), Handbook of survey research (pp. 495-546). New York: Academic Press. Blishen, B. R., Carroll, W. K., & Moore, C. (1987). The 1981 socioeconomic index for occupations in Canada. Canadian Review of Sociology and Anthropology. 24, 465-488. Brenneman, D. S. (1992). From a woman's point of view: The use of reasonable woman standard in sexual harassment cases. Cincinnati Law Review. 60(4). 1281-1306. Brewer, M. (1982). Further beyond nine to five: An integration and future directions. Journal of Social Issues. 38(4). 149-158. 138 Bularzik, M. (1978). Sexual harassment at the workplace: Historical notes. Radical America .12(4). 24-43. Bumiller, K. (1990). Fallen angels: The representation of violence against women in legal culture. Journal of the Sociology of Law. 18. 125-142. Burt, M. (1980). Cultural myths and supports for rape. Journal of Personality and Social Psychology. 38. 217-230. Calvert, S., Cocking, R. , & Smrcek, M. (1991). AIDS public service announcements: A paradigm for behavioral science. Journal of Applied Developmental Psychology. 12. 255-267. Cammaert, L. (1985). How widespread is sexual harassment on campus? International Journal of Women Studies. 8. 388-397. Clarke, P., & Kline F. (1974). Media effects reconsidered: Some new strategies for communication research. Communication Research. 1. 224-240. Cohen, J. , & Cohen, P. (1983). Applied multiple regression/correlation analysis for the behavioral sciences. (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum Associates, Publishers. Collins, E., & Blodgett, T. (1981). Sexual harassment... some see it... some won't. Harvard Business Review. 59. 76-95. Demarest. K., & Garner, J. (1992). The representation of women's roles in women's magazines over the past 30 years. The Journal of Psychology. 126(4). 357-369. de Vaus, D. (1990). Surveys in social research (2nd ed.). London: Unwin Hyman Dzeich, B. , & Weiner, L. (1984). The lecherous professor: Sexual harassment on campus. Boston: Beacon Press. Equal Employment Opportunity Commission (1980). Guidelines on discrimination because of sex. Federal Register. 45(219), 74676 -74677. 139 Erbring, L. , Goldenberg, E. , & Miller, A. (1980). Front-page news and real-world cues: A new look at agenda-setting by the media. American Journal of Political Science. 24(1). 16-49. Ettema, J. , & Kline, F. (1977). Deficits, differences, and ceilings: Contingent conditions for understanding the knowledge gap. Communication Research. 4,179-202. Faludi, S. (1991). Backlash: The undeclared war against American women. New York: Crown. Farley, L. (1978). Sexual shakedown: The sexual harassment of woman on the iob. New York: McGraw-Hill. Fisher, G. (1989). Mass media effects on sex role attitudes of incarcerated men. Se£Ro!gs,20, 191-203. Fitzgerald, L , Shullman, S., Bailey, N., Richards, M., Swecker, J., Gold, Y. , Ormerod, M., & Weitzman, L. (1988). The incidence and dimensions of sexual harassment in academia and the workplace. Journal of Vocational Behavior. 22,152-175. Gerber, G. , Gross, L., Morgan, M., & Signorielli, N. (1980). The "mainstreaming" of America: Violence profile no. 11. Journal of Communications. 30(3). 10-29. Glazer, N. (1980). Overworking the working woman: The double day in a mass magazine. Women's Studies International Quarterly. 3_, 79-93. Gutek, B. (1985). Sex and the workplace: Impact of sexual behavior and harassment on women, men and organizations. San Francisco: Jossey-Bass. Gutek, B., & Morasch, B. (1982). Sex-ratios, sex-role spillover, and sexual harassment of women at work. Journal of Social Issues. 3JL(4), 55-74. Gutek, B. , Morasch, B. , & Cohen, A. (1983). Interpreting social - sexual behavior in a work setting. Journal of Vocational Behavior. 22. 30-48. 140 Gutek, B., Nakamura, C., Gahart, M., Handschumacher, I., & Russell, D. (1980). Sexuality and the workplace. Basic and Applied Social Psychology. 1. 255-265. Gruber, J. (1990). Methodological problems and policy implications in sexual harassment research. Population Research and Policy Review. 9_, 235-254. Gruber, J. (1992). A typology of personal and environmental sexual harassment: Research and policy implications for the 1990s. Sex Roles. 26. 447-464. Hans, V. , & Dee, J. (1991). Media coverage of law: Its Impact on juries and public. American Behavioral Scientist. 35(2). 136-149. Hemming, H. (1985). Women in a man's world: Sexual harassment. Human Relations. 38(1). 67-69. Hill, D. (1985). Viewer characteristics and agenda setting by television news. Public Opinion Quarterly. 49. 340-350. Hubbard, J. , Defleur, M., & Defleur, L. (1975). Mass media influences on public conceptions of social problems. Social Problems. 23. 22-34. Jaschik, M. , & Fretz, B. (1991). Women's perceptions and labeling of sexual harassment. Sex Roles. 25. 19-23. Jensen, I. , & Gutek, B. (1982). Attributions and assignment of responsibility in sexual harassment. Journal of Social Issues. 38(4). 121-136. Kanter, R. (1977). Some effects of proportions on group life: Skewed sex ratios and responses to token women. American Journal of Sociology. £2(5), 965-990. Klassen, M. , Jasper, C., & Schwartz, A. (1993). Men and women: Images of their relationships in magazine advertisements, Journal of Advertising Research. 3J, 30-39. Konrad, A., & Gutek, B. (1986). Impact of work experiences on attitudes toward sexaul harassment. Administrative Science Quarterly. 31. 422-438. 141 LeBlanc, A. (1993). Examining HlV-related knowledge among adults in the U. S. Tournal of Health and Social Behavior. 34, 23-36. Leff, D. , Protess, D. , & Brooks, S. (1986). Crusading journalism: Changing public attitudes and policymaking agendas. Public Opinion Quarterly. 50. 300-315. Lill, D. , Gross, C. , & Peterson, R. (1986). The inclusion of social-responsibility themes by magazine advertisers: A longitudinal study. lournal of Advertising. 15(2). 35-41. Lott, B., Reilly, M., & Howard, D. (1982). Sexual assault and harassment: A campus community case study. Signs: Tournal of Women in Culture and Society, & 296 -319. Lysonski, S. (1985). Role portrayals in British magazine advertisements. European lournal of Marketing. 19(7). 37-55. MacKinnon, C. (1979). Sexual harassment of working women: A case of sex discrimination. London: Yale University Press. Mazer, D. , & Percival, E. (1989). Ideology or experience? The relationships among perceptions, attitudes, and experiences of harassment in university students. Sex Roles. 20.135-147. Pearson, R. , Ross, M., & Dawes, R. (1992). Personal recall and the limits of retrospective questions in surveys. In J. Tanur (Ed.), Questions about questions: Inquiries into the cognitive bases of surveys (pp. 65-94). New York: Russell Sage Foundation. Pedhazur, E. (1982). Multiple regression in behavioral research: Explanation and prediction (2nd ed. ). New York: CBS College Publishing. Peitchinis, S. (1989). Women at work: Discrimination and response. Toronto: McClelland & Stewart. 142 Petty, R., & Cacioppo, J. (1984). The effects of involvement on responses to argument quantity and quality: Central and peripheral routes to persuasion. Journal of Personality and Social Psychology. 46. 69-81. Pollay, R. (1986). The distorted mirror: Reflections on the unintended consequences of advertising. Journal of Marketing. 50(2). 18-36. Potter, W. (1988). Perceived reality in television effects research. Journal of Broadcasting & Electronic Media. 32, 23-41. Powell, G. (1986). Effects of sex role identity and sex on definitions of sexual harassment. Sex Roles. 14. 9-19. Prentice, A. , Bourne, P. , Brandt, G. , Light, B. , Mitchinson, W. , & Black, N. (1988). Canadian women: A history. Toronto: Harcourt Brace Jovanovich Canada Inc. Pryor, J. (1987). Sexual harassment proclivities in men. Sex Roles. 17. 269-290. Reese, S. , & Miller, M. (1981). Political attitude holding and structure: The effects of newspaper and television news. Communication Research. 2, 3-23. Riger, S. (1991). Gender dilemmas in sexual harassment policies and procedures. American Psychologist. 46. 497-505. Reilly, M., Lott, B. , & Gallogly, S. (1986). Sexual harassment of university students. Sex Roles. 15. 333-358. Renn, J . , & Calvert, S. (1993). The relation between gender schemas and adults' recall of stereotyped and counterstereotyped televised information. Sex Roles. 28. 449-459. Roberts, D., & Bachen, C. (1981). Mass communication effects. Annual Review of Psychology. 32. 307-356. Ruggiero, J., & Weston, L. (1985). Work options for women in women's magazines: The medium and the message. Sex Roles. 12, 535-547. Saal, F., Johnson, C. , & Weber, N. (1989). Friendly or sexy? It may 143 depend on whom you ask. Psychology of Women Quarterly. 13. 263-276. Samuelson, M., Carter, R., & Ruggels, L. (1963). Education, available time, and use of mass media. Journalism Quarterly. 40. 491-493. Sandler, B. (1990). Sexual harassment: A new issue for institutions. Initiatives. 51(4), 5-10. Sandroff, R. (1988, Dec). Sexual harassment in the fortune 500. Working Women. p. 69-73. Sandroff, R. (1992, June). Sexual harassment: The inside story. Working Women. p. 47-51. Schneider, B. (1982). Consciousness about sexual harassment among heterosexual and lesbian women workers. Journal of Social Issues. 3J£(4), 75-98. Shotland, R. & Craig, J. (1988). Can men and women differentiate between friendly and sexually interested behavior? Social Psychology Quarterly. 51(1), 66-73. Signorielli, N. (1989). Television and conceptions about sex roles: Mainstreaming and conventionality and the status quo. Sex Roles. 21.341-359. Skelly, G. , & Lundstrom, W. (1981). Male sex roles in magazine advertising, 1959-1979. Journal of Communication. 31(4). 52-57. Slater, D. , & Elliot, W. (1982). Television's influence on social reality. Quarterly Tournal of Speech. 68, 69-79. Stark, S. (1992, May). Sexual harassment in the workplace: Lessons from the Thomas Hill hearings. Trial. 116-122. Statistics Canada (1990). Women in Canada: A statistical report (2nd ed . ) . Ottawa, Ontario: Ministry of Supply and Services. Sullivan, G. , & O'Connor, P. (1988). Women's role portrayals in magazine advertising: 1958-1983. Sex Roles. 18. 181-188. 144 Swim, J. , & Borgida, E. (1987). Public opinion on the psychological and legal aspects of televising rape trials. Journal of Applied Social Psychology. 17(5). 507-518. Tabachnick, B., & Fidell, L. (1989). Using multivariate statistics (2nd ed.). New York: Harper Collins. Tangri, S., Burt, M. , & Johnson, L. (1982). Sexual harassment at work: Three explanatory models. Journal of Social Issues. 38(4). 33-54. Terpestra, D. , & Baker, D. (1986). A framework for the study of sexual harassment. Basic and Applied Social Psychology. 7(1). 17-34. Terpestra, D., & Baker, C. (1989). The identification and classification of reactions to sexual harassment. Journal of Organizational Behavior. 10. 1-14. Thompson, J. (1988). Mass communication and modern culture: Contribution to a critical theory of ideology. Sociology. 22(3). 359-383. Tichenor, P., Donohoe, G., & Olien, C. (1970). Mass media flow and differential growth in knowledge. Pnhir opinion quarterly. 34. 158-170. Till, F. (1980). Sexual harassment: A report on the sexual harassment of students. Washington, DC: National Advisory Council on Women's Educational Programs. U. S. Merit Systems Protection Board (1981). Sexual harassment in the federal workplace: Is it a problem? Washington D. C: U.S. Government Printing Office. Villemez, W., & Touhey, J. (1977). A measure of individual differences in sex stereotyping and sex discrimination: The "Macho" scale. Psychological Reports. 41. 411-415. Volgy, T. , & Schwarz, J. (1980). Television entertainment programming and sociopolitical attitudes. Journalism Quarterly. 57(1). 150-155. 145 Sandler, B. (1990). Sexual harassment: A new issue for institutions. Initiatives. 52(4), 5-10. Wade, S., & Schramm, W. (1969). The mass media as sources of public affairs, science, and health knowledge. Public Opinion. 33. 197-209. Walker, G. , Erickson, L. , & Woolsey, L. (1985). Sexual harassment: Ethical research and clinical implications in the academic setting. International Journal of Women's Studies. 8(4). 424-433. Wampold, B. , & Freund, R. (1987). Use of multiple regression in counseling psychology research: A flexible data-analytic strategy. Journal of Counseling Psychology. 34. 372-382. Weathers, P. , Furlong, M., & Solorzano, D. (1993). Mail survey research in counseling psychology: Current practice and suggested guidelines. Journal of Counseling Psychology. 40. 238-244. Wilson, B., Linz, D., Donnerstein, E., & Stipp, H. (1992). The impact of social issue television programming on attitudes toward rape. Human Communication Research. I9_ (2), 179-208. Wolf, N. (1991). The beauty mvth. Toronto: Vintage Books. Yankelovich Clancy Shulman (1991a, June 3). Do you believe a woman who is raped is partly to blame? Time, p. 51. T H E U N I V E R S I T Y O F B R I T I S H C O L U M B I A 140 APPENDIX A: Introduction Letter and Survey Department of Counselling Psychology Faculty of Education 5"?80 Toronto Road Vancouver. B.C. Canada V6T IL2 Tel: (604) 822-5259 Fax: (604) S22-232S Nov. 26, 1993 Dear Sir or Madam, I am a graduate student at the University of British Columbia and I am conducting a study on sexual harassment in the workplace in cooperation with your union, BCGEU. The study is entitled, "The relationship of television viewing and demographic variables with knowledge and attitudes about sexual harassment". Although this study is a research requirement for my Master's degree, it has the additional purpose of aiding your union in assessing how union members view sexual harassment in the work place. Five hundred members of your union have been randomly selected by computer as potential participants in this study. Although your name was one of those randomly selected, your participation is completely voluntary and I will assume that you have given consent to participate if you do complete the survey. I have sent you a survey that takes approximately 15-20 minutes to do. It asks questions on your understanding of sexual harassment; how you obtained knowledge about it; and your experiences with it. There is no means of identifying you as a participant since your name or where you work is not required on the survey. You will notice, however, that there is a number on the back page of the questionnaire. This number will enable me to identify participants who have completed the survey so that I will not need to recontact them. The number coding of each survey has been done in such a way that I do not have access to your name or address and your union does not have access to your completed questionaire. In this way I have protected your confidentiality and anonymity. If you choose not to participate, this will have no effect on your standing in the union because you cannot be identified, but it is very important that both men and women who were selected complete the survey. I need everyone's view, including people who aren't interested in sexual harassment as well as people who know very little about it. I have enclosed a stamped addressed return envelope in which I would like you to send back your completed survey. If you decide to participate please return the survey within 10 days. It is certainly your choice whether you participate in the study and I hope you do because the information gained from this study may be very useful in finding ways of improving your fellow workers' environment and working conditions. If you should have any questions about this study, please contact my supervisor, Dr. Richard Young (822-6380) or me through the Counselling Psychology Department of U. B. C. (822-5259). Thank you for your interest and time. Sincerely, Lorri Johnston 147 The University of British Columbia Department of Counselling Psychology SEXUAL HARASSMENT: KNOWLEDGE AND ATTITUDE SURVEY Section 1: People feel very differently about what should or should not go on at work. I would like your opinion about different kinds of sexual behavior that can happen at work. How much do you agree or disagree with each statement below? For each statement, X the box which best describes your opinion. strongly no strongly disagree disagree opinion agree agree a. All this concern about sexual harassment makes it harder for men and women to have normal relationships. b. I believe that sexual intimidation is a serious social problem. c. It is only natural for a man to make sexual advances to a woman he finds attractive. d. Many charges of sexual harassment are frivolous or vindictive. e. A lot of what people call sexual harassment is just normal flirtation between men and women. f. Most women who are sexually insulted by a man provoke his behavior by the way they talk, act, or dress. g. An attractive woman has to expect sexual advances and should learn how to handle them. h. One of the problems with sexual harassment is that some women can't take a joke. i. It is only natural for a woman to use her sexuality as a way of getting ahead in school or at work. Section 2: In your opinion, which of the following behaviors are sexual harassment? Answer each item with an X strongly no strongly disagree disagree opinion agree agree a. An employee hears co-workers (of the opposite sex) talking about him/her in a highly sexualized manner in the immediate work area. O O O D D b. The presence of pornographic pictures or posters, or sexual grafitti in the workplace. D • O O O c. An employee discovers that he/she is the object of rumors or gossip among co-workers who have made remarks about his/her sexuality or appearance. O 0 0 0 0 d. An employee is sexually assaulted or raped at work. O O 0 0 0 e. An employee is asked to go for a drink after work by a fellow worker. O O 0 0 0 0 0 0 0 0 0 0 0 0 0 a a 0 a a a a a 0 a 0 a 0 0 0 0 a 0 a a 0 0 0 0 0 a 0 0 a a 0 a 0 0 a 148 Section 2: In your opinion, which of the following behaviors is sexual harassment? Answer each item. strongly no strongly disagree disagree opinion agree agree f. An employee is promised a promotion in exchange for sexual favors with: a) a supervisor O D O O O b) a co-worker O O 0 0 0 g. An employee is threatened with loss of job for refusing to have sex with: a) a supervisor O O 0 0 0 b) a co-worker O O 0 0 0 h. Important information or job training is withheld from an employee because of refusal to have sex with the employee's: a) supervisor O O 0 0 0 b) co-worker O O 0 0 0 i. An employee fears job consequences for refusing to date: a) a supervisor O O 0 0 0 b) a co-worker O 0 0 0 0 j . An employee receives uninvited letters, telephone calls, or materials of a sexual nature from: a) a supervisor 0 O 0 0 0 b) a co-worker O O 0 0 0 k. Important information or job training is withheld from an employee because of refusal to date: a) a supervisor 0 0 0 0 0 b) a co-worker O O 0 0 0 I. An employee is given a poor work evaluation by his/her supervisor. O O 0 0 0 m. An employee receives uninvited and deliberate touching, leaning over, cornering, or pinching by: a) a supervisor 0 O 0 0 0 b) a co-worker O 0 0 0 0 n. An employee is complimented on how good his/her work is by the supervisor. 0 O 0 0 0 o. An employee is repeatedly "badgered" or pressured for a date by: a) the employee's supervisor O O 0 0 0 b) a co-worker 0 0 0 0 0 p. An employee receives sexually suggestive looks, stares, or gestures from a: a) supervisor 0 O 0 0 0 b) co-worker 0 0 0 0 0 q. An employee receives unwanted small gifts over several months by a: a) supervisor O O 0 0 0 b) co-worker O 0 0 0 0 r. An employee is given a poor reference or poor work evaluation because of refusal to: a) date a supervisor O O 0 0 0 b) have sex with a supervisor O O 0 0 0 s. Workers often tease one of their peers about his/her sex life. O O 0 0 0 149 Section 3: To what extent have you ever experienced any of the following behaviors at work or behaviors related to your job in the past two years? Mark one. box for each item with an X a. Co-workers talked about you in a highly sexualized manner in your immediate work area. O b. The presence of pornographic pictures or posters, or sexual grafitti in the workplace. O c. Was the object of rumors or gossip among co-workers who made remarks about your sexuality or appearance. D d. Sexually assaulted or raped at work. O e. Asked to go for a drink after work by a fellow employee. Q f. Promised a promotion in exchange for sexual favors with a supervisor or co-worker. D g. Threatened with loss of job for refusal to have sex with a supervisor or co-worker. D h. Important information or job training was withheld because of refusal to have sex with a co-worker or supervisor. O i. Feared job consequences for refusing to date a supervisor or co-worker. D j . Received uninvited letters, telephone calls, or materials of a sexual nature from a supervisor or co-worker. O k. Important information or job training was withheld because of refusal to date a supervisor or co-worker. O I. Given a poor work evaluation by your supervisor. D m. Received uninvited and deliberate touching, leaning oyer, cornering, or pinching by a supervisor or co-worker. O n. Complimented on your work by a supervisor. O o. Repeatedly "badgered'' or pressured for a date by a supervisor or co-worker. O p. Received sexually suggestive looks, stares, or gestures from a supervisor or co-worker. • q. Received unwanted small gifts over several months by a co-worker or supervisor. • r. Was given a poor reference or poor work evaluation because of refusal to a) date a supervisor O b) have sex with a supervisor Q once sometimes often very often • a a a a a a a a a a a o a a a a a a a a a a o a a a a a a a a • D D a • D a a a a a a a a a a a o a a a a a a a a a o a a a a a s. Workers teased you about your sex life. O 150 Section 4: The following section asks questions about how you acquired knowledge of sexual harassment In your opinion, how much knowledge of sexual harassment have you gained from the following sources? Mark each category with an X nans at aH vary little some quite a lot a great deal a. Newspaper b. Radio c. Television d. Magazines e. Friends, relatives, or co-workers f. Information from BCGEU (e. g. pamphlets, workshops, etc.) g. Other (please specify) • a a a a a a a a a a D a a a D D a D a a a a D a a a a a 0 a a a a • Section 5: Please indicate if you watched the sexual harassment episode on any of the following programs. I saw on TV Please add any other TV shows dealing with sexual harassment that you have seen. a. Herman's Head b. Designing Women c. Nurses d. Reasonable Doubts e. Golden Palace f. Murphy Brown g. Sex, Power & the Workplace h. The Young and the Restless i. Clarence Thomas Senate Hearings j . 60 Minutes k. A sexual harassment commercial 1. Grace Under Fire m. Catwalk n. TV movie - Business as Usual y « a a a a a a a a a a a a 0 a no a a a D a a a u U u a a • • 1 don't know a a a a a a a a a a a a a a If you did not watch any TV programs dealing with sexual harassment, move to question #7 Section 6: Since viewing any of the programs listed (or programs you specified) in section 5, did your views change in regard to sexual harassment? Answer each statement with an X strongly disagree) dteegree undeoded agree a. It affected my views on sexual harassment. b. I did not know what sexual harassment was before I saw this program(s) but I have a better idea now. c. It reinforced my views on sexual harassment. d. I sought out more information about sexual harassment after seeing the program(s). e. It made me realize that I have experienced sexual harassment in my job. strongly a a a a a a a • a • 151 Section 6: Since viewing any of the programs listed (or programs you specified) in section 5, did your views change in regard to sexual harassment? Answer each statement with an X. f. I would be more likely to report sexual harassment in the future if it happens to me. g. I learned that I wasn't alone; other women (men) have also experienced this problem. h. It made me realize how easy it is to discredit the victim, i. It made me realize how differently people interpret sexual harassment, j . I became more aware of sexual harassment. k. I no longer consider sexual harassment as seriously as I once did. This section is very important for the statistical analysis of the survey. This information is used in order to compare views of people of different ages, education level, ethnic backgrounds, etc. and is one of the most important parts of the survey. Remember that your answers are completely confidential and anonymous. If you do not wish to respond to a particular question, this is your right to do so. 7. Your age is 8. Your gender male E female E 9. What is your marital status? single ^ remarried married E divorced 10. Highest level of education completed? Less than high school E High school graduation B Technical training or apprenticeship B 11. What is your household income? less than 20,000 a 20,000 - 29,999 n 30,000 - 39,999 a 40,000 - 49,999 a 12. With what ethnic group do you most closely identify? ( e. g. Chinese; Japanese; African; East Indian; Western European; Native Indian; etc.) 13. What is your occupation or job? 14. What is your present work status? unemployed E welfare n casual B part time n full time ^ 15. What is your BCGEU component? Feel free to make any comments on the back of this page. Thank you for participating. strongly stn>ngly disagree disagree undecided agree agree D D D D D D • D D D • • D • D D • • D D a a • D D • D • D • widowed ^ common-law E separated D Some college or university a University graduate B Advanced degree (Master's; PhD) B 50,000 - 59,999 60,000 - 69,999 70,000 - 79,999 more than 80,000 T H E U N I V E R S I T Y O F B R I T I S H C O L U M B I A APPENDIX B: Follow Up Letters Department of Counselling Psychology Faculty of Education 5780 Toronto Road Vancouver, B.C. Canada V6T 1L2 Tel: (604) 822-5259 Fax: (604) 822-2328 Jan. 3, 1994 Dear Sir or Madam, Recently I sent you a survey on sexual harassment in the work place. As I mentioned in my first letter, your name was drawn through a scientific sampling procedure in which every fellow member of BCGEU had an approximately equal chance of being selected. I am writing to you again because of the significance each person's response has to the usefulness of this study. If you have already completed and returned the survey, I would like to thank you for participating. I have enclosed another survey in case you didn't complete the questionnaire and would like to do so but no longer have the original letter and survey that I sent you approximately five weeks ago. I would like to mention again that your participation is completely voluntary and your confidentiality and anonymity is totally assured if you choose to answer the survey. The questionnaire will take approximately 15-20 minutes to complete. If you choose to complete the survey, please complete it within 10 days and mail it in the stamped addressed envelope which I have provided. Do not complete this survey if you have already mailed the first one. Once again I would like to thank you for your interest and time. If you have any questions about this study, please contact my supervisor, Dr. Richard Young (822 -6380) or me through the Counselhng Psychology Department of U.B.C. (822-5259). Sincerely, Lorri Johnston T H E U N I V E R S I T Y O F B R I T I S H C O L U M B I A 153 Department of Counselling Psychology Faculty of Education 5"S0 Toronto Road Vancouver. B.C. Canada V6T 1L2 Tel: (604) 822-5259 Fax:(604) 822-2328 Jan. 26, 1994 Dear Sir or Madam, Recently I sent you a survey on sexual harassment in the work place. As I mentioned in my first letter, your name was drawn through a scientific sampling procedure in which every fellow member of BCGEU had an approximately equal chance of being selected. I am writing to you again because of the significance each person's response has to the usefulness of this study. If you have already completed and returned the survey, I would like to thank you for participating. I have enclosed another survey in case you didn't complete the questionnaire and would like to do so but no longer have the original letter and survey that I sent you approximately eight weeks ago. I would like to mention again that your participation is completely voluntary and your confidentiality and anonymity is totally assured if you choose to answer the survey. The questionnaire will take approximately 15-20 minutes to complete. If you choose to complete the survey, please complete it within 10 days and mail it in the stamped addressed envelope which I have provided. Do not complete this survey if you have already done the survey and sent it back. Once again I would like to thank you for your interest and time. If you have any questions about this study, please contact my supervisor, Dr. Richard Young (822 -6380) or me through the Counselling Psychology Department of U. B.C. (822-5259). Sinrerelv. Lorri Johnston 154 APPENDIX C SOURCE OF KNOWLEDGE: OTHER This appendix contains categories listed by subjects constituting the "other" category of sources of knowledge. Subjects responded to "In your opinion, how much knowledge of sexual harassment have you gained from the following sources?" Subjects were asked to identify "other." "Other" sources of information Frequency School 9 Work 15 Personal experience 6 Seminars/workshops 8 Government information 4 Women's Resource Centre 1 Books 2 Women's Action Committee Handouts 1 Spouse 1 Support group through mail 1 Political party 1 Women's Shelter 1 Court proceedings 1 Women's issues 1 Family values 1 Common Sense 1 Not specified 6 n = 60 APPENDIX D Table D-l Ethnicity Responses Exactly as Subjects Answered Ethnicity Response Frequency Western European Canadian WASP French Canadian English Scottish Scottish/French North American white white Caucasian North American (Britsh descent) Finish Dutch Eastern European Russian Ukranian Native Indian Chinese Chinese/East Indian Canadian English/Chinese South East Asian Japanese East Indian Indo - Canadian African Latin American Third Nation Pict No Response other none - - possibly Japanese any none all N/A Canadian and all ethnic groups 90 48 5 2 5 3 2 1 1 2 1 1 2 6 1 1 12 4 1 1 1 2 2 1 4 1 1 1 33 1 1 1 1 5 3 1 Total - 248 156 APPENDIX E Description of Missing Data by Section Section l(Attitude): There were 12 subjects who were missing items from this section. Nine of these subjects were missing 1 item. The three remaining subjects were missing 2, 3, and 5 items respectively. Section 2 (Knowledge): There were 16 subjects who were missing 1-2 items and three subjects who were missing the entire knowledge section. Section 3 (Experience): There were 10 subjects missing one item and one subject missing 2 items. Section 4 (Source of knowledge): There were 16 subjects missing one item; 3 subjects missing 2 items; one subject missing 3 items; one subject missing 5 items; and one subject missing the entire section. Section 6 (Change of view): Twenty subjects watched TV programs but did not do this section. There were 19 subjects that were affected by the typing error (scores were replaced with subject's mean score). There were 12 subjects with missing items. There was one subject who did not identify any TV programs watched, but answered section 6 (subject eliminated from regression analysis). Missing Demographics; Income (12); Education (2); SES (2); Work status (1); Component (1) All demographics missing (4) Subjects Dropped from Analysis : There were a total of 248 completed surveys. Nine subjects were dropped from analysis because subjects were missing all demographic information or were missing entire sections of the dependent variables or missing more than half the items in more than one section of either a dependent or independent variable. 157 APPENDIX F Table F-l Survey Return Rates by Mailout Total Completed Returns Mailout Female Male Group n % n % n % First mailout 85 28.9 56 27.2 141 28.2 (N = 500) Second mailout 45 21.5 20 13.4 65 18.0 Third mailout 25 16.2 15 11.2 41 14.0 Total 155 52.7 92 44.7 247 49.4 Note. First mailout was to 290 (58%) females and 210 (42%) males. Four subjects identified themselves as female that were listed by their union as male. Mailout and return rates were adjusted for this. Adjusted rate for mailout was 294 (58.8%) females and 206 (41.2%) males. There was one subject who completed a survey who could not be identified by gender. There were 18 surveys sent back because of address change. There were 7 surveys sent back uncompleted and 247 completed surveys for a total of 254 or a 50.8% return rate. 158 Appendix G Table G-l Perceived Change of View Means for Women and Men with Typing Errors Scored as Strongly Agree Change of View Statements Women Men Mean SD Mean SD a. It affected my views on sexual harassment. b. I did not know what sexual harassment was before I saw this program(s) but I have a better idea now. c. It reinforced my views on sexual harassment d. 1 sought out more information about sexual harassment after seeing the program(s). e. It made me realize that I have experienced sexual harassment in my job. f. I would be more likely to report sexual harassment in the future if it happens to me. g. I learned that I wasn't alone; other women (men) have also experienced this problem. h.It made me realize how easy it is to discredit the victim, i. It made me realize how differently people interpret sexual harassment, j . I became more aware of sexual harassment 3.03 1.07 2.85 1.06 2.05 0.93 1.88 0.93 3.66 0.83 3.18*1.03 2.26 0.89 2.06 0.83 2.20 1.08 2.06 0.92 3.65 0.92 2.88*0.98 3.03 0.97 2.61 1.02 3.71 0.94 3.43 1.04 4.04 0.74 3.85 0.97 3.52 0.99 3.34 1.05 Note, n ranged from 66 to 68 for men and 91 to 97 for women. * Mean scores of men and women are significantly different, rj < .005. 159 Appendix H Table H-l Descriptive Statistics of Respondents with the Highest Total Experience Sums Case # Experience Age Gender Educ Marital SES Component Sum Status _ _ 34 85 52 35 32 159 20 5 45 63 91 165 135 133 212 103 94 192 219 176 191 185 182 Note. Education levels: (2) high school graduation; (3) technical training or apprenticeship; (4) some college or uninversity; (5) university graduate; (6) advanced degree. Note. Mean age of women (n =13) was 34.5 (SD 8.%) and mean age of men (a =11) was 34.2 (§E=7.55). 17 17 17 15 14 13 12 10 9 9 9 9 9 18 17 16 15 13 13 13 12 12 10 9 20 30 32 49 26 29 42 37 26 34 47 31 45 35 23 24 37 35 35 36 39 24 46 42 F F F F F F F F F F F F F M M M M M M M M M M M 4 6 5 2 4 4 4 4 4 6 6 5 4 4 4 4 2 3 5 5 4 4 4 2 single divorced single common married single single common divorced divorced married single divorced married single single single common married married separated married remarried separated 36.89 57.55 49.87 25.56 41.82 39.01 39.01 41.82 44.39 60.11 59.94 60.11 39.01 52.86 48.72 52.86 48.72 52.86 60.11 54.05 38.15 60.73 59.94 25.90 02 12 11 02 12 12 12 12 06 06 12 06 12 09 06 09 11 09 06 09 12 07 12 02 160 Appendix H Table H-2 Experiences of sexual harassment for all survey respondents in terms of frequency for each i tem of section 3 Experience Statement Categories Never Once Sometimes Often Very often % % % % % a. Co-workers talked about you in a highly sexualized manner in your immediate work area.... 76.6 b. The presence of pornographic pictures or posters, or sexual grafitti in the workplace 60.5 c. Was the object of rumors or gossip among co-workers who made remarks about your sexuality or appearance 80.6 d. Sexually assaulted or raped at work 98.7 f. Promised a promotion in exchange for sexual favors with a supervisor or co-worker 99.2 g. Threatened with loss of job for refusal to have sex with a supervisor or co-worker 99.2 h. Important information or job training was withheld because of refusal to have sex with a a co-worker or supervisor 100.0 i. Feared job consequences for refusing to date a supervisor or co-worker 99.2 j . Received uninvited letters, telephone calls, or materials of a sexual nature from a supervisor or co-worker 93.3 5.1 14.3 7.1 24.4 4.6 13.1 0.8 0.8 0.8 0.0 0.8 2.9 0.0 0.0 0.0 0.0 0.0 2.9 2.5 0.8 4.2 3.8 1.3 0.4 0.4 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.4 161 Table H-2 (continued) Experiences of sexual harasssment for all survey respondents in terms of frequency for each item of section 3 Experience Statement Categories Never Once Sometimes Often Very often % % % % % k. Important information or job training was withheld because of refusal to date a supervisor or co-worker m. Received uninvited and deliberate touching, leaning over, cornering, or pinching by a supervisor or co-worker...... o. Repeatedly "badgered" or pressured for a date by a supervisor or co-worker p. Received sexually suggestive looks, stares, or gestures from a supervisor or co-worker q. Received unwanted small gifts over several months by a supervisor or co-worker r. Was given a poor reference or poor work evaluation because of refusal to date a supervisor s. Was given a poor reference or poor work evaluation because of refusal to have sex with a supervisor t. Workers teased you about your sex life 99.6 0.4 96.7 2.5 73.2 95.0 98.7 97.9 62.3 6.3 2.5 1.3 1.3 0.0 81.2 6.7 10.9 0.8 17.6 0.8 0.0 0.0 0.0 0.0 0.8 0.4 0.0 0.0 1.7 0.0 1.3 0.0 6.7 25.5 0.0 0.0 3.3 0.0 0.0 2.1 Note. Sample size for items f, g, h, i, k, m, o, r, t was 239; for items b, d, j , q was 238; for items a, c, s was 237; and for item p was 236. 162 Appendix H Table H-3 Experience of Being Sexually Harassed: Percentage of Males a n d Females Responding to Each Item by Category Experience Statement Frequency in % Never Once Sometimes Often Very often M F M F M F M F M F a. Co-workers talked about you in a highly sexualized manner in your immediate work area.... 64.4 85.0 5.6 4.8 22.2 9.5 5.6 0.7 2.2 b. The presence of pornogrpahic pictures or posters, or sexual grafitti in the workplace 45.6 69.6 8.9 6.1 33.3 18.9 6.7 2.7 5.6 2.7 c. Was the object of rumors or gossip among co-workers who made remarks about your sexuality or appearance 70.0 87.1 6.7 3.4 20.0 8.8 3.3 - - 0.7 d. Sexually assaulted or raped at work 100.0 98.0 - 1.4 - - - 0.7 f. Promised a promotion in exchange for sexual favors with a supervisor or co-worker 100.0 98.6 - 1 . 4 - - - -g. Threatened with loss of job for refusal to have sex with a supervisor or co-worker 100.0 98.6 - 1.4 - -h. Important information or job training was withheld because of refusal to have sex with a co-worker or supervisor. 100.0 100.0 - - - . _ . . . i. Feared job consequences for refusing to date a supervisor or co-worker. 98.9 99.3 1.1 0.7 -j . Received uninvited letters, telephone calls, or materials of a sexual nature from a supervisor or co-worker 91.2 95.2 3.3 2.7 4.4 2.0 1.1 163 Table H-3 (continued) Experience of being sexually harassed: Percentage of males a n d females responding to each item by category Frequency in % Experience Statement Never Once Sometimes Often Very often M F M F M F M F M F k. Important information or job training was withheld because of refusal to date a supervisor or co-worker 100.0 99.3 - 0.7 - - - - -m. Received uninvited and deliberate touching, leaning over, cornering, or pinching by a supervisor or co-worker..... 79.1 82.4 7.7 6.1 11.0 10.8 1.1 0.7 1.1 o. Repeatedly "badgered" or pressured for a date by a supervisor or co-worker. 100.0 94.6 - 4.1 - 1.4 -p. Received sexually suggestive looks, stares, or gestures from a supervisor or co-worker 74.4 74.0 6.7 6.2 15.6 19.2 3.3 0.7 - 0.7 q. Received unwanted small gifts over several months by a supervisor or co-worker 95.6 95.3 1.1 3.4 2.2 - 1.1 1.4 r. Was given a poor reference or poor work evaluation because of refusal to date a supervisor 98.9 98.6 1.1 1.4 s. Was given a poor reference or poor work evaluation because of refusal to have sex with a supervisor. 98.9 98.6 1 .11 .4 t. Workers teased you about your sex life 45.1 73.0 8.8 5.4 37.4 18.2 5.5 2.0 3.3 1.4 Items a, b, c, p, q, s (males n = 90); items d ,f, g, h, i, j , k, m, o ,r, t (males n = 91) Item p (female n = 1); a, c, d, j , s, (females n_= 147); b, f, g, h, i, k, m, o ,q ,r, t (female n = 148). 164 APPENDIX I Table 1-1 Sources of Knowledge: Mean a n d Standard Deviation Scores Sources of Knowledge Entire Group Women Men Mean SD n Mean SD n Mean SD n a. Newspapers 2.97 0.90 237 2.99 0.92 146 2.93 0.88 91 b. Radio 2.61 0.92 234 2.67 0.91 144 2.52 0.93 90 C. Television 3.15 0.98 236 3.21 0.99 145 3.06 0.97 91 d. Magazines 2.87 1.03 231 3.03 1.01 142 2.61*1.03 89 e. Friends 2.86 0.94 228 2.93 0.98 140 2.75 0.86 88 f. Union 2.83 1.10 228 2.76 1.08 140 2.93 1.13 88 * Scores for men and women significantly different, Q .006 165 APPENDIX I Table 1-2 Sources of Knowledge: Frequencies and Percentages of Women and Men Responding to Each Category Sources of Knowledge Perceived amount of knowledge from source none at all very little some n % n % n °A quite a lot a great deal n % n % a. Newspapers b. Radio c. Television d. Magazines e. Friends f. Union Total F M Total F M Total F M Total F M Total F M Total F M 13 7 6 30 15 15 15 9 6 28 13 15 15 10 5 36 23 13 5.5 4.8 6.6 12.8 10.4 16.7 6.4 6.2 6.6 12.1 9.2 16.9 6.6 7.1 5.7 15.8 16.4 14.8 49 33 16 67 42 25 35 19 16 45 21 24 61 31 30 40 25 15 20.7 22.6 17.6 28.6 29.2 27.8 14.8 13.1 17.6 195 14.8 27.0 26.8 22.1 34.1 175 17.9 17.0 119 68 51 103 65 38 105 63 42 98 65 33 105 69 36 93 63 30 50.2 46.6 56.0 44.0 45.1 42.2 43.9 43.4 46.2 42.4 45.8 37.1 46.1 493 40.9 40.8 45.0 34.1 44 30 14 31 19 12 62 41 21 49 34 15 35 19 16 46 21 25 18.6 205 15.4 13.2 13.2 133 263 283 23.1 21.2 23.9 16.9 15.4 13.6 18.2 20.2 15.0 28.4 12 8 4 3 3 -19 13 6 11 9 2 12 11 1 13 8 5 5.1 5.5 4.4 13 2.1 -8.1 9.0 6.6 4.8 63 2.2 53 7.9 1.1 5.7 5.7 5.7 Note. Total = entire group (n ranged from 228 to 237). Females (n ranged from 140 to 146); males (n ranged from 89 to 91). 166 APPENDIX J Table J-l Number of Television Programs Watched by Episode Number and percent of each episode seen Episodes with sexual harassment entire group women men Herman's Head Designing Women Nurses Reasonable Doubts Golden Palace Murphy Brown Sex, Power & the Workplace The Young and the Restless Clarence Thomas Senate Hearings 60 Minutes Sexual harassment commercial Grace Under Fire Catwalk Business as Usual (TV movie) Other n 19 42 16 14 5 54 9 22 104 78 88 33 1 8 27 % 8.0 17.6 6.7 5.9 2.1 22.6 3.8 9.2 433 32.6 36.8 13.8 0.4 3.3 113 n 9 30 9 10 2 31 7 18 61 48 53 24 0 3 15 % 8.0 26.8 8.0 8.9 1.8 27.7 6.3 16.1 54.5 42.9 47.3 21.4 0 2.7 13.4 n 10 12 7 4 3 23 2 4 43 30 35 9 1 5 12 % 13.9 16.7 9.7 5.6 4.2 253 2.8 5.6 58.9 41.7 48.6 123 1.1 6.9 16.7 Table J-2 Total Number of Programs Watched Number of programs watched Entire group women men a % a % a % 0 1 2 3 4 5 6 7 8 9 10 55 50 45 35 19 19 7 6 1 1 23.0 20.9 192 14.6 7.9 7.9 2.9 2.5 0.4 0.4 36 243 33 223 23 153 20 13.5 12 8.1 15 10.1 5 3.4 4 2.7 -_ 19 20.9 17 18.7 23 253 15 163 7 7.7 4 4.4 2 2.2 2 2.2 1 1.1 1 1.1 Mean number of programs watched 2.18 SD 1.95 2.16 1.95 2.19 1.95 167 APPENDIX K Table K-l Number and Percentage of Survey Respondents in Each Union Component Union Component Entire Group n % 10 4.1 7 2.9 Women n % 1 5 0.4 2.1 Men n % 9 3.7 2 0.8 Correctional Services (1) Hospital & Allied Services (2) Retail Stores & Warehouse (5) Social, Education & Health Services (6) 42 Education, Scientific, Technical & Administration (7) 12 Environment, Resources & Operations (9) Engineering, Technical & Inspection (11) Acuninistrative Service (12) Total 17.2 4.9 24 9.8 18 7.4 2.0 2.9 44 20 104 244 18.0 8.2 44.7 17 4 _2§ 154 7.0 1.6 40.2 27 16 11 90 11.0 6.6 4.5 Table K-2 Comparison of Survev Respondents to Union PoDulation bv Component Union Components Survey Sample % Union Population % Correctional Services (1) Hospital & Allied Services (2) Retail Stores & Warehouse (5) Social, Education & Health Services (6) Education, Scientific, Technical & Administration (7) Environment, Resources & Operations (9) Engineering, Technical & Inspection (11) Administrative Service (12) 4.2 2.5 -17.6 5.0 18.5 8.4 43.7 5.1 5.3 8.5 16.2 -10.5 9.3 43.0 Note. Component 5 was not sampled and component 7 was unexpectedly sampled. 168 APPENDIX K Table K-3 Comparison of Percentage of Subjects in the Respondent Sample to the Entire Union Population by Union Component. Adjusted for Inclusion Error Survey Union Union Components Sample Population Correctional Services (1) Hospital & Allied Services (2) Social, Education & Health Services (6) Education, Scientific, Technical & Administration (7) Environment, Resources & Operations (9) Engineering, Technical & Inspection (11) Administrative Service (12) % 4.2 2.5 17.6 5.0 18.5 8.4 43.7 % 4.9 5.0 15.4 14.9 10.0 8.8 40.9 169 Appendix L Table L-l Correlation Matrix for Attitude a n d Knowledge Regressions Attsum Knolsum Age Sex Expsum Educ SES TVProg Attsum Knolsum Age Sex Expsum Educ SES TVProg Attsum Knolsum Age Sex Expsum Educ SES TVProg Resol Reso2 Reso3 Reso4 Reso5 Reso6 -.484** .191** -.418** .136* -.110 .016 -.167* Attsum -.024 .044 -.020 -.119 -.192** .045 -.047 .391** -.159* .070 -.128* .009 -.122 -.258** -.133* -.022 -.001 Knolsum Resol .064 -.011 .083 .041 .145* -.083 -.024 .064 .142* .006 -.076 -.092 -.045 .089 .442** .410** .299** .098 -.030 -.244** -.012 -.160* .013 Reso2 .044 -.011 .045 .061 .031 -.087 -.014 -.007 .280** .232** .115 .070 .050 .010 -.010 Reso3 -.020 .083 -.032 .046 -.128* -.181** -.127 .283** .437** .121 .044 .405** -.067 Reso4 -.119 .041 .017 .170** -.125 -.090 -.163* .236** .108 .011 -.028 Reso5 -.192** .145* -.141* .056 .001 -.019 -.034 .159* .225** Reso6 .045 -.083 .064 -.082 -.060 -.203** .007 .102 -.030 N = 231 Note. Variables were designated by: attsum (total attitude sum); knolsum (total knowledge sum); age (age in years); sex (gender of respondent); expsum (total experience sum of being sexually harassed); Educ (education level); SES (socioeconomic status); and TVProg (total number of television programs watched). Sources of knowledge were designated by: resol (newspapers); reso2 (radio); reso3 (television); reso4 (magazines); resoS (friends); and reso6 (information from union). *E<.05. **£<.01. 170 Appendix L Table L-2 Correlation Matrix for Perceived Change of View Regression ViewlOm Age Sex Expsum Educ SES TVProg ViewlOm Age Sex Expsum Educ SES TVProg ViewlOm Age Sex Expsum Educ SES TVProg Resol Reso2 Reso3 Reso4 Reso5 Reso6 n= 161 Note. Variables were designated by: viewlOm (total perceived change of view mean); age (age of respondent); sex (gender of respondent); expsum (total experience sum of being sexually harassed); Educ (education level); SES (socioeconomic status); and TVProg (number of television programs watched). Sources of knowledge were designated by: resol (newspapers); reso2 (radio); reso3 (television); reso4 (magazines); resoS (friends); and reso6 (information from union). *E<.05. **JB_<.01. .032 .233** -.153 -.198* -.264** .146 -.122 -.258** -.133* -.022 -.001 ViewlOm Resol .008 .038 .179* .330** .064 .123 .008 .142* .006 -.076 -.092 -.045 .089 .442** .410** .299** .098 -.030 -.244** -.012 -.160* .013 Reso2 .038 .045 .061 .031 -.087 -.014 -.007 .280** .232** .115 .070 .050 -.010 -.010 Reso3 .179* -.032 .046 -.128* -.181** -.127 .283** .437** .121 .044 .405** -.067 Reso4 .330** .017 .170** -.125 -.090 -.163* .236** .108 .011 -.028 Reso5 .064 -.141* .056 .001 -.019 -.034 .159* .225** Reso6 .123 .064 -.082 -.060 -.203** .007 .102 -.030 171 APPENDIX M Table M-l Correlation comparisons of men and women Correlations Variables Attsum Knolsum ViewlOm F M F M F M Attsum Knolsum Age Expsum Educ SES TVProg Resol Reso2 Reso3 Reso4 Reso5 Reso6 .13 -.09 -.12 -.10 -.18* -.00 .03 .00 -.11 -.09 .02 .19 .16 -.14 -.02 -.19 -.05 .15 -.00 .02 -.35** .01 -.29** .18* -.13 .02 -.03 -.01 .17* .06 .13 .05 .09 -.16* -.44** -.20 -.04 .15 -.05 -.01 -.06 -.13 .00 -.14 .18 .06 .11 -.16 -.24* -.36** .24* .06 -.01 .17 .27** .08 .06 -.00 -.03 -.15 -.07 -.04 -.09 .05 .19 .34** .00 .28* Note. Variables were designated by: attsum (mean total attitude sum); knolsum (mean total knowledge sum); age (age in years); Expsum (total experience sum of being sexually harassed); Educ (education level); SES (socioeconomic status); and TVProg (total number of television programs watched. Sources of knowledge were designated by: resol (newspapers); reso2 (radio); reso3 (television); reso4 (magazines); resoS (friends); and reso6 (information from union). Correlations of independent variables with attsum and knolsum were from sample sizes of: females (ri = 143) and males (n = 87). Correlations of independent variables with perceived change of view mean were from sample sizes of: female (fl = 95) and males (n = 66). *£<.05 , **fi<.01. 172 APPENDIX N Table N-l Combined Percentage of Agreement and Disagreement with Attitude Statements of Section 1 for Men and Women. Disagree No Opinion Agree F M F M F M Attitude Statements % % % % % % a. All this concern about sexual harassment makes it harder for men and women to have normal relationships. 67.3 54.4 4.8 4.4 27.9 41.1 b. I believe that sexual intimidation is a serious social problem. 14.3 24.2 6.8 8.8 78.9 67.0 c. It is only natural for a man to make sexual advances to a woman he finds attractive. 87.2 73.0 6.8 7.9 6.1 19.1 d. Many charges of sexual harassment are frivolous or vindictive. 69.9 37.8 17.1 36.7 13.0 25.5 e. A lot of what people call sexual harassment is just normal flirtation between men and women. 73.6 42.2 10.1 23.3 16.3 34.4 f. Most women who are sexually insulted by a man provoke his behavior by the way they talk, act, or dress. 89.9 77.8 3.4 7.8 6.8 14.4 g. An attractive woman has to expect sexual advances and should learn how to handle them. 86.5 60.4 3.4 7.7 10.1 30.9 h. One of the problems with sexual harassment is that some women can't take a joke. 81.1 61.1 8.1 13.3 10.8 25.6 i. It is only natural for a woman to use her sexuality as a way of getting ahead in school or at work. 97.3 83.5 1.4 6.6 1.4 9.9 Note. "Strongly disagree" and "disagree" responses were collapsed into the disagree category. "Strongly agree" and "agree" responses were collapsed into the agree category. N for females ranged from 146 to 148; n for males ranged from 89 to 91. 

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.831.1-0054038/manifest

Comment

Related Items