Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The development and evaluation of a measure of proximal correlates of male domestic violence Starzomski, Andrew J. 1999

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

Item Metadata


831-ubc_1999-464318.pdf [ 9.01MB ]
JSON: 831-1.0099464.json
JSON-LD: 831-1.0099464-ld.json
RDF/XML (Pretty): 831-1.0099464-rdf.xml
RDF/JSON: 831-1.0099464-rdf.json
Turtle: 831-1.0099464-turtle.txt
N-Triples: 831-1.0099464-rdf-ntriples.txt
Original Record: 831-1.0099464-source.json
Full Text

Full Text

T H E D E V E L O P M E N T A N D E V A L U A T I O N OF A M E A S U R E OF PROXIMAL C O R R E L A T E S OF M A L E DOMESTIC V I O L E N C E by A N D R E W J. STARZOMSKI  B.Sc. (Honours) St. Mary's University (1991) M.A. The University of British Columbia (1993)  A THESIS SUBMITTED IN PARTIAL F U L F I L L M E N T OF T H E REQUIREMENTS FOR T H E D E G R E E OF D O C T O R OF PHILOSOPHY in T H E F A C U L T Y OF G R A D U A T E STUDIES (Department of Psychology) We accept this thesis as conforming Jo'the required standard  T H E UNIVERSITY OF BRITISH C O L U M B I A July 1999  ©Andrew J. Starzomski, 1999  In presenting degree  this  at the  thesis  in  partial fulfilment  of  University of  British Columbia,  I agree  freely available for reference copying  of  department publication  this or  and study.  of this  his  or  her  Department of The University of British Columbia Vancouver, Canada  Date  DE-6 (2/88)  that the  representatives.  may be It  thesis for financial gain shall not  permission.  requirements  I further agree  thesis for scholarly purposes by  the  is  an  advanced  Library shall make it  that permission for extensive granted  by the  understood be  for  head  that  allowed without  of  my  copying  or  my written  Development and Evaluation of the PCERQ  Abstract  This dissertation examined how psychological variables associated with selfcontrol related to abusiveness in situations of intimate conflict. The variables of interest were efficacy, need for power and responsibility. These variables were examined relative to other predictors of abuse such as the Abusive Personality (Dutton, 1994b), a construct of personality features that predispose some men to intensely aversive emotional arousal in their intimate relationship, leading to abusiveness. The research is relevant to the experience of those men with the characteristics of Abusive Personality, as well as those who may not have those predispositional features. The first step of the project was the development of the Power, Conflict Efficacy and Responsibility Questionnaire (PCERQ), with its four sub-scales: (1) Conflict Ineffectiveness (CI; lack of conflict efficacy), (2) N-Power (NP; need for power), (3) Standards of Non-Abusiveness (SNA; one part of responsibility), and (4) Exonerative Rationalizations (ER; cognitions complicit with inconsistent self-control - a second part of responsibility). These sub-scales were developed on the basis of data collected from samples of undergraduate males in dating relationships (n = 147), men in treatment groups for wife assault (n = 50), and a community sample of men (n = 27). Results from regression equations predicting self-reported abuse with the PCERQ sub-scales, along with other theoretically-relevant measures, found that CI was a prominent and consistent predictor of both verbal and physical abuse. The interaction of the NP and E R sub-scales significantly predicted physical abuse, as did the interaction of the CI sub-scale with the Abusive Personality (the most abusive participants had the highest scores on both Abusive Personality and Conflict Ineffectiveness). These results show the importance of considering both situational conflict experiences, along with personality and life history variables, when examining wife assault. ii.  Development and Evaluation of the PCERQ  TABLE OF CONTENTS  Abstract  ii  List of Tables  vi  List of Figures  vii  Dedication  viii  Acknowledgements  ix  Chapter I  1  Overview and Summary  Overview Literature Review: Existing Research on the Causes of Wife Assault Social Learning Theory Personality and Individual Differences Sociocultural and Feminist Models Relationship Interactions and Wife Assault Sociobiological Factors and Wife Assault On Extending the Merits of Existing Explanations  Chapter II  The Rationale for Investigating and Measuring Power, Conflict Efficacy and Responsibility 16 Conflict-Efficacy Need for Power Responsibility Summary of the PCERQ's Rationale, Categories and Themes  Chapter III  1 2 3 6 10 11 13 14  17 18 . 1 9 23  Intentions and Methods of the Research  25  Research Intentions and Hypotheses Anticipated Internal Properties Anticipated Validity Properties Method Participants  25 25 28 31 31  iii.  Development and Evaluation of the PCERQ  Materials Procedure Chapter IV  Chapter V  Chapter VI  32 36  Descriptive Characteristics of the Samples  .  .  .  .  40  Demographics Exposure to Conflict Tactics in Family of Origin ( F O O ) . . . . Conflict Tactics in Respondents' Ongoing Intimate Relationship . Links Between Abuse in Ongoing Relationship, SES and Relationship Satisfaction Synopsis  40 44 45 48 49  Derivation and Internal Properties of the PCERQ Sub-Scales  51  The Item Generation Process General Considerations and Methods of Sub-Scale Formation. . (I) Deriving the Conflict Ineffectiveness (CI) Sub-Scale (II) Deriving the Need for Power (NP) Sub-Scale (III) Deriving the Standards of Non-Abusiveness (SNA) Sub-Scale.. (IV) Deriving the Exonerative Rationalizations (ER) Sub-Scale.. (V) The Final Step of Sub-Scale Construction. An ObliquelyRotated Factor Analysis Relationships Among the PCERQ Sub-Scales Temporal Stability Summary of the Sub-Scale Derivation and Internal Properties  51 53 56 58 60 62  Concurrent Validity of the PCERQ Sub-Scales  74  64 65 70 72  The PCERQ in Relation to Demographic Variables 75 The PCERQ and the Family of Origin (Global Impressions and Abuse). 77 The PCERQ and Social Desirability . 78 The PCERQ and the SPQ 78 The PCERQ and the Abusive Personality 79 The PCERQ and Personality and Attitude Measures 79 The PCERQ and Narrative Measures 83 Summarizing the PCERQ's Concurrent Validity 84 Chapter VII  The Criterion Validity of the PCERQ  iv.  86  Development and Evaluation of the PCERQ  A Comprehensive Consideration of Social Desirability in Relation to the PCERQ-Abuse Analyses 86 Correlations Between the PCERQ and Abuse Across the Three Samples 90 Regression Analyses Predicting Abuse with Multiple Independent Measures 90 The PCERQ Sub-Scales in Isolation: Main Effects and Interaction Terms 93 On the Generation and Control and Aggressive Impulses: Predicting Abusive Behaviour by Integrating the PCERQ and PAS . . . . 96 Other Criterion Validity Considerations with the CI Sub-Scale . .105  Chapter VIII Discussion  108  Evaluation of Hypotheses and Interpretation of Major Findings . Methodological Issues and Limitations of the Current Research . The Findings in Light of the Existing Wife Assault Literature Avenues of Clinical Application Suggestions for Future Research Synopsis: What Has Been Accomplished Here? End Notes.  .  .  132  References. Appendix A (Descriptive Properties of the Samples)  108 118 .124 126 128 130  136 .  .  .  148  Appendix B (The PCERQ and Concurrent Validity)  154  Appendix C (The PCERQ and Criterion Validity)  158  Appendix D (Measures)  161  Appendix E (Consent and Debriefing Forms)  187  Development and Evaluation of the PCERQ  List of Tables  Table 1  Overview of Demographic Features by Sample  41  Table 2  Abuse in Ongoing Relationship by Sample  47  Table 3  Conflict Ineffectiveness (CI) Sub-scale: Item Content and Correlations with Four Obliquely-Rotated PCERQ Factors  66  Table 4  Table 5  Table 6  Need for Power (NP) Sub-Scale: Item Content and Correlations with Four Obliquely-Rotated PCERQ Factors .  .  .  67  Standard to be Non-Abusive (SNA) Sub-Scale: Item Content and Correlations with Four Obliquely-Rotated PCERQ Factors . . .  68  Exonerative Rationalizations (ER) Sub-Scale: Item Content and Correlations with Four Obliquely-Rotated PCERQ Factors .  .  .  69  Table 7  Intercorrelations Among PCERQ Sub-Scales  71  Table 8  Correlations of PCERQ Sub-Scales with Personality Measures  81  Summary of Regression Equation Predicting Verbal Abuse with BIDR and PCERQ  95  Summary of Regression Equation Predicting Physical Abuse with BIDR and PCERQ  97  Table 9  Table 10  Table 11  Table 12  Table 13  Summary of Regression Equation Predicting Verbal Abuse with PAS and PCERQ  100  Summary of Regression Equation Predicting Physical Abuse with PAS and PCERQ  102  Mean Scores of PCERQ Sub-Scales as a Function of Sample  vi.  .  106  Development and Evaluation of the PCERQ  List of Figures  Figure I  Figure 2  Figure 3  The Interactive Effect of the Need for Power by Exonerative Rationalizations Sub-Scales in Predicting Physical Abuse The Interaction of the Propensity for Abuse Scale and the Conflict Ineffectiveness Sub-Scale in Predicting Physical Abuse . . . . . . . . Overview Summary of the PCERQ Sub-Scales' Properties  vii.  98  103 .  109  Development and Evaluation of the PCERQ  DEDICATION  The completion of this project was enabled in very large part through the patience and understanding of Sappho Griffin; she dealt with this large addition to our little family with her natural grace.  viii.  Development and Evaluation of the PCERQ  ACKNOWLEDGEMENTS  I wish to first extend thanks to Donald Dutton, who has been a supportive advisor throughout my grad school career; our collaborations formed the nuclei of this research. The members of my departmental committee, namely Jennifer Campbell, Jim Enns, Boris Gorzalka, Peter Suedfeld and Del Paulhus, provided essential feedback and insight that was helpful and prompt, particularly as the project neared completion. The Social Sciences and Humanities Research Council (SSHRC) and University Graduate Fellowship (UGF) provided funding for me in recent years. The statistical consulting help of Roger Tweed and Tom Ruteledge was greatly appreciated. Larry Axelrod has been a reliable and guiding force through the various phases of this project, and it could not have come together without this steadfast influence. Kelley Persoage and others in the lab were of invaluable help over a long period time in helping with data collection. Outside of the university community there have been many people who supported this work and helped me with various aspects of it. The staff of the Psychology Department at Riverview Hospital, the Regional Health Center, Jim Hamilton, Zender Katz, David Kolton and Diana Mawson are among those who have been helpful along the way. Robert Konopasky, David Nussbaum and Carson Smiley have all gone beyond the call of duty for me and have been wonderful role models. Many people from outside the academic community who have not always understood what this was all about, but cared nonetheless. The post-modern anarchist John McNally, the man that time forgot Russ Pitts, the odysseian Tim Brierley, my former room-mates Adam DiPaula and Steve Heine and the artist Sean Alward have all been among those caught in the maelstrom of this project at some point. The support and presence of my father, mother, Mark, Joanne and Dusty have collectively and individually been a part of this journey far from home. ix.  Development and Evaluation of the PCERQ  1  Chapter I Overview and Literature Review  Overview  Despite being examinedfroma variety of perspectives, there remain important gaps in causal conceptions of wife assault. Until the mid-1990's researchers focused mostly on how 1  various dispositional factors contributed to intimate violence. These models evoked the image of men being propelled toward abusive behaviour on the basis of long-standing predispositional characteristics. Little attention was given to psychological phenomena experienced more proximal to the abusive behaviour, namely during conflict situations, that could regulate or moderate abusiveness. An exploration of these various perspectives follows in the remainder of Chapter 1. Chapter 2 provides the rationale for this dissertation's examination of appraisals and motivations related to power, conflict efficacy and responsibility that are hypothesized to influence abuse. These concepts are believed to be especially relevant to determining whether conflicts become abusive. The constructs being examined include: (a) the desire for power over one's partner, (b) appraisals of self-efficacy to non-abusively manage conflict, and (c) the stability of non-abusive conflict intentions. Examining such phenomena is consistent with a more recent trend in the research which examines intimate violence as the product of The term "wife assault" will be used throughout this thesis to refer to acts of physical aggression committed by men towards their intimate female partner. The terms "assaultive men" and "wife assaulter(s)" refer to men known to be physically assaultive towards their wife, common-law partner, or girlfriend. Physical assaults refer to events such as pushing, slapping, punching, kicking, biting or otherwise physically injuring their partner as reflected in the physical abuse items of the Conflict Tactics Scale (CTS: Straus, 1979). These terms are not intended to refer to men violent with other men but not with their female partner. Ultimately, they are intended to reflect the perpetration of physical abuse, which is almost always accompanied by some variety of emotional abuse. Men who are solely emotionally or psychologically abusive towards their female partner but have not engaged in physical assaults of the sort listed above would not be classified as assaultive men according to this definition. 1  Development and Evaluation of the PCERQ  2 interwoven patterns of relationship interaction and conflict management (Holtzworth-Monroe & Smutzler, 1996; Gortner, Gollan & Jacobson, 1997). Integrating these psychological elements into a psychometric tool (the Power, Conflict Efficacy and Responsibility Questionnaire, or PCERQ) is viewed as a useful goal in furthering our capacity to understand and assess forces that disinhibit aggression which are not primarily the product of longstanding personality dysfunction. Chapter 3 consists of an overview of the purposes and methods employed in this dissertation. This is followed by descriptions of the samples' background characteristics in Chapter 4. The PCERQ sub-scales' construction through theoretical and empirical considerations appears in Chapter 5, along with an evaluation of their internal properties. Concurrent validity of the PCERQ sub-scales are presented and evaluated in Chapter 6, while Chapter 7 focuses on the capacity of the new instrument to predict self-reported abusiveness. The discussions of Chapter 8 offer interpretations of the findings in light of the initial hypotheses, experimental limitations, relationships to the existing literature, clinical implications and future research possibilities.  Literature Review: Existing Research on the Causes of Wife Assault. This dissertation is ultimately about refining our understanding of how men experience intimate conflict, which is seen as contributing to abusive behaviour. This section provides background on other facets of abusiveness which have emerged in the literature. It will be seen that some of the more progressive research on assaultive men has investigated the prominent role of long-standing personality traits that generate intensely negative affect within the intimate relationship. Eventually it will be seen that this dissertation's task is the measurement and evaluation of psychological processes related to the (mis)management of impulses generated by those traits during conflict, enabling a more thorough appreciation for how such impulses contribute to emotional and physical abuse.  Development and Evaluation of the PCERQ 3  Social Learning Theory Social learning theory (SLT) considers behaviour to be the product of one's experience in various learning environments. Of particular importance within this theory are the social and psychological contexts in which behaviour is learned, developed and maintained. The most relevant components of SLT to wife assault are: a) that behaviour is learned by performance or observation, b) that the labelling and processing of physiological and affective states arising in interpersonal situations are learned, and c) that behaviour is maintained and regulated by various cognitive appraisals and contingencies. Strong links between the observation and subsequent use of aggression in close relationships has been well documented. It has been shown, for example, that men who witnessed paternal abuse of their mother in childhood were three times more likely to have assaulted their wives than men who had not seen such behaviour (Straus, Gelles, & Steinmetz, 1980). In a similar vein, a prospective study by Widom (1989) indicated that men who witnessed abuse in childhood were more violent than those who had not. Social learning theorists have identified reinforcing aspects of abusiveness that facilitate the development and maintenance of such behaviour, including the termination of aversive situations, increased agency in a potentially threatening or unpredictable situation, and gaining compliance of one's partner are all possible reinforcers of intimate aggression (Dutton, 1995b). In line with this view of how abusiveness is learned, it has been found that many wife assaulters do not acquire their maritally violent behaviour through modeling, but rather adopt it through their own experiences using it in their first adult relationship (Kalmuss & Seltzer, 1986). Research in SLT has also considered how cognitive processing of affective states is learned. This work has examined how people label their emotions and what they attribute them to, along with the triggering effects of the labelling-attribution process on subsequent arousal and behaviour. In the area of abuse, Novaco (1976) proposed that anger defends men against other less acceptable emotions by providing them with a sense of control. He also  Development and Evaluation of the PCERQ 4  advanced the notion that experiencing emotions from a bias o f anger becomes the preferred mode for men in relationships; various uncomfortable emotions can get labelled as 'anger' so as to enhance a feeling o f control over the conflict situation. In a similar vein, Gondolf (1985) explained that, by interpreting many situations from an 'anger' standpoint, men limit their subsequent range o f responses, and can end up blaming others for their emotional discomfort. Dutton, Van Ginkle, and Starzomski (1995) found that assaultive men tend to report more shaming experiences from childhood. The vulnerability to re-experiencing shame in these men is high, and they were found to defend against this self-directed negative affect through hostile projections that trigger anger. Starzomski (1993) showed that seeing one's girlfriend as responsible for relationship discord was positively correlated with male undergraduates' self-reported partner abuse. Together, these various studies suggest that many maritally-abusive men learn to process intimate relationship dysphoria with blame and anger, determining they are 'angry' due to their partner's conduct. A SLT construct that would seem to have great potential in understanding how men regulate their behaviour in relationship conflict is self-efficacy, or an individual's beliefs about their capabilities to exercise control over events affecting their lives (Bandura, 1992). Selfefficacy is domain-specific, in that one's sense o f agency and competency is relatively unique to the activity in which one is engaging (Miller, et al., 1986; Schwarzer, 1992). This dissertation introduces and measures the concept of'conflict efficacy': an individual's sense that they can non-abusively resolve interpersonal difficulties. Balshaw (1993) observed that previously-violent or at-risk men desisting from violence had developed self-efficacy to use non-violent conflict tactics. In one o f the few studies exploring self-efficacy and abuse, Tolman, Edleson and Fendrich (1996) found that perceived control over one's own behaviour was strongly related to intentions to be nonabusive. That study focused on how easily men believed they would be able to avoid getting violent in various scenarios as a measure o f efficacy. They found that perceived behavioral control played a significant role in predicting abusive behaviour.  Development and Evaluation of the PCERQ 5  Prince and Arias (1994) conducted one of the only other studies investigating selfefficacy in relation to wife assault. They investigated self-efficacy, desirability of control over partner, and self-esteem in a sample of assaultive men and found two clusters of respondents accounting for 68% of their sample (n=25). While low self efficacy for gaining control over one's partner was common to both clusters, one group had high self-esteem and a high desire for control while the second group had low self-esteem and low desire for control. The authors did not report any data regarding differences in abusiveness between the two groups. They speculated that those low on both self-esteem and desire for control would be aggressive out of frustration, whereas those high on these two variables would be aggressive to create a more secure, or comfortable and dominant, relationship position. In sum, SLT has provided a framework that has increased the knowledge base regarding the learned nature of abusiveness. Studies of how men learn to process conflict in ways that foster emotional reactivity and abusive behaviour is a particularly valuable SLT application. Conflict efficacy remains a relatively unexamined aspect of the wife assault research. This thesis is advancing the notion that conflict-efficacy could be a useful one, and the instrument being developed in this dissertation will assess the degree to which men feel capable of non-abusive intimate conflict resolution. This new tool, specific to the forum of the intimate relationship, is more comprehensive than the general "violence avoidance" view of efficacy used by other researchers (Tolman et. al, 1996). Another SLT deficit in previous studies of abuse is the lack of attention paid to intentions or motivations for (non-) abusive behaviour. By gaining a more specific awareness of what men want from conflict, and experience during it, an understanding may emerge about why some men retain aggression as a conflict tactic. Overall, SLT provides some basis for understanding how powerfully aversive states of arousal may be learned, generated and maintained. This dissertation evaluates how conflict efficacy relates to the management of those feelings and the aggressive impulses they generate.  Development and Evaluation of the PCERQ 6  Personality and Individual Differences This domain of psychological research has been particularly productive, especially in recent years, in accounting for impulsive intimate violence. The focus of impulse management which runs through this thesis has emergedfromthefindingsof this personality-oriented literature. Early reviews of wife assault writings by Hotaling and Sugarman (1986) and Tolman and Bennett (1990) indicated wife assaulters on the whole had poor self-esteem, exaggerated dependency on their partner, rigid patriarchal values, substance abuse problems, abnormally high levels of anger and jealousy, as well as deficits in a range of social, academic and vocational skills. The Abusive Personality, conceptualized and researched by Dutton and his colleagues (Dutton, 1994b; Dutton & Starzomski, 1997), is a fusion of personality variables strongly related to abusiveness in intimate relationships. It emphasizes the impulsive nature of intimate violence by a sub-group of men. The variables aggregated within the Abusive Personality are: borderline personality organization (Kernberg, 1977), chronic anger, trauma symptoms, fearful attachment style, and parental abuse in childhood. The three hallmark features of borderline personality organization as outlined by Gunderson (1984) that connect so strikingly and specifically with wife assault are: (a) a proclivity for intense, unstable interpersonal relationships; (b) an unstable sense of self, intolerance of being alone and abandonment anxiety; and (c) intense anger, demandingness, and impulsivity (Dutton & Starzomski, 1993; Dutton, 1994b). Fearful attachment style is the tendency to view both oneself and one's partner as incapable of contributing to a safe and trusting intimacy climate (Dutton, Saunders, Starzomski, & Bartholomew, 1994). The sub-grouping approach to perpetrators of wife assault represented by the Abusive Personality is part of an emerging trend of searching for patterns of heterogeneity within samples of abusers, particularly based on personality differences. This type of research has become a focus in the mental health and forensic literatures more generally (Andrews & Bonta, 1994; Haslam & Beck, 1994; Marengo, 1994). Considering personality diversity allows for examinations of variation in different patterns of abusiveness. Early writings on  Development and Evaluation of the PCERQ 7  possible personality differences among wife assaulters was theoretical in nature (Elbow, 1977) or based on sparse empirical support (Caesar, 1986). Elbow (1977) proposed a theoretical typology outlining four groups of assaultive men, which suggested that men in each of the sub-groups were motivated to use violence by different forces. Caesar's (1986) typology, derived from psychological testing and interviews with a sample of assaultive men, also consisted of four sub-groups. The importance of this early work was for its focus on classifying types of assaultive men, as details of the typologies advanced by these writers did not meet with much subsequent research or clinical use. Hamberger and Hastings (1986; 1988) were among the first researchers to empirically classify assaultive men on the basis of heterogeneous personality features. They found that assaultive men could be clustered into three groups based on personality disorder symptoms: (a) schizoidal / borderline, (b) narcissistic / antisocial, and (c) passive dependent / compulsive. Saunders (1992) also identified three sub-groups varying on personality-related variables like jealousy and coping style, in addition to descriptive data regarding violent behaviours, family of origin violence, and alcohol use. His descriptors for the groupings he uncovered were: (a) generally violent (psychopathic features of impulsivity and various anti-social behaviors), (b) emotionally volatile (borderline personality features resulting in abuse primarily within the intimate dyad), and (c) overcontrolled (avoidant of confrontation and conflict, unassertive, with abuse arising in a defensive/reactive way). Holtzworth-Monroe and Stuart (1994) reviewed sub-grouping studies with samples of assaultive men. Their appraisal outlined three sub-types that have emerged across many studies and that vary in violence frequency, targets, and personality functioning. These groupings were labelled family-only (low frequency, violence only in family, absence of severe personality disorder; -50% of wife assaulters), dysphoric-borderline (moderate-high frequency, violence mainly in family, borderline or schizoid personality disorder; -25% of wife assaulters), and violent/antisocial (moderate-high frequency, violent in many interpersonal domains, antisocial personality or psychopathy; -25% of wife assaulters). The  Development and Evaluation of the PCERQ 8  Abusive Personality conceptualization advanced by Dutton corresponds to the dysphoricborderline sub-group. Each of these groups were posited to have unique causal factors for their violence, relating to both temporally-distal influences (e.g., childhood events) and contemporary ones (e.g., social skill deficits, nature of emotional attachments formed with loved ones, attitudes, and relationship interactions). Dutton (1997) outlined how the dysphoric-borderline / Abusive Personality cluster of assaulters have impulsivity as a core feature of their behaviour, as opposed to the cool and calculating violence of the violent/antisocial abuser. Gottman, Jacobson, Rushe, Shortt, Babcock, La Taillade and Waltz (1995) described the results of a study that revealed two categories of assaultive men. A central difference between the two types was whether heart rate decreased within five minutes of a marital conflict interaction: Type 1 assaulters had, on average, a substantially lower heart rate minutes after a disputes and also tended to be the more severely violent of the two groups, with higher levels of antisocial personality disorder and violence outside the family. In relation to the categories described by Holtzworth-Monroe and Stewart (1994), personality measures suggested the Type 1 men corresponded to the violent/antisocial category, and Type 2 men were similar to the dysphoric/borderline group. Whereas most of the studies mentioned above have dealt primarily with variables linked to personality disturbance, Barnett and Hamberger (1992) produced one of the few studies to examine non-clinical personality features of assaultive men. Using Gough's California Psychological Inventory (CPI) they found that men in non-violent control samples had higher levels of such traits as Socialization, Responsibility, Self-Control, and Tolerance than maritally-violent men. Relative to non-violent men, those who were maritally-violent were described as less resourceful, creative and planfiil in their management of intimate conflict. A concluding statement of their work echoes the philosophy underlying the conceptualization of violence being advanced in this thesis:  Development and Evaluation of the PCERQ 9  More research needs to be done... to clarify batterer attributes in terms of "normative", "everyday" interpersonal processes. Such research will have particular relevance for developing batterer interventions which emphasize development and enhancement of problem-solving and interpersonal conflict-resolution skills. (Barnett & Hamberger, 1992, p. 26). The psychological processes investigated in this dissertation (i.e., conflict efficacy, need for power, responsibility) are very much in keeping with these observations and suggestions in that they are not inherently rooted in psychopathology or personality dysfunction. Overall, studies of personality disturbances in assaultive men converge on two main conclusions. First, the prevalence of personality dysfunction among wife assaulters is greater than one sees in the general population, despite discrepancies as to exactly what this prevalence may be. Second, borderline and antisocial personality disorders appear more frequently in samples of wife assaulters than other personality dysfunctions. While it is apparent that samples of wife assaulters show levels of personality dysfunction higher than normative levels, by the opposite side of the same coin there is variability in the proportion of samples that show insignificant diagnosible personality pathology . These conclusions 2  demonstrate the importance of incorporating, but going beyond, personality dysfunction in causal accounts of wife assault. Comprehensive etiological explanations are needed for those abusive men who do not show prominent signs of personality disorder. The variables of interest in this study, which relate to the management of one's behaviour and impulses during conflict, apply to those both with and without strong aggressive impulses generated by personality disturbances.  Estimates of the proportion of assaultive men who show no indication of personality dysfunction range from 12% (Hamberger & Hastings, 1986; 1988; Hart et al, 1993 via the MCMI-II; Hale, Zimostrad, Duckworth, & Nicholas, 1988) to upwards of 50% (Hart, et al, 1993 via the PDE; Flournoy and Wilson, 1991). Explanations for this variability between samples may relate to criminal justice system contact (Dutton & Starzomski, 1994) and/or choice of assessment instruments (Hart, et al, 1993).  Development and Evaluation of the PCERQ 10  Sociocultural - Feminist Accounts of Wife Assault Theoretically, feminist positions on wife assault posit that men are socialized into a gender role that values dominance and power over women (Bograd & Yllo, 1988; Dobash and Dobash, 1979; Pence and Paymar, 1993). Empirically, research in this vein often relies on socioeconomic indicators to study the extent of male power over, and accompanying abuse of, women. Hypotheses directing these studies have postulated that male abusiveness is the product of having sociocultural-structural advantages (as indicated by relativefinancialand educational status), and hence power over, women. These theorists also suggest that perceived threats to, or relative loss of, socioeconomic status results in reactive abuse (Yllo & Straus, 1990). Impulsivity does not occupy a conspicuous place in this model, which explains abuse as methodically oriented toward establishing and maintaining control over women. Some parts of this model have been empirically supported, although consistency has been lacking. Hornung, McCullough and Sugimoto (1981) found that women with jobs of higher status than their husbands were more likely to experience life-threatening violence from their partner than women in jobs of equal status to their husbands. Although Hotalling and Sugarman (1986) concluded that if a woman has more education or makes more money than her husband she is more likely to be abused, Babcock et al., (1993) failed to replicate this finding. Yllo and Straus (1990) found a curvilinear relationship between relative socioeconomic status of women and wife assault rates, such that those American states with the highest and lowest statuses of women had higher rates of husband violence than states with a more average status of women. The authors suggested that women with the least social power were the victims of oppressive power-wielding males, while women with the relatively greatest socioeconomic status were abused by resentful men fearful of further power loss. See Dutton (1994c) and Renzetti (1994) for elaboration on the debates within this perspective.  Development and Evaluation of the PCERQ 11  The general finding from this stream of research is that gender-based social power differences are of some value in accounting for variance in patterns of marital violence. Attitudes about relationship influence and roles, shaped by society, have been shown to strongly correspond with the personality features comprising the Abusive Personality (Dutton & Starzomski, 1997). These empirical observations hint at the possibility that personality and sociocultural experience and values interact, and culture may serve as a potential 'carrier' of abuse-enabling values to which personality-disturbed men gravitate. What remains to be more completely considered is the role of socioeonomically-based power elements in relation to personality and proximal psychological phenomena.  Relationship Interactions and Wife Assault. In the 1990s many researchers have come to devote attention to particular elements of marital interaction accompanying conflict and abuse. This has resulted in a body of knowledge cataloguing interpersonal characteristics, events and exchanges unique to violent couples. Researchers use observational methodologies to describe interaction patterns, and formulate theoretical explanations for the patterns of influence and behaviour of interest. The focus of this approach is the interpersonal level of abuse, as opposed to the intrapsychic processes of the parties involved. Hence, impulsivity and impulse management is not of foremost concern to these researchers. Nonetheless, consideration of their work is warranted, as its integration with new findings from intrapsychic-oriented research is likely to be profitable in the future. A majorfindingabout relationship interactions linked to male abuse is called negative reciprocity. This is a pattern observed in male-violent relationships involving the escalation (rather than the ceasing) of reciprocally negative or aversive behaviour once it has begun (Cordova, Jacobson, Gottman, Rushe & Cox, 1993). Other studies in this area have reported that maritally-violent men react with levels of anger rated as substantially higher than those of non-violent men in response to partner behaviours. Among the interpersonal cues eliciting  Development and Evaluation of the PCERQ 12  this high level of anger among maritally-violent men are changes in conversation topic during conflict, angry wife behaviors, and other 'negative' wife behaviors such as withdrawal and expressions of fear and sadness (Burman, Margolin & John, 1993). Jacobson, Gottman, Waltz, Rushe, Babcock and Holtzworth-Monroe (1994) examined arguments of violent and non-violent couples and found that wives (and not husbands) were afraid during violent arguments, and that violent men were more likely to use provocative anger than non-abusive men. Their research also indicated that no wife behaviours were able to terminate husband violence once it had begun. Relationship interactions have been studied in relation to the course of violent relationships. Jacobson, Gottman, Gortner, Berns and Shortt (1996) found that violent marriages led to divorce or separation in cases where wives assertively defended themselves. The authors also indicated that relationships that continued were characterized by a decrease in husband violence over time, but unchanged levels of emotional abuse. Holtzworth-Munroe and Smutzler (1996) determined that assaultive men were less supportive and caring when their spouse was distressed, and they were also less likely to discuss the spouse-distressing issue than non-assaultive men. Furthermore, the assaultive men indicated a tendency to get more angry and respond in a hostile manner to distress relative to the responses offered by non-assaultive men. One of the more recent relationship interaction studies examined husbands' rejection of wife attempts to influence them. Coan, Gottman, Babcock and Jacobson (1997) found that Type 1 batterers (resembling the violent/antisocial sub-type of Holtzworth-Monroe and Stuart) consistently sought to undermine "any and all" efforts by their partner to affect some change in her behaviour. From their findings, the authors made a case for linking the initiation of negative reciprocity, escalation of intimidation and emotional intensity and the 'surrender' of the wife's intention of influence (based on the woman's experience of fear during such encounters). The authors also posited that rejecting spousal influence is symbolic to the instrumental Type 1 batterer, representing a level of independence as well as suppression of  Development and Evaluation of the PCERQ 13  what is perceived as a threatening and even humiliating advance. In other words, Type 1 men appear to be using violence as a means to control their wife through intimidation. Type 2 batterers (akin to the dysphoric/borderline subtype), they suggest, respond with impulsive or reactive aggression to scenarios of perceived abandonment and greater partner independence so as to engage their partner in a sustained encounter or event. Relationship interaction research has identified processes and patterns of conflict differentially operating in violent versus non-violent relationships. Nonetheless, despite some work contrasting interaction patterns of the Type 1 and Type 2 batterer, explaining why assaultive men have problems recognizing, decoding and/or responding to the emotional distress of their partner in a non-assaultive and self-controlled manner is not a strong suit of interactionism research. The links between their internal psychological experiences during these times of dyadic or partner distress and their interactive responses and aggression remains unclear. While it is not the task of this dissertation to investigate interactionism in relation to the concepts being studied (i.e., conflict efficacy, need for power, responsibility), alerting the reader to this stream of research allows for a more complete framing of how the thesis fits within the abuse area as a whole.  Sociobiological Factors and Wife Assault Sociobiology considers aggression and other social behaviours as related to evolutionary principles of species survival. From this theoretical perspective, reproduction of the species is a fundamental motivator of conduct, particularly in relationships that relate to potential reproduction opportunities (or threats to reproduction). Dutton (1995b) summarized sociobiological formulations of aggression in heterosexual relationships and highlighted the paradoxical implications of this theory: how could physical attacks on one's reproductive partner be advantageous to the propagation of the species? In other words, while it fits with evolutionary principles that men would attack other men who they perceive  Development and Evaluation of the PCERQ 14  as rivals for mating with their female partner, optimizing species survial through chronic abusiveness toward their own partner appears counterproductive. A way that sociobiological theory has not been thoroughly assessed, however, is by considering the possibility that abusive men end up single or estranged more often than nonabusive men (and subsequently less likely to pass on their genetic composition). If such a pattern were found, it would suggest that abusive behaviour has the effect of distancing men from access to reproductive opportunities. Nonetheless, much research has addressed the cycle of violence describing how abusive relationships persist despite the violence and often produce children. Daly and Wilson (1988) produced a cornerstone piece of empirical research on sociobiological influences in the area of violence, particularly homicide, in heterosexual relationships. They found that men in Detroit who were in relationships theoretically associated with the greatest threats to a male's reproductivefitness(i.e., when men are involved with women much younger than themselves) killed their female partner at rates much higher than men more equal to, or younger than, their partner. Nonetheless, the claim that these results strongly support the evolutionary perspective is undermined by the empirical confounding of race and cultural issues in that data (Dutton, 1995b). It will become apparent in the ensuing chapters that the new conceptual material being explored in this dissertation is not rooted in, or explicitly linked to, sociobiology. Nonetheless, this section has been presented simply to alert the reader to another perspective on male aggression in heterosexual relationships.  On Extending the Merits of Existing Explanations Despite the contributions of the various perspectives described above, there are inevitably some weaknesses within them. This dissertation aspires to integrate some of the more notable findings while bolstering knowledge in some of the areas less-well explored. Research on personality disturbance in wife assaulters has suggested that many of these men  Development and Evaluation of the PCERQ 15  do not show strong personality disturbance. Personality research has also been unable to explain variability in the (mis)management of in-the-moment aggressive impulses associated with personality-generated aversive emotional states. SLT presents concepts, particularly self-efficacy, that are likely integral to the management of impulses and the regulation of behaviour during conflict (including subsequent to abusive impulses). Psychological indicators of cultural messages condoning of abusive dominance by men in relationships, as suggested by sociocultural / feminist scholars, has yet to be operationalized or studied as a proximal mechanism that disinbits abusive behaviour. From this stream, need for power and responsibility can be conceptualized as variables influential in the modulation of impulses to re-assert dominance over one's partner by aggressive means. Overall then, this dissertation seeks to build on the strengths of these existing pillars of research by adding elements related to the management of interpersonal conflict and the aversive emotional states and aggressive impulses that may accompany it.  Development and Evaluation of the PCERQ 16  Chapter H  The Rationale for Investigating and Measuring Power, Conflict Efficacy and Responsibility.  The task of this research was to develop a measure of psychological variables believed to strongly affect the experience (i.e., motivations, appraisals, and phenomenology) of intimate distress and conflict so as to increase to likelihood of abuse. Variables were targetted for study that had not yet been examined in much detail by wife assault researchers, but which appeared to complement the streams of research discussed in the first chapter. Of particular interest were: (1) need for power (over one's partner), (2) conflict-efficacy (one's perceived adequacy for being non-abusive), and (3) responsibility (the nature and strength of resolutions to be non-abusive). It was hypothesized that these variables would bear on how men appraised conflict and thereby generated and evaluated the behavioral choices available. Additionally, it was forecast that these variables would interact with one another, and with the more distal influences discussed in the previous section, to either facilitate or avert abuse. What follows is the rationale for arriving at the categories of the measure being developed, called the Power, Conflict Efficacy and Responsibility Questionnaire (PCERQ). In essence, this rationale can be summarized as probing the links between situational or 'in-themoment' motivations and abuse, as well as examining the associations between personal accountability for non-abusiveness in the face of a desire to dominate one's partner or a relationship conflict. Running through these areas of investigation is the common thread that thoughts and feelings that may facilitate abuse which arising in the context of conflict may not be based in personality pathology or long-standing dispositional influences. These more proximal types of cognitions, motivations and perceptions tend to be the foci of therapy with assaultive men, but have not been subjected to much in the way of research or comparison with more distal contributors to violence. This chapter offers background on the themes  Development and Evaluation of the PCERQ 17  underlying the development of the item pool, while the reduction of the PCERQ into conceptually-meaningful sub-scales is described in Chapter 4.  Conflict Efficacy Some PCERQ items were written to assess areas deemed relevant to the construct of conflict efficacy. In actuality, item wording resulted in a the scale of interest measuring 'inefficacy' as opposed to 'efficacy'. The label attached to this scale was Conflict Ineffectiveness (CI). The content of the CI items followed from research on marital locus of control (MLOC), which applies self-efficacy to marital problem-solving (Miller, Lefcourt, Holmes, Ware, & Saleh, 1986). Their research suggested that married individuals with a high internal marital locus of control (who believed they could positively affect the course of marital interactions and outcomes) were more satisfied with their relationships and the quality of resolutions reached with their partner. They also reported that M L O C was uncorrelated with the use of destructive tactics in marriage. Balshaw (1993) concluded, based on her qualitative study with men who had come to refrain from abusiveness, that developing a sense of confidence in one's ability to use non-abusive methods was a big part of becoming nonviolent with their partner. Findings from these various areas led to the writing of items intended to specifically assess one's perceived ability to use non-abusive methods during conflict. Additionally, these items were to measure the extent to which respondents believed they were capable of coping with situations ranging in emotional intensity versus feeling that there was little that they could do to contain conflict or their behaviour during it. "When my partner and I have a disagreement things sometimes speed up and get out of control.", reflecting a low opinion of one's skill base for coping with conflict, is an item from the CI category. The SPQ is a brief questionnaire, developed by the writer for a previous study, that measures aspects of the conflict efficacy. It was evaluated in a pilot study with 48  Development and Evaluation o f the PCERQ  18  undergraduate couples conducted by the researcher in the 1995-96 academic year. The SPQ was found to correlate strongly with reports of male verbal and physical abuse provided by both men (verbal abuse: r = .68, p < .001; physical abuse: r = .35, p < .05) and their female partners (verbal abuse: r = .54, p < .001; r = .36, p < .05).The SPQ items are presented in Appendix D-10.  Need for Power Power has been a major topic in family violence research, but it has been conceptualized and studied in an idiosyncratic fashion. A consensus in this area in recent years suggests that power in intimate relationships is best viewed in a multi-faceted way (Babcock, Waltz, Jacobson, & Gottman, 1993), a perspective adopted in the present study. In keeping with this multi-faceted outlook, the two aspects of power explored in this dissertation are power bases (various interpersonal influence assets at one's disposal) and need for power (desire or motivation to influence). Power bases are assessed in the introductory demographic questions of the questionnaire battery, while need for power (n-power) is being examined within the PCERQ. Blau (1964) suggested that power bases can be defined as resources that are either intrinsic/psychological (e.g., emotional stability, communication skills) or extrinsic/social status (e.g., relative job prestige, financial independence, commitment to a relationship). Such resources relate to one's capacity to influence a relationship partner. Research indicates that coercion and abusiveness are used more often when intrinsic power bases are functioning poorly (Babcock, etal, 1993; Dutton & Strachan, 1987). For example, assaultive men may find that their intrinsic power bases are chronically failing them and therefore be tempted to use coercive methods to restore perceived losses in relative power within the marital relationship. On a related note, Howard, Blumstein, and Schwartz (1986) found that such factors as gender identification, relative power and dependency were related to influence strategies in close relationships. Those who were more feminine, had less structural/extrinsic  Development and Evaluation of the PCERQ 19  power and were more committed to and dependent on their partner tended to rely on weaker tactics, such as manipulation and supplication, to influence their partners. In sum, the extent to which a man perceives himself as lacking power within the relationship appears to relate to how likely he is to use aggression as a way to "re-balance" the perceived discrepancy. Winter (1973) described need for power as a motivation to have influence over, or impact upon, others. It is also associated with sensitivity to influence attempts (either as target or initiator) within one's social environment. When one's n-power is threatened, Winter (1973) suggested that attempts to gain control will increase in proportion with the level of drive and threat. In support of this, studies have shown that people high in n-power engage in arguments, yell in traffic, destroy furniture, and insult store clerks more often than those with low n-power (McClelland, 1975; Winter, 1973). Winter and his colleagues have evaluated the importance of n-power in intimate heterosexual relationships, and found that men with high n-power expect more problems in relationships, distrust and exploit their partners, and also attempt to suppress their partner's socioeconomic status (Stewart & Rubin, 1976). Studies by Dutton and Strachan (1987) and Mason and Blankenship (1987) have reported that men with high n-power inflict more abuse on their partners than those with low n-power. In spite of these results showing an association between n-power and anti-social behaviour, the discussion of responsibility that follows shortly will outline how the relationship may not be that simple. A novel feature of the PCERQ is that it contains items written to specifically apply Winter's n-power notions to the domain of intimate relationship influence. Examples of PCERQ items from the n-power category are: "In my relationship I like to be the one who guides the decisions." and "I debate with my partner and try to convince him/her into my point of view."  Responsibility The nature of one's personal responsibility to refrain from intimate violence has not, to this time, been the subject of substantial empirical examination. A survey of thoughts and  Development and Evaluation of the PCERQ  20  writings about one's responsibility to refrain from physical aggression should help the reader more fully understand its place within this dissertation. Essentially, responsibility is anticipated to play an important role in how one evaluates one's options during conflict (particularly the extent to which aggressive options are considered and utilized). This variable is being examined, in part, based on the theoretical input of feminist scholars who have suggested that men receive inconsistent messages about the unacceptability of aggression toward women. From such a vantage point, one's feelings of accountability for being nonviolent toward one's partner may be eroded or compromised by undertones in the culture that violent lapses or strategies are tolerable. The moral responsibility of the abuser for his acts seems to sometimes be overlooked or distorted. Remarks from a Nova Scotia judge justifying his minimal sanction for a man convicted of severely battering his wife (he was shortly thereafter convicted of the woman's murder) is an example: "I don't know... whether it is your own fault or you happen to have a very sensitive mate who was easily rattled." (Cox, 1995). In academic circles, Lamb (1991) analyzed wife assault articles and suggested that male academics writing on wife assault tend to adopt a stylistic tone more exonerative of the abuse than men writing with women or women alone. That an abuser's responsibility for the violence he commits is a variable typically absent from models of aggression has been noted with concern in recent years. Cather's (1995) commentary on a multivariate model of violence that failed to incorporate personal accountability summarized the nature of this concern: "This view of behaviour as determined by 'biopsychosocial' factors interacting in ways 'at least as complex as metabolic pathways' not only relieves people of responsibility for their behaviour, it also minimizes the role of human agency..." (p. 413). Burman and her colleagues have stressed the importance of being mindful of the accountability of the violent individual in mapping causes of aggression. They comment that: " presume that working with the victim would reduce the partner's physical aggression ignores the power of the physical aggressor and that person's decision to  Development and Evaluation of the PCERQ 21  use physical aggression" (p. 38; emphases added). In reference to the increased attention paid by researchers to interpersonal conflict events they state: "In the interests of stopping aggression in marriage, one would not focus solely on these interactions without also focusing on the aggressor's responsibility, prior experiences with aggression, and the social systems that support aggression." (p. 38). Margolin, Gordis, Oliver, and Raine (1995) stated that "What must be kept at the forefront of our thinking about battering is that men who batter make the decision to be violent... no finding regarding an interactional pattern could challenge the responsibility of the person exhibiting the aggression." (p. 260). Collectively, these remarks point toward an increased need to understand the choice-making process and responsibility of men for managing their behaviour so as to be non-violent. The personality trait literature is an obvious area of reference in trying to develop a measure and understanding of responsibility to refrain from aggression. Of the Big Five personality traits, Conscientiousness would seem to relate most strongly to the responsibility concept being considered in this research. The questionnaire measure of this trait "assesses the degree of organization, persistence, control, and motivation in goal-directed behaviour." (p. 3, Costa, Jr. & Widiger, 1994). While this aggregation of psychological features fits with the self-control and self-regulation notions of responsibility being advanced in this dissertation, Big Five traits other than Conscientiousness have been shown to be crucial precursors of aggression. Research by Caprara, Barbaranelli, Pastorelli and Perugini (1994) found that impulsive-reactive aggression was strongly related to Neuroticism and instrumental-planned violence was particularly associated with low Agreeableness. They found that Conscientiousness took a less important role in aggression relative to these traits. Although others have also psychometrically explored the concept of responsibility (e.g, Gough & Bradley, 1992; Suedfeld, Hakstian, Rank, & Ballard, 1985), the responsibility concept adopted for the present research is based on the work of Winter and Barenbaum (1984). According to their model, responsibility has several components: (a) respect for standards of behaviour, (b) altruistic concern for the impact of one's behaviour on others, (c)  Development and Evaluation of the PCERQ 22  forethought and concern for the consequences of one's actions (d) use of a self-judgment or reflective process as to the effects of one's conduct, and (e) valuing the making and keeping obligations with others, this concept of responsibility by coding narratives written in response to pictures. Those with high levels of responsibility drank less alcohol, engaged in fewer fights and impulsive acts, were less sexually possessive of their partner, and protective of their children during divorce proceedings (Winter, 1988). Within this theory, responsibility is learned and refined across the lifespan through such activities as domestic tasks like caring for younger siblings and doing household chores. The Standards to be Non-Abusive (SNA) category of PCERQ items applies these concepts to the intimate domain. The Exonerative Rationalizations (ER) pool of PCERQ items was written to augment Winter's standards-oriented responsibility components by assessing possible fragility of one's willingness to adhere to non-abusive standards. These items attempt to measure one's resolve to live by non-abusive intentions in even the most challenging relationship situations. Those scoring high on E R items report difficulty in consistently abiding by non-abusive intentions. Balshaw (1993) found the combination of non-abusive intentions and the strong resolve to abide by them recurred in interviews with previously-abusive men who had come to desist from abuse. It may be that simply having non-abusive standards is insufficient for one to find alternatives to violence: that being aggression toward one's spouse is reprehensible under any conditions may be a more important cognitive set for supporting a non-abusive style. The Responsibility by N-Power Interaction. Of particular importance for this study is Winter'sfindingthat responsibility moderated how individuals expressed their n-power. Whereas his earlier work had determined that the dominance of n-power typically appeared to be problematic, this new result suggested a more positive set of outcomes could be associated with prominent levels of n-power. The observed interaction was such that only those high in n-power and low in responsibility were prone to using aggressive influence. Those high on measures of both responsibility and n-power used interpersonal influence tactics that were assertive and direct; proactive abuse more commonly followed from a  Development and Evaluation of the PCERQ 23  combination of low responsibility and high n-power. This interaction will be explored in this dissertation for its potential importance in clarifying the motivation of men during conflict. Particularly, it could be that high n-power need not be associated with aggressive outbursts during conflict among men with high responsibility.  Summary of the PCERQ's Rationale, Categories and Themes. The primary purpose of this research is to measure and evaluate parts of the conflict situation and experience that likely bear on one's decision to resort to physical aggression against one's partner. From this investigation it may be possible to better understand how various psychological features of conflict may contribute to abusive conduct. The PCERQ is a questionnaire measure that has been developed to assess these dimensions. The PCERQ's item categories and their meanings are as follows: (a) Conflict Ineffectiveness, namely the extent to which one feels capable of dealing with interpersonal problems in a non-abusive fashion, (b) Need for Power, which is the degree of influence one seeks to have over one's partner, (c) Standards to be Non-Abusive, indicating the value one places on abstaining from aggressive tactics toward one's partner, and (d) Exonerative Rationalizations, which assesses for the presence of cognitions that erode the stability of non-abusive standards. It is of great interest to examine these variables in relation to longer-standing personality problems and reports of abuse. That is, the psychological features assessed on the PCERQ scales may help us understand how emotional distress and aggressive impulses (generated from long-standing traits) are constrained or released. From such inquiry it should be possible to more thoroughly comprehend the processes by which different groups of men adopt aggression toward their partner or not. The PCERQ works toward this goal by aiming a spotlight at the conflict interaction itself, delineating the self- and partner-appraisals arising during such exchanges, and hopefully casting light on how men elect to be aggressive toward their partner. For such goals to be addressed the PCERQ scales must first be  Development and Evaluation of the PCERQ 24  psychometrically refined, which is the subject of Chapter 5. What follows are two chapters describing methodological strategies and sample characteristics.  Development and Evaluation of the PCERQ 25  Chapter m Intentions and Methods of Research The purpose of this dissertation is to refine and evaluate the P C E R Q sub-scales, so that they can be used to ascertain how conflict phenomena relate to partner abuse and a variety o f other abuse predictors. This chapter offers an in-depth presentation of the experimental hypotheses underlying the study, along with an account of the methodological strategies used in addressing those questions. The discussion o f hypotheses in this chapter has three components: ( 1 ) intercorrelations among P C E R Q sub-scales, as those constructs are anticipated to show a certain pattern of relationships with one another, ( 2 ) relationships with variables that relate to the P C E R Q ' s external or concurrent validity, and ( 3 ) relationships with abusiveness. The methodology discussion summarizes the steps and procedures undertaken in the collection of the data for the study.  Research Intentions & Hypotheses This part of the chapter provides the reader with an overview of the ideas and anticipated relationships among variables that guided the study. These relate to the P C E R Q sub-scale's interrelationships, their relationships with measures that speak to concurrent validity and their anticipated relationships with abuse criterion measures.  Anticipated Internal Properties The P C E R Q sub-scales were developed to capture the four categories described in the previous chapter: conflict ineffectiveness, n-power, standards to be non-abusive and exonerative (irresponsible) rationalizations. These four P C E R Q categories were believed to be interrelated and not to represent entirely unique or orthogonal psychological phenomena  Development and Evaluation of the PCERQ  26  relative to one another. These domains were anticipated to relate to one another in the following manner: Conflict Inffectiveness and N-Power: Research by Prince and Arias (1994) suggested that these two variables could be relatively independent of one another. While they reported uniformly low self-efficacy in the assaultive men they studied, they found variability in how much importance men placed on obtaining influence over their partner. Dutton and Strachan (1987), on the hand, found that assaultive men in their study had a strong inverse relationship between communication skills and need for power. From these results, it is unclear whether or not the Conflict Ineffectiveness sub-scale will be correlated with the n-power measure of the PCERQ. Conflict Ineffectiveness and Standards to be Non-Abusive: PCERQ items measuring these two areas were were hypothesized to correlate at least mildly in a negative direction with one another. This result was anticipated given the difficulty one was expected to have in maintaining a high standard to be non-abuse if one did not feel capable of fulfilling such personal expectations. Conflict Ineffectiveness and Exonerative Rationalizations: These two scales were expected to show a positive relationship with one another. This hypothesis follows from logic that one's level of ineffectiveness will coincide with cognitions that erode high expectations for being non-abusive. In other words, feeling very ineffective and having high expectations of refraining from abuse would appear to create a high level of dissonance in terms of appraising one's conduct. Hence, strong feelings of ineffectiveness and the presence of exonerative rationalizations appears to be a combination that would keep dissonance at a minimum. N-Power and Standards to be Non-Abusive: Winter and Barenbaum (1984) found that their standards-oriented measure of responsibility was uncorrelated with n-power. On the other hand, high n-power, and the implication of low responsibility (manifest in abusive behaviour), has been reported among assaulters relative to non-abusive men (e.g., Dutton &  Development and Evaluation of the PCERQ  27 Strachan, 1987). Outcomes from these two competing lines of research failed to offer a clear hypothesis about relationships that could be observed between power and responsibility items in this study. Nonetheless, given the fact that the PCERQ concepts of n-power and standards to be non-abusive follow from Winter's theories, a nil relationship would seem most likely between these variables. N-Power and Exonerative Rationalizations: It was not anticipated that these two variables would correlate with one another. If the n-power scale was comprised of items consistent with abusive conduct, one would expect a positive relationship between the two variables. Due to the fact that the n-power scale was designed to be descriptive of influenceseeking, with no particular implications of disrespect as part of that process, such a lack of relationship was foreseeable. Standards to be Non-Abusive and Exonerative Rationalizations: These two variables were not expected to correlate with one another. The rationale here was that one could claim to have non-abusive standards, but have either high or low resolve to consistently abide by those standards. Hence, non-abusive men would be expected to report high standards and low levels of exonerative rationalizations whereas abusive men could be high or low on standards but, likely, low on exonerative rationalizations. This pattern of hypothesized relationships among the PCERQ sub-scales, basically suggesting that several scales should relate to one another, indicated that an oblique (intercorrelated factors) extraction procedure would need to be considered during the factor analytic phase of the study. More on the scale and refinement process is presented in Chapter 5. Scale Properties: Temporal stability: The concepts assessed by the PCERQ sub-scales are conflict- and influence-related variables, and it is anticipated that they become established, utilized and engrained through life experience. These psychological features are expected to be relatively stable over time, as they are envisioned to reflect patterns of one's experience and management of conflict and influence experiences. In this light, test-retest reliability of  Development and Evaluation of the PCERQ 28  the PCERQ scales is expected to show stability over time. Despite this anticipated stability, change to these variables is believed to be plausible and many current treatment approaches target them. In other words, should a man come to view his proximal-to-conflict experience as problematic or if he has the adequacy of it meaningfully challenged, he may engage in a formal (counseling) or informal (reflection or discussion with partner or friends) process of change. Although it is acknowledged that such catalytic events arise naturally, it is anticipated that most men would not encounter such identity-challenging events very often. Scale Properties: Internal consistency: Factor analysis of the PCERQ sub-scales and Cronbach's alpha will each be consulted in determining the internal consistency of the PCERQ's subscales. Given that items in each of the scales were written to reflect cohesive conceptual themes, it is expected that the subscales of the PCERQ will have an acceptable alpha level of internal consistency. Indeed, it is necessary for the purposes of subsequent analyses relative to other variables for these sub-scales to have acceptable internal stability.  Anticipated Validity Properties Convergent and divergent validity: Inspecting the degree of a measure's correlations with conceptually-similar variables (convergent validity) and its lack of relationship with conceptually dissimilar variables (divergent validity) is a crucial step in understanding what a scale means. In terms of convergent validity, the PCERQ items were written to tap conceptualizations of responsibility, power, and marital self-efficacy / locus of control somewhat similar to those in the literature. What follows is a listing of corroborative measures used to cross-validate the degree to which PCERQ sub-scales relate to constructs that already exist: 1) N-power's convergent validity was anticipated to result in a significant correlation with Winter's measure of n-power and an inverse correlation with the Big Five personality trait of Agreeableness. Costa Jr and Widiger (1994) describe low scorers on Agreeableness as  Development and Evaluation of the PCERQ 29 "...cynical, rude or even abrasive, suspicious, uncooperative, and irritable, and can be manipulative, vengeful, and ruthless." (p. 3; emphases added). 2) Convergent validity of conflict ineffectiveness was investigated by examining correlations with Marital Locus of Control Scale (MLOCS; Miller, et al., 1986), strategies for managing conflict scenarios based on coded narratives, and the Rosenberg Self Esteem Inventory (RSEI; Rosenberg, 1965). Conflict ineffectiveness was also anticipated to be associated with moderate to high levels inadequate modeling of non-abusive conflict tactics, in the form of exposure to abuse tactics in childhood. 3) Convergent validity of Standards to be Non-Abusive and Exonerative Rationalizations, both generally related to responsibility, was anticipated to be demonstrated through correlations with Winter's measure of responsibility derived from coded text and the Big Five personality trait of Conscientiousness. Despite the fact that the PCERQ was expected to correlate with these various measures, it is important to mention that the PCERQ item content was written to narrowly and uniquely focus on conflict and influence in the intimate relationship. Hence, it was anticipated that these new sub-scales would result in low to moderate correlational patterns with measures that apply these constructs to more general interpersonal contexts, suggesting substantial unique variance in the new scale's constructs relative to these broader-based measures (e.g., measures of the Big Five traits). It was anticipated that relationship satisfaction would correlate with conflict ineffectiveness, standards to be non-abusive (inversely) and exonerative rationalizations; npower was not expected to relate to relationship satisfaction. As it was intended for the PCERQ scales to be related to psychological functioning in a circumscribed fashion, it was hypothesized that they were anticipated to be unrelated to socioeconomic status. With respect to divergent validity, the PCERQ sub-scales are intended to measure relationship-oriented psychological phenomena that ought not to be related to such demographic factors as age, cultural background, or socioeconomic status. On this basis, nil  Development and Evaluation of the PCERQ 30  relationships were anticipated between the PCERQ and these potential influences on behaviour. Criterion validity. A primary goal of this thesis was to investigate how well the PCERQ could measure abuse-facilitating psychological features arising during intimate conflict. This issue was evaluated by determining the PCERQ's capacity to account for significantly unique variance in two forms of intimate abusiveness relative to other variables. Regression equations were the means to examine this issue. Personality disturbance, social desirability, relationship satisfaction, and family of origin abuse have all shown relationships with intimate abuse in previous studies. It was surmised that the PCERQ would explain unique variance in reports of intimate abusiveness relative to these predictor variables, as its conceptual was anticipated to be relatively unique compared to these other variables. Of perhaps greater utility would be the possible manner in which the PCERQ could relate to the PAS, a measure of the abusive personality which has been shown to be robustly predictive of abuse. While the PCERQ sub-scales assess conflict-specific experiences, the PAS measures longer-standing traits that contribute to emotional activation and impulsivity during intimate discord. While it is known that the PAS will predict abusiveness, it is hoped that components of the PCERQ will enhance our understanding of how intrapsychic events relate to abusiveness. By entering the PAS into regression analyses prior to testing the PCERQ components, it will be possible to examine the temporal nature of the intrapsychic antecedents of abuse. The conflict ineffectiveness measure is expected to function as a significant main effect, and the interactions of n-power with standards to be non-abusive and with exonerative rationalizations are also anticipated to be useful terms. The main effects of those two responsibility-oriented scales may not retain significance (given the relatively poor track record of Conscientiousness as a predictor variable). In keeping with Winter's finding that the n-power's interactions with responsibility are the particularly salient predictive element involving n-power, it is not anticipated that n-power as a main effect will be a strong predictor of abusiveness relative to other PCERQ terms.  Development and Evaluation of the PCERQ 31  Method  Participants.  Only a brief overview of participant characteristics is provided here, as the following chapter describes them in much greater detail.  Undergraduate Dating Sample (UPS). This sample (n = 147) was comprised of young men in introductory-level psychology courses at the University of British Columbia. In order to optimize the degree to which data would be comparable to the other samples with greater relationship experience, only men who were currently in a heterosexual relationship of at least four months' duration participated. Men in the UDS tended to be almost 20 years of age (mean = 19.81, standard deviation = 2.0).  Treatment Group Sample (TGS). Data collected from a total of 54 men court-mandated to attend treatment groups for those convicted of (wife) assault. These groups typically occur in three-hour evening sessions, once per week for approximately 16 weeks. Group therapists and participants in various communities in southwestern British Columbia allowed the researcher to collect data from the men in their programs. Groups in the cities of Chilliwack, Maple Ridge, Abbotsford, North Vancouver, Vancouver, Coquitlam, and Port Coquitlam were involved. The average age of the men in the TGS was 38.08 years (8.6 year standard deviation).  Community Sample ICS). Men in this sample (n = 27) were currently involved in marital relationships and responded to recruitment advertisements in community newspapers. The average age of men  Development and Evaluation of the PCERQ 32  in this group was 37.56 years (10.4 year standard deviation). While it would have been optimal to randomly-select men for this sample from a population similar to the TGS on such variables as socioeconomic status and level of criminal justice contact, this was not possible in this research project.  Materials.  Measures Completed By All Respondents. 1) Power, Conflict Efficacy and Responsibility Questionnaire - 90 Item Version (PCERQ90; Appendix D-l). An item pool, written by the author, assessing concepts at the core of this research. Itsframeworkand rationale are described in Chapter 2. This questionnaire also contained the three items that combine to form a scale called the SPQ that was pilot-tested by the researcher in an earlier study with undergraduate dating couples. The SPQ was found to be strongly associated with partner reports of abusiveness and was the foundation of the conflict ineffectiveness notion elaborated upon in this dissertation. 2) Conflict Tactics Scale - Family of Origin Form (CTS-FOOF; Gelles & Straus, 1988; Appendix D-2). The CTS-FOOF is an 18-item measure on which respondents describe their exposure to conflict tactics in their home during their childhood. Subscales ask questions about tactics parents used with one another, and how conflict was managed between each parent and the respondent. 3) Conflict Tactics Scale - Marital Form (CTS-MF; Straus, 1979; Appendix D-3): On the CTS respondents indicate how often they have engaged in (and been the recipient or victim of) 18 different conflict behaviors with their partner in the last year, and also how often their partner has committed these acts towards them. For each dyad ("you to partner" and "partner to you") three subscales are formed by item aggregation: (a) Verbal Reasoning, (b) Verbal Abuse, and (c) Physical Abuse. Much in the way of debate and empirical study has been undertaken regarding the relative merits and shortcomings of self-reported incidence of  Development and Evaluation of the PCERQ 33  intimate violence (e.g., Hilton, Harris & Rice, 1998), with the prevailing tone questioning the wholesale validity of such an approach. The risk of inaccurate portrayals of one's own use of violence, a possibility with CTS-MF, is being considered in light of variables like social desirability that could potentially influence self-reporting. 4) Propensity for Abuse Scale (PAS; Dutton, 1995a; Appendix D-4): The PAS has been found to be a very useful tool in the measurement of personality factors relevant in the context of intimate abuse. The PAS is a 38-item scale that was constructed by identifying items from various personality questionnaires which correlated highly with partner reports of abuse. The measure is comprised of three orthogonal factors: (a) emotional lability, (b) experiencing symptoms of post-traumatic stress, and (c) shame-inducing paternal maltreatment during childhood. This measure is best seen as an indicator of long-standing traits associated with predispositions to aversive emotional arousal in the context of close relationships. As such, the PAS assesses respondents' proneness to experience a form of emotional reactivity and dyscontrol associated with impulsivity. 5) NEO-FFI (Costa & McCrae, 1989). This is a 60-item brief form of the larger NEO-PI measure of the Big Five personality traits. The authors report that this shorter form's five 12item subscales (Openness, Conscientiousness, Extraversion, Agreeableness, and Neuroticism) account for 75% of the variance in the convergent criteria of the larger instrument. 6) Balanced Inventory of Desirable Responding (BEDR; Paulhus, 1984; Appendix D-5). This scale is a 16-item abridged version of the 40-item BIDR, a measure of socially desirable response bias. This abridged version, created to shorten the psychometric battery, was enabled by the high inter-item consistency of the overall measure (Paulhus, personal communication). The BIDR measures two distinct forms of social desirability, namely Impression Management and Self-Deceptive Enhancement. Impression Management is the tendency to deliberately alter one's presentation to fit with ideas of normative behaviour. SelfDeceptive Enhancement is the tendency to be excessively generous or benevolent in views of  Development and Evaluation of the PCERQ 34  one's own behaviour, which serve a self-protective function; it can also be conceptualized as an indicator of narcissism. Eight items from each subscale were selected. 7) Marital Locus of Control Scale. (MLOCS; Miller, Lefcourt, Holmes, Ware, & Saleh, 1986; Appendix D-6). The 44-item M L O C S was constructed to assess the extent to which individuals had an internal versus external locus of control (LOC) for events in their marital relationship. High scores on the MLOCS reflect a high external/low internal L O C , whereas low scores indicate a low external/high internal L O C . The measure has four subscales: (a) ability, (b) effort, (c) chance, and (d) luck. 8) Green's Socioeconomic Status Ranking Scheme (Green, 1970). This ranking scheme quantifies participants' socioeconomic status on the basis of reports of their current and parental occupation and educational level. Composite SES was determined according to Green's method, namely SES = (.7 x employment score) + (.4 x education score). 9) N-Power Coding Scheme (Winter, 1973). This coding method was used to assess one's motivation toward power over others on the basis of narrative material. The narrative material, most commonly in the form of imaginative stories based on pictures, is evaluated for references to influence attempts or the desire to have an affect on others' behaviour or emotions. In terms of coding categories, the general "Power Imagery" category was chosen for the proposed study. Power Imagery encompasses references to influence attempts (everything from persuasion attempts to attacks), wanting to affect others' emotions, and attempts to teach or help outside of a circumscribed role. 10) Responsibility Text Coding Scheme (Winter, 1989): This measure evaluates responsibility from narrative material along five dimensions: respect for standards, concern for others, self-judgment, valuing interpersonal obligations and forethought / concern for consequences of one's actions. 11) Hypothetical Conflict Scenarios (Appendix D-7). This is a series of situations of intimate conflict, developed by the author, and being used for the first time. Respondents are asked to write descriptions of how they would react and respond to four hypothetical relationship  Development and Evaluation of the PCERQ 35  events that could give rise to conflict. Lastly, they are also asked to describe a recent conflict they have had with their partner. These narratives were examined by the author in an attempt to determine ways in which responses were varying in ways that could be relevant to the management and experience of conflict. Theoretical and therapeutic principles were also used in identifying a range of categories were that were scored from these narratives. Coding was done by the author, who was blind as to respondents' scores on other measures. The coding scheme appears in Appendix D-8 and include: (1) affective tone, (2) number of positive attributions regarding one's partner, (3) number of negative attributions regarding one's partner, and (4) number of potential (unprompted) solutions generated to the conflict scenario. The narrative responses were also coded with Winter's n-power and responsibility schemas. 12) Demographics Form (Appendix D - l 1). This form was used to collect information about a range of characteristics, such as age, cultural background, family of origin features, socioeconomic status and relationship perceptions.  Measures Completed By UPS Only. 13) Rosenberg Self-Esteem Inventory (RSEI; Rosenberg, 1965; Appendix D-9). The RSEI is one of the most widely-utilized measures of self esteem. It is a measure of generalized feelings of self-worth. It was employed, in part, as a measure concurrently relevant to the conflict ineffectiveness scale of the PCERQ.  Measures Completed bv the TGS and CS Only. 14) Dyadic Adjustment Scale (DAS; Spanier, 1976). This scale is a 32-item measure of marital satisfaction that has been used extensively in studies of relationship functioning and family violence. As this scale had only a peripheral role in the study it is not included in the Appendixes.  Development and Evaluation of the PCERQ 36  Narrative Stimuli and Response Forms flJDS Only). Five pictures were selected for use as stimuli in the generation of narratives used in cross-validating the PCERQ. The pictures were selected on the basis of being interpersonal in nature, relatively ambiguous as to the emotions, motivations, or nature of the interpersonal contact, and relatively unfamiliar images (e.g., well-known photos or paintings were not appropriate). The pictures were assembled into booklets in counterbalanced order so as to avoid any picture-order effect on story writing. Participants wrote their narrative responses on standard forms based on those used by Winter (1973). Instructions on each page asked participants to write a story in the space provided addressing who the characters were, what seemed to be happening, what the individuals were experiencing and what would happen in the future.  Procedure  Undergraduate Dating Sample (UPS) The recruitment of UDS participants was conducted in two ways. The first was by posting notices on the experiment bulletin board in the Department of Psychology at the University of British Columbia. Those who expressed an interest in participating by signing the posted notice were contacted by phone. The second manner of recruitment was by calling students who had completed forms in their Psychology course authorizing researchers to contact them to see if they wished to be involved in studies. UDS data was collected between September 1996 and February 1998. When contacted via telephone, potential participants were first informed that the study was only open to those who were currently involved in a dating relationship of at least fourth months' duration. The study was described as an experiment in two parts: (1) writing brief stories in response to pictures, and (2) completing an assortment of questionnaires  Development and Evaluation of the PCERQ 37  related to their dating relationship. Participants were informed that they would receive 1.5 credits toward their Psychology course grade in exchange for participating in the study. Individuals wishing to come in for the study after phone contact were assigned a time to arrive at the lab. Upon arrival, subjects were met by a research assistant (two female undergraduate research assistants conducted the UDS portion of the research). Participants were reminded of the general information about the study and were told that they could ask questions of the experimenter or even withdraw at any time if they were uncomfortable due to questions or procedures therein. In terms of actual experimental involvement, participants first completed the storytelling part of the study. Each participant was supplied with their own booklet and were told they have five minutes to write each of five stories; the experimenter kept time during this phase and prompted subjects to begin writing about each subsequent picture. After the five stories have been written, participants completed the questionnaire battery. Men completed the study in groups of up to six, depending on the availability of space in the laboratory and respondent availability. A thorough de-briefing concluded the experiment, following which participants were given their participation credit forms. A randomly-selected subsample of men were asked if they wished to return on a second occasion three months later to complete additional questionnaires and receive three dollars for their efforts. A total of 33 men consented to come on a second occasion for the re-test component of the study. The men completed a new demographic measure assessing change in their dating relationship since initial testing, plus the PCERQ and CTS. Coding of Narrative Data. The researcher was self-trained in Winter's n-power and responsibility methods, in accordance with published and required standards Winter has outlined elsewhere (Winter, 1973, 1989). The training involved learning the conceptual categories and practicing in the application and use by scoring published narratives for which there are answer keys. Upon reaching required standards of reliability in relation to these  Development and Evaluation of the PCERQ 38  answer keys, experimental protocols (both imaginative stories and narratives of conflict) were evaluated and scored.  Treatment G r o u p Sample  The researcher approached a variety of wife assault group treatment providers in greater Vancouver and inquired about asking men in their programs to complete the questionnaires. Typically, these groups operate on a contract basis with the British Columbia Ministry of the Attorney General. Upon approval from the therapists involved (acquired in every instance), the researcher entered treatment groups in weeks three to five of the 16week treatment programs. By way of introducing the study, it was made explicit to potential respondents that participation was strictly voluntary and independent of their participation or evaluation in treatment. Those men who were interested were given a questionnaire to take home. The researcher returned two weeks later to collect the completed work, individually debrief participants and offer a $12 stipend to those who completed the battery. Those who had not completed the questionnaires were given stamped envelopes to mail them to the researcher's laboratory at UBC. For those who mailed questionnaires to the laboratory, a cheque and debriefing form were mailedby the researcher. The breakdown of the number of men approached, number of who took questionnaires, and number who returned them was as follows: 250 approached (estimate), 160 distributed, and 55 returned. The researcher did not actively inquire as to why men did not initially wish to participate, but such factors as illiteracy, concern for protection of privacy, lack of trust and disinterest were spontaneously offered as reasons by some men. Data for the treatment sample was collected between September 1996 and April 1998.  C o m m u n i t y Sample  The CS was recruited through ads placed in local public libraries and the classified sections of local community newspapers around Vancouver. The ad indicated that, by calling  Development and Evaluation of the PCERQ 39  the laboratory number provided, respondents could participate in a study open to men married to, or co-habitating with, the same woman for a minimum of 1 year. The ad also stated that those who participated would receive $10 for completing an assortment of questionnaires relating to that relationship. The questionnaires, along with a stamped return envelope, were mailed to those who wished to participate after they heard more about the study from a research assistant over the telephone. Remunerative funds and a debriefing form were mailed once the completed questionnaires had been received at the laboratory. While the majority of men who wished to participate completed and returned the package, approximately 20% of those who received questionnaires did not return them.  Development and Evaluation of the PCERQ 40  Chapter IV  Descriptive Characteristics of the Samples  Prior to describing the development of the PCERQ and its application in the following chapters, it is first desirable to present a descriptive account of the men who provided the data for this thesis. While Tables 1 and 2 offer concise overviews of participants' essential background characteristics, the interested reader may consult detailed supplementary information in the various components of Appendix A.  Demographics. Family structure. The vast majority of the Undergraduate Dating Sample (UDS), 87%, reported they were raised in a two-parent family by their biological parents. In the Treatment Group Sample (TGS) there was a smaller proportion, but still a large majority, of men who were raised in two-parent families. The value from the Community Sample was similar to the UDS data. Men in the TGS tended to have more brothers and sisters than those in the UDS and CS. On a 9-point scale assessing amount of joy experienced in childhood, with 1 representing no joy and 9 representing very joyful, the median responses were 8 for the UDS and 7 in the CS and TGS. Socioeconomic Status. The modal annual income range of the family of origin of those in the UDS was $50,000 to $100,000; 32.5% of the sample fell into this range, and 27.2% fell into the $25,000 to $50,000 range. The modal family income in both the CS and TGS was $20,000 to $50,000 per year, with about 40% of each sample endorsing that range. Similar proportions in both samples fell above and below that modal range. In addition to being assessed within the categorical demographic questions described earlier, socioeconomic status (SES) was also evaluated in accordance with Green's (1970)  Development and Evaluation of the PCERQ  41 Table 1. Overview of Demographic Features bv Sample. Sample UDS (n = 147) Mean (SD) or n (%)  TGS (n = 48) Mean (SD) or n (%)  CS (n = 27) Mean (SD) or n (%)  Years of age  19.81 (2.0)  38.25 (8.6)  37.56 (10.4)  Raised by both bio. parents  128 (88%)  34 (69%)  23 (85%)  Born in Canada/USA  109 (74%)  43 (86%)  17 (63%)  Parents born Canada/USA  55 (38%)  37 (74%)  15 (55%)  English-onlyat home  93 (63%)  42 (86%)  19 (70%)  Length of relship (yrs)  1.39(1.1)  8.01 (8.1)  9.24 (7.3)  Socioeconomic status  68.33 (2.3)  57.10(7.0)  67.29 (7.0)  Number of arrests  .07 (.3)  3.82 (4.5)  .19 (.8)  Inter-parental V A  5.02 (4.8)  6.28 (6.2)  6.30 (4.6)  Inter-parental P A  1.12 (.6)  2.54 (4.8)  1.12(2.5)  Victim of parental V A  4.32 (4.5)  5.13 (6.0)  4.22 (4.2)  Victim of parental P A  1.67 (3.5)  3.05 (4.6)  1.56(2.8)  Perp. of V A to parents  5.21 (4.3)  5.44 (4.6)  4.94 (4.3)  Perp. of P A to parents  0.73 (2.5)  0.48(1.1)  0.26 (0.7)  Note: UDS denotes Undergraduate Dating Sample, TGS denotes Treatment Group Sample andCS denotes Community Sample, (n) denotes number of participants. (%) denotes percentage of sample. V A represents verbal abuse, P A represents physical abuse.  Development and Evaluation of the PCERQ 42  coding scheme. Values for education level generated by this system are: less than high school = 46, high school graduate = 53, 1 to 3 years of college = 62 and 4 or more years of college = 69 (those are average values, gradients exist within each category). Values for occupational type (again, with gradients within each range) are as follows: labourers = 42, service workers = 46, semi-skilled workers = 49, craftsmen = 52, clerical = 56, managerial = 59 and professional = 63. The mean (and standard deviation) for SES in each sample were as follows: 68.33 (2.3) in the UDS, 57.10 (7.0) in the TGS and 67.29 (7.0) in the CS. The constricted range of SES scores in the UDS, and the lower SES scores in the TGS compared to the other two samples, presented some limitations for subsequent analyses. It will be seen later that, because of this confounding of SES and abusiveness, within-samples statistics were conducted in order to examine the role of SES relative to other predictors of relationship aggression. Number of Arrests. There was minimal exposure to the criminal justice system in the UDS and the CS. Of the 147 participants in the UDS, 138 (94%) had never been arrested, 7 (5%) reported one arrest, and 2 (1%) reported two. No undergraduate participant had been arrested more than twice. Arrests were relatively uncommon among the CS, with 88% . reporting no history of arrest. One man had been arrested once, another four times. The average number of arrests in the TGS was 3.85 (4.5 standard deviation), with a mode of 1 (31 % of the sample). In the TGS, 25% of the sample had been arrested more than 5 times. Relationship Variables. As can be seen in Table 1, the mean length of current dating relationship in the UDS students was 1.39 years (standard deviation =1.1 years). Marital relationships in the TGS had a mean length of 7.41 years (standard deviation of 7.8 years), and in the CS it was 9.24 (standard deviation of 7.3 years). Appendix A - l provides an  Development and Evaluation of the PCERQ 43  overview of current relationship status. In the UDS, the vast majority of respondents did not live with their girlfriend, although 2 men were in commonlaw relationships. In the TGS, 12.0% were single, 51.0% lived with (were married or common-law) a partner, 37.0% were separated or divorced. Those no longer with their wife against whom the assault occurred had been asked to answer relationship-oriented questions based on that relationship. Men in the TGS who were currently involved were compared with those who were not through a Hotellings T-Square, testing a variety of variables to determine if the groups were potentially different: verbal and physical abusiveness to one's partner, and family of origin abuse victimization. The results demonstrated no difference between the groups (p > .90). This similarity allowed for the aggregation of TGS men within one grouping for the purposes of analyses. In the CS, 85% of the sample was married or common-law, with the remainder indicating single as their status (i.e., not living with their partner). Cultural Background. The three samples had somewhat different properties in terms of cultural backgrounds. In particular, there was more cultural variation (non-North American) influence in the UDS and CS than the TGS. While the majority of respondents across samples were themselves born in North America, the percentage with parents bora in North America was below and close to half in the UDS and CS, respectively. For the interested reader, Appendix A-2 provides a more detailed overview of cultural influence within samples than Table 1. In the UDS, 74% of respondents were born in North American, 22% in Asia and 4% elsewhere. The two most-common locations of parents' birth in the UDS were Asia (42%) and North America (38%). English was the sole language spoken in the home of 63 % of UDS growing up. While the majority of men in the CS indicated a North American heritage,  Development and Evaluation of the PCERQ 44  30% reported not solely speaking English in the home growing up. There was a relatively prominent European component to the CS, as compared to the Asian one noted in the UDS. The TGS was predominantly drawnfrommen born in Canada or the United States (85%) raised by North American-born parents (74%) who grew up in English-speaking homes (85%). The CS was comprised of some cultural variability as well. In order to consider the possible effect of cultural background on various subsequent analyses in the UDS, participants were grouped based on their linguistic background. Participants who spoke English exclusively in their home growing up were differentiated from those who spoke English not at all or only some of the time. A total of 54 participants (37% of the sample) did not speak English exclusively in their home growing up. Within this group, 46% were themselves born in Asia and 85% had parents from Asia. Although Asian extraction was the predominant feature of this non-English-speaking sub-group, 4 men (7%) of the sample had parents born in Europe. It will seen at various points in this thesis that this grouping was the basis for checking the possible role of cultural differences on test scores.  Exposure to Conflict Tactics in Family of Origin (FOO) Verbal abuse frequency levels. The reports of family of origin verbal abuse appear in Appendix A-3. In absolute terms, indices of abuse showed that the vast majority of participants reported most forms of verbal abuse occurred "at least once" in all family relationships. In terms of a coarse average, between 80% and 85% of the participants acknowledged witnessing, receiving and perpetrating verbally abusive tactics in their FOO relationships growing up. With respect to the parent / participant relationship, there appeared to be a trend toward being more abusive toward one's parents than they were in return. A M A N O V A that evaluated scores on FOO verbal abuse found no significant differences between the samples [ F (12, 394) = .64, p = .80)].  Development and Evaluation of the PCERQ 45  Physical abuse levels. The frequencies of physical abuse in participants' families of origin are reported in Appendix A-4. Across all samples the values were solidly in the lower end of the "at least once" range. There was also a trend within each sample for participants to report being the victim of physically abusive tactics in the parent / son relationship more often than they used such tactics toward their parents. A MANOVA that contrasted FOO physical abuse scores between the samples found a main effect for sample membership [F (12, 392) = 2.01, p < .05]. Univariate F-tests indicated this overall significance was caused by significantly higher "Father to You" physical abuse levels in the TGS, relative to the UDS and CS [F (2, 210) = 5.61, p<.01]. An overview of FOO abuse and its relation to this research. Those in all three samples witnessed marital relationships wherein both parents were relatively equal in their levels of verbal and physical abusiveness. There also appeared to be a trend toward inflicting more verbal abuse toward one's parents than was receivedfromthem, across the samples. On the other hand, there appeared to be a more physical abuse receivedfromone's parents than inflicted upon them. It was observed that UDS participants who grew up speaking English inflicted significantly more verbal abuse toward their fathers than those in the non-Englishspeaking group [T-Square = 12.25; F (4, 136) = 3.00, p < .02]; levels of being a witness and victim did not differ between the groups. Overall, the data is consistent with various normative surveys that have indicated abusive tactics within the family are not limited to clinical populations. Rather, the behaviours measured by the CTS are not unfamiliar to the experience of large tracts of the population at large.  Conflict Tactics in Respondents' Ongoing Intimate Relationship.  Perpetration of Verbal Abuse: Information on abuse within respondents' current relationships appears in Table 2. Approximately 90% of the men in each of the samples reported they had engaged in some degree of verbally abusive conduct toward their partner in the past year. There was a strong trend toward reporting more verbal abuse victimization  Development and Evaluation of the PCERQ 46  than perpetration in each of the samples, although this achieved statistical significance only in the UDS [t (144) = 6.72, p < .001). A one-way ANOVA that contrasted samples on perpetration of verbal abuse was significant [F (2, 218) = 34.62, p < .001]. Moreover, Scheffe contrasts revealed that the three samples were significantly differentfromone another (P<05). Perpetration of Physical Abuse. Relatively low levels of physical abusiveness in the last year were reported in the UDS and CS: 10% in the UDS and 4% in the CS. On the other hand, 71% of those in the TGS acknowledge some degree of physically abusive tactic in the last year. Those in the TGS and UDS reported being a victim of physical abuse more often than perpetrating it themselves [TGS: t (50) = 3.83, p < .001); UDS: t (143) = 2.79, p < .001]. The three samples were found to be significantly differentfromone another on infliction of physical abuse through Scheffe contrasts (p < .05) following a one-way ANOVA [F (2, 218) = 28.41, p<.001]. Results of analyses to test possible differences on measures of violence as a function of cultural background in the UDS indicated that culture was not related to levels of selfreported verbal and physical abusiveness [T-Square = 2.57; F (4, 140) = .629, p > .05]. This finding set the stage for subsequent analyses with the CTS data, allowing it to be used without concern that culture would be a problematic covariate. On the physical abuse "You to Partner" subscale of the CTS a subgroup of 4 TGS men had outlying high scores relative to the sample (scores of 13 or higher, 5 standard deviations above the TGS mean). The effect of their outlying scores on the TGS data is most obvious when consulting Appendix A-6. The mean level of physical abuse toward one's partner dropped to 2.38 from 3.50 with their exclusion. An inspection of the data indicated that these 4 most-violent men, relative to the other TGS men, showed particularly high levels of verbal and physical abuse by Father to Mother and Father to You, and very low levels of verbal and physical abuse by Mother to Father. It was decided that the TGS data would be  Development and Evaluation o f the P C E R Q  47  ON  00  O O  o o  o vd  o o  o o  CN  CM  -*'  o  O  o V")  C/2  Q  VD ro'  C/3  00  o S  o  VO  D T3 O  ON CN o  s  VO C/3  II  o VO  TD  0O  Q c/3  od  f-  CO  o  T3  o ro  o  © 00  CN CN O  o  o  o CN  O CO  CO  o  CO  CN 00 ro CN  Ov  —i  o  o o  o Os'  © vi  o ON  O + ^3 *2 ->  o ON o ©'  00  O  o o  ©'  ro ©'  CD Q  C/5  vd  c o  CN Os  1  VO  r-'  ro Os ro  in Os Os  ©'  tal e  Os  —'  £ 3 2  'o  e  SSI.  cx o  CN  >  '53  3  3  §  •s o CO  cx  2 "? 2 CN  Development and Evaluation of the PCERQ 48  considered in subsequent analyses both with and without these men. Further support for this decision stems from the research's focus on factors related to abuse by men presenting without signs of prominent personality disturbances (who tend to inflict abuse in less frequent and severe levels). As a result, subsequent tables will report data for the TGS without this subgroup of most-violent men, although any changes in results due to including these 4 mostviolent participants in analyses will be discussed in the text as necessary.  Links Between Abuse in Ongoing Relationship, SES and Relationship Satisfaction. Questions assessing participants' perceptions of relative commitment, attractiveness and disruption if the relationship were to end were considered measures of extrinsic power. Extrinsic power within a relationship is theoretically linked to seeing oneself as more attractive, less committed and less likely to experience disruption at the end of a relationship. In the UDS, no correlations between these measures and the Verbal and Physical Abuse Youto-Partner achieved significance (all p's > . 10). In the TGS, abusiveness to one's partner was correlated with believing relationship breakup would be more likely to detrimentally affect one's partner [verbal abuse: r (44) = .35, p < .05; physical abuse: r (44) = .26, p = .08]. N o correlations between abusiveness and these relationship factors were statistically significant in the CS. There were no significant correlation coefficients within the three samples between ongoing abusiveness and SES (all p's > .05; SES as indicated by Green's coding scheme). Additionally, groups were formed on the basis of whether the SES of one's partner was greater than one's own SES. Contrary to other studies showing more abuse among men with lower SES than their partners (Hornung et al., 1981), results in the samples of this study showed no significant differences (all p's > .05). Relationship satisfaction and abusiveness. Those in all three samples indicated their level of satisfaction with their relationship on a 9-point scale, from "not enjoyable" to "extremely enjoyable". A one-way A N O V A testing this scale across the three samples was  Development and Evaluation of the PCERQ 49  significant [F (2, 220) = 14.40, p < .001], and Scheffe contrasts showed each of the samples was differentfromone another (on level of enjoyment, TGS < UDS < CS). In the UDS, this value was inversely correlated with verbal abuse [r (144) = -.17, p < .05] and close to significance with physical abuse [r (144) = -.14, p < .10]. This 9-point measure was unrelated to abusiveness within the TGS, or within verbal abuse in the CS. Men in the TGS and CS also completed the Dyadic Adjustment Scale (DAS), a more comprehensive measure of relationship satisfaction. In the TGS, the DAS correlated at r (49) = .46 (p < .05) with the 9-point satisfaction measure; it correlated at r (27) = .23 (n.s.) in the CS. The DAS was uncorrelated with abusiveness in the TGS. In the CS, however, the DAS was very strongly and inversely related to the perpetration of verbal abuse [r (27) = -.61, p < .001].  Synopsis  It was the purpose of this chapter to outline some of the more meaningful and important facets of the samples used in the research. Some variation was noted within and between the samples, particularly with respect to cultural background, socioeconomic status and abusiveness. The UDS and CS were noted to be somewhat heterogeneous in cultural composition, with salient Asian and European backgrounds respectively; the TGS was homogeneously North American in its cultural background. It was shown that culture will not need to be considered a covariate in subsequent analyses of violence, as UDS sub-groups formed on the basis of culture were relatively similar in reports of FOO and dating abuse. Similar FOO experiences with verbal abuse were noted in each sample, however there was more exposure to physical abuse among the TGS men. The TGS was the most abusive sample, followed by the UDS, then the CS. It was noted that some under-reporting of abuse levels, attributable to socially desirable responding, was likely. The response style issue will be more fully evaluated in analyses later in the thesis. Intrinsic power and SES were noted to be unrelated to abuse within each of the samples. The were differentfromone another in  Development and Evaluation of the PCERQ 50  level o f current relationship enjoyment, but links between lack o f enjoyment and abusiveness were inconsistent.  Development and Evaluation of the PCERQ 51  Chapter V Derivation and Internal Properties of the P C E R Q Sub-Scales. This chapter describes the approaches used in constructing the PCERQ sub-scales. These sub-scales are the building blocks of the dissertation, as they ultimately will enable the testing of experimental hypotheses related to abusiveness in later chapters. Included in the first part of this chapter are the general theoretical and statistical considerations that guided the process of aggregating the PCERQ questionnaire items into sub-scales. This informs the reader about the purposes and criteria which governed item selection and the evaluation of interim outcomes on the way to arriving at 'final' sub-scale combinations of questionnaire items. This overview of the germane aspects of the sub-scale construction process is followed by specific descriptions of how each of the four PCERQ sub-scales were developed. The chapter concludes with descriptions of how the sub-scales were found to relate to one another and the extent to which these observed relationships fit with hypotheses.  The Item Generation Process Items for the PCERQ were written by the author to assess concepts that were relatively unexamined by other questionnaires already in use. Among the key areas of interest that were targetted in the writing of items were: feelings of self-control in relationship conflict, desire for power over one's partner, the value placed on restraint from aggression toward one's partner, and the tendency to make excuses for falling short of expectations for being non-violent. Other areas were of interest at the time of scale development but became too peripheral as the focus of this document took shape; items written to assess such categories as 'relationship interactions/events', and 'exposure to positive conflict role models' were not incorporated into the scale development process described in this chapter.  Development and Evaluation of the PCERQ 52  In terms of how the items were developed, several goals were paramount. First, it was considered essential that items reflect events within an intimate adult relationship (items were phrased so as to be gender-free and potentially completed by men or women in heterosexual or homosexual relationships). In other words, care was taken to focus specifically on the interactions and appraisals surrounding a committed romantic relationship and the conflicts and challenges that arise within it. Second, care was taken to develop items that were not especially vulnerable to social desirability biases; subtlety, using the sorts of 'mild' exonerative or minimizing views expressed by men in treatment groups for wife assault were incorporated into item content as appropriate. An example of this would be using relatively modest forms of control over partner being sought, as opposed to clearly inappropriate versions that would lead to minimal variability in responding. Third, items were typically written from positive and negative viewpoints to allow for reverse-scoring of items and scales that were not clearly and uniformly descriptive of either 'desirable' or 'pathological' functioning. The large PCERQ-90 battery of 90 questions that participants in this research completed was assembled based on pilot testing of a small group of items that formed the SPQ and revealed promising outcomes. Unfortunately, some lack of planning by the researcher as well as a lack of temporal and fiscal resources resulted in a lack of formal itemtesting prior to the distribution of the large 90-item battery. It would have been more desirable, for example, to have raters consider the degree to which items were 'prototypical' for the categories of interest, as well as how vulnerable they could be to response bias. That may have enabled a more strong set of items to be carried forward into this research. Nonetheless, the items were shared informally with various psychology graduate students familiar with the research area and personality and attitude measurement, and some adjustments were made to the item pool based on their feedback.  Development and Evaluation of the PCERQ 53  General Considerations and Methods of Sub-Scale Formation There were several issues that related to the construction of the PCERQ's sub-scales. In terms of the desired properties of the final instruments, it was considered optimal to generate scales that (1) corresponded strongly to theoretical conceptualizations of the variable of interest, and (2) represented a single unitary conceptfroma statistical perspective. The crafting of the sub-scales, then, proceeded by using factor analytic methods to reduce a large pool of theoretically-related items down to smaller cohesive clusters that met requisite statistical standards. An outline of these standards is presented in the following paragraphs. Thefirststep with each sub-scale was the selection of itemsfromthe large PCERQ item pool that would be considered for factor analysis. Items' theoretical relevance to the construct of interest, as determined by the researcher, was the sole criterion in deciding on whether they would be used in the analyses. Each sub-scale description that follows will explain the theoretical considerations that guided this initial item selection. Theoretical issues remained important throughout the item reduction process, in that decisons about whether items would be dropped in successive steps were weighed with both theoretical and statistical criteria in mind. The statistical side of the sub-scale development process can also be summarized in general terms. The procedure essentially involved conducting sequences of factor analyses with progressively smaller groups of items, with the goal of arriving at a sub-scale possessing an optimal level of statistical cohesiveness. It should be noted that the factor analytic procedures used the responses of all three samples (Undergraduate Dating Sample, Treatment Group Sample and Community Sample) in working toward thefinalsub-scale compositions . A statistical consideration weighed at intermediary steps prior to deciding on 3  The combined sample for the sub-scale creation analyses had approximately the following ratio: 66% Undergraduate Dating Sample, 22% Treatment Group Sample, and 12% Community Sample. Hence, while the sub-scales were generatedfromresponses of men who are or have recently been involved in intimate heterosexual relationships, the sample (relative to the 'population of interest' in the community as a whole) was younger and likely more  3  Development and Evaluation of the PCERQ  54 the final sub-scales involved evaluating items that failed to load on the main factor of interest in multi-factor solutions. It will be seen that items were dropped when their unrotated factor loadings were substantially higher on factors other than the factor of interest; items could remain within the solution if they loaded somewhat higher on the non-essential factor, provided they showed some relatively strong presence on the factor of interest. For example, if an item had unrotated loadings of .20 on the factor of interest and .65 on a rival factor, it was typically dropped (pending theoretical justification as well). On the other hand, if an item in that same analysis had unrotated loadings of .40 on the factor of interest and .45 on the rival factor, it was typically kept for the next step (in order to see how it would perform once statistically "distracting" items had been eliminated). Indicators of scale unidimensionality (Eigenvalue patterns and unrotated factor loadings) and internal consistency (Cronbach's alpha) were the main statistical criterion used in judging if a given item cluster was acceptable as a cohesive 'final sub-scale' or if more refinement was needed. Ideally, the final item cluster was hoped to result in the extraction of only one factor with an Eigenvalue greater than 1.0. On the other hand, it was also possible that the "Eigenvalue greater than 1.0" criterion could result in a the exclusion of theoreticallyrelevant items that had broken off into, apparently, a separate factor with such a decision rule. It will be seen that in some cases, where Eigenvalues were marginally greater than 1.0 for a second factor created primarily by one or two items but the item content of that second  intimately violent than a random sample of the 'population of interest' would be. In essence, there was a higher base rate of intimate abuse in this sample than one would expect in the 'population of interest' (e.g., all men in intimate heterosexual relationships). As a result, the items that 'rose to the top' of the factor analytic procedures and formed the PCERQ subscales may be biased by the over-representation of young men and aggressive men in the sample relative to the population of interest. It will be explained later that any potentially misleading effects of such a biased scale formation process was investigated by examining the scales' relationships with other variables within each sub-sample. Those analyses showed that the scales functioned in similar ways across each of the three samples in their associations with measures of personality and conflict experience. Ultimately, the sub-scales should be assessed (through factor analysis and correlations with other measures) to more completely assess their statistical stability.  Development and Evaluation of the PCERQ  55  factor was highly relevant to the theory underlying the sub-scale of interest, a one-factor solution was accepted. In terms of sub-scale internal consistency, one tries to construct scales that have coefficients of Cronbach's alpha in the .80's. This heuristic of pursuing sub-scales with high alpha coefficients was weighed against the unidimensional factor criteron described above. In other words, if a given item created a second factor with an Eigenvalue greater than 1.0 and resulted in a sharp loss of internal consistency, it was typically dropped despite possible theoretical relevance. The rationale for this was the need to create scales that were as reliable as possible; maximally reliable scales would lead to the most stable and strong pattern of correlational and regression outcomes relative to other measures in later analyses. If this facet of the project was a small part of a much larger multi-sample research program, such items would be included on the grounds that reduced alpha could be a sample-based artifact. However, given the smaller parameters of this dissertation, the alpha-reduction criterion had a more prominent role in decision making than it would have in a more expansive project. The final statistical consideration in creating the sub-scales was the running of an oblique (inter-correlated factors) factor analysis that forced a four-factor solution (one factor per sub-scale). Item loadings and factor groupings in the pattern matrix outcome were inspected to ascertain the degree to which each given sub-scale remained statistically intact when considered relative to items from the other sub-scales. Items that loaded much higher on a non-primary factor than on the primary sub-scale (e.g., loading at >.40 on an unexpected factor and < .20 on its anticipated factor) would be considered for elimination at this stage. This analysis also had undertones of being a confirmatory process, as it could offer evidence for the relative distinctiveness of the sub-scales generated through the process outlined above. It was not a true confirmatory process, however, which would test the initial solution's structure in an altogether separate sample. In review, a range of both theoretical and statistical considerations were utilized in the generation ofthe PCERQ sub-scales. Theory guided decisions about which items would  Development and Evaluation of the PCERQ 56  initially be used at the starting point with each sub-scale. Once items were selected, a variety of theoretical and statistical criteria (i.e., factor structure, Eigenvalues, items' relative unrotated loadings, Cronbach's alpha) were used to guide decisions in the process leading to the generation of final sub-scales. This framework runs through the sub-scale development sequences that follow for each of the four PCERQ sub-scales: (1) Conflict Ineffectiveness (CI), (2) Need for Power (NP), (3) Standards for Non-Abusiveness (SNA), and (4) Exonerative Rationalizations (ER). Following the development of sub-scales, items from those sub-scales were to be entered into an obliquely-rotated factor analysis which extracted four factors; this step was both a final step of sub-scale development and a pseudoconfirmatory endeavour . 4  I. Deriving the Conflict Ineffectiveness (CD Sub-Scale. In the initial step for building the Conflict Ineffectiveness (CI) sub-scale, items from the 90-item PCERQ pool were evaluated for theoretical correspondence to this notion. Those items touching on feelings of incompetency for non-abusively managing conflict, an inability to understand or manage one's emotion-based impulses during conflict, and perceiving one's partner (rather than oneself) as determining the course of conflict were included in the initial item pool. Experiencing conflict as a sped-up and uncontrollable phenomenon, lacking perceived options for dealing with disputes and having little confidence for non-abusively resolve discord also qualified items for inclusion in the pool. A total of 15 items (2, 8, 14, 20, 26, 32, 36, 40, 48, 50, 56. 62, 68, 72, 86) had content related to these various conflict ineffectiveness indicators and were used in the initial factor analysis. The first factor analysis (all factor analyses used the Principle Components extraction technique) of these 15 items yielded five factors with Eigenvalues greater than 1. A Scree The label 'pseudo-confirmatory' is used here as a true confirmatory factor analysis would be conducted on an altogether separate sample. In this case, the factor organization is being evaluated on the same sample from which the factors / sub-scales were developed, a lessoptimal procedure than a true confirmatory process. 4  Development and Evaluation of the PCERQ 57  plot of the solution, however, suggested that two factors were explaining the preponderance of the variance. From this, a second factor analysis was run that forced a two-factor solution. The two-factor solution (Eigenvalues of 5.32 for Factor 1 and 1.44 for Factor 2) showed that items #50 and #56 were the only two items with unrotated loadings higher on Factor 2 (.50, .64) than on Factor 1 (.29, .40). The valence of the unrotated loadings for items 50 and 56 indicated that Factor 2 was about feeling effective, which was apparently not simply the opposite pole of the Factor l's ineffectiveness (as evidenced by the separation of those factors). These two items were droppedfromthe item pool and a third factor analysis that forced two factors was run, continuing the search for a unidimensional scale. The third factor analysis (Eigenvalues of 5.13 for Factor 1 and 1.17 for Factor 2) found that only two items (40 and 48) had unrotated loadings higher for Factor 2 (-.56, .59) than Factor 1 (.44, -.34). These items were considered for omission from the pool based on their higher loadings on Factor 2 and relatively low Factor 1 loading (in the case of item 48 particularly). Considering the conceptual nature of those two strong Factor 2 items suggested they were about reflecting on or evaluating one's emotions during conflict, as opposed to describing the quality of the emotional experience per se, which was part of the focus of Factor 1 (i.e., comparing one's conflict behaviour to that of one's parents and an inability to track own emotions during conflict). While this "reflecting on emotions" material was of interest in the bigger picture of ineffective conflict management, it appeared to be relatively peripheral (statistically and conceptually) to the core notion of interest as it was emerging on Factor 1. Dropping those two items led to a fourth factor analysis. The fourth factor analysis yielded two factors (Eigenvalues of 4.87 for Factor 1 and 1.08 for Factor 2). Item #26 was the only item to have unrotated loadings higher on Factor 2 (.62) than Factor 1 (-.42). Based the fact that item #26 did not load primarily or particularly strongly on Factor 1, and that its content was in part a reversal of content (same question in the negative form) of an item already on Factor 1, it was droppedfromthe pool. From this step, a fifth factor analysis was conducted that produced a unidimensional solution  Development and Evaluation of the PCERQ 58  (Cronbach's alpha = .87). This was the Conflict Ineffectiveness grouping entered into the later oblique factor analyis of all the sub-scale items collectively.  n.) Deriving the Need for Power Sub-Scale. The first step in working toward the N-Power (NP) sub-scale involved selecting items from the 90-item PCERQ pool that were theoretically akin to the n-power concept espoused by Winter (1973). To qualify, items had to have content related to the pursuit and/or exercising of influence over their partner's behaviour and/or emotional state. Items that related to perceptions of one's partner as easily influencable or justifying influence attempts, as well as items showing a keen sensitivity to being influenced oneself, were also part of the initial pool of items. In total, 14 items of the 90 PCERQ items fit in some notable way with these requirements (1, 7, 13, 19, 31, 37, 43, 45, 49, 55, 61, 67, 79, 85). A first factor analysis with these items extracted four factors with Eigenvalues greater than 1.0, although a Scree plot suggested three factors were accounting for most of the variance. From this consideration a second factor analysis was run that forced a 3-factor solution, and unrotated item loadings were examined in order to make conceptual sense of the factor structure. Factor 1 of this solution was built around items 1, 45, 37 and 55: item content reflecting a desire to influence, guide and teach, plus viewing one's partner as somewhat weaker than oneself. This content was very close to the theoretical view of N Power being sought in the dissertation. Factor 2 was anchored by items 7, 19, 31, and 67. These items and the valence of their unrotated loadings described a de-valuing of seeking influence and dominance. These items, which had been written as 'reverse' items of n-power, were apparently not functioning quite in that way. If they had simply been the reverse of 'seeking influence', they would have loaded on Factor 1 or at least correlated with it (which they did not, when sub-scales of Factor 1 and Factor 2 items were computed). Factor 3 was comprised primarily of item #43  Development and Evaluation of the PCERQ 59  ("enjoy seeing partner's reaction to what I do"). The remainder of items showed some loading on Factors 1 and 2. As Factor 1 was closest to capturing the desired outlook on an N-Power construct within a relationship context, work was undertaken to isolate the items that could form a unidimensional and effective N-Power scale with those items at its core. The items most unrelated or orthogonal to Factor 1 (i.e., items that loaded on Factors 2 and 3) were omitted in a third factor analysis designed to investigate what would happen to items that had unrotated loadings on all three factors. Hence, items 43 (Factor 3) and 19, 7, 31 and 67 (Factor 2) were omitted from a third factor analysis. The third factor analysis resulted in two factors with Eigenvalues greater than 1.0 (2.56 and 1.31). Items #13 and #61 loaded strongly on the second factor (-.71, -.47) without comparatively strong Factor 1 loadings (.32, .03), and were therefore considered for elimination from the item pool. While these items involved commanding and exercising authority over one's partner, they showed virtually no relationship with the items on Factor 1. Given the items loading on Factor 1 were felt to capture a n-power notion not statistically enhanced by items #13 and #61, those two items were removed from the N-Power sub-scale pool for a fourth factor analysis. The fourth factor analysis yielded two factors with Eigenvalues greater than 1.0 (2.49 and 1.07). Item #85 was the only one to have an unrotated loading higher on Factor 2 (.65) than Factor 1 (.49). Given Factor 2's Eigenvalue was only marginally greater than 1.0, plus the fact that all items (except #85) that showed some loading on Factor 2 had greater unrotated loadings on Factor 1, as well as the observation that item #85 also loaded relatively highly on Factor 1, a single-factor extraction was forced. In essence, a single-factor solution that included the guidance-oriented content of item #85 seemed reasonable and optimal. This grouping was considered the tentative N-Power sub-scale (Cronbach's alpha = .69), subject to re-evaluation in the four-factor oblique factor analysis that examined items from all four sub-scales relative to one another simultaneously.  Development and Evaluation of the PCERQ 60  DJ.) Deriving the Standards to be Non-Abusive (SNA) Sub-Scale. The notion of responsibility is of particular interest in this dissertation due to its potential interaction with need for power in relation to abusiveness. Specifically, Winter (1984) demonstrated that need for power is not inherently a problematic interpersonal trait; rather, he found its combination with low responsibility to be the source of problematic behaviours associated with dominance. He operationalized responsibility as internalized standards of respectfulness and valuing rules of conduct. That trait will be applied more focally in this research with respect to having standards to be non-abusive and to engage fairly in conflicts with one's partner. The 90-item PCERQ array was inspected so as to identify questions with content related to non-abusive standards for guiding one's conflict behaviour. Items were initially considered relevant that had content dealing with beliefs about right versus wrong ways of resolving conflict, guidelines about non-abusive behaviour, beliefs about the importance of maintaining self-control and intentions of monitoring one's actions during conflict. A total of 9 items (5, 65, 23, 53, 24, 66, 84, 89, 54) that had been written to assess this dimension were accepted for the factor analytic process. The solution to the first factor analysis contained three factors with Eigenvalues greater than 1.0 (Eigenvalues of 2.21, 1.71, 1.07 respectively), consistent with the interpretation of the Scree plot. Inspection of unrotated factor loadings, however, found that no items were predominantly loading on Factor 3 relative to the other factors. The highest unrotated loadings on Factor 3 were for item #89 (unrotated loadings of -.54 on Factor 3 and -.52 on Factor 1) and item #54 (unrotated loadings of -.50 on Factor 3 and .54 on Factor 1). No clear conceptual meaningfulness appeared to underlie the third factor, and it also had a small Eigenvalue relative to the first two factors. These considerations led to the forcing of a two-factor solution in the second factor analysis. The two-factor solution found that items loading most uniquely on Factor 1 (84, 23, 89, and 54) related to endorsing personal accountability for one's own behaviour, particularly  Development and Evaluation of the PCERQ 61  being non-abusive, in conflict. Items most strongly associated with Factor 2 (items 66 and 65 loading at .67 and -.58 respectively) related to intentions to monitor one's conduct during conflict; they also loaded moderately on Factor 1 (.45 and -.32 respectively). Item 66 loaded strongest on Factor 2 and its content was somewhat conceptually peripheral to the the 'standards of accountability' theme of Factor l's items; on this basis, it was removed from the pool. Item #65 was left in because its content was felt to be conceptually imperative to notions of having personal rules that guide one's conduct. A third analysis was run with the remaining items that forced a solution of not more than two factors. The solution of this third factor analysis was comprised of two factors with Eigenvalues of 2.12 and 1.34. It was noted that the three items (54, 23 and 5) showing the strongest loadings on Factor 1 showed minimal loading on Factor 2 (Factor 1, Factor 2: .61, .00; .62, .00; .56, .30). On the other hand, the three items (89, 84 and 24) that loaded stronger on Factor 2 than Factor 1 also loaded at least moderately on Factor 1 (Factor 1, Factor 2: -.54, .45; -.54, .48; .40, -.46). Hence, Factor 1 contained items assessing the concept closest to the 'non-abusive standards' notion corresponding to the notion of responsibility/standards put forward by Winter. Alternatively, any conceptual distinctiveness of Factor 2's item content was not apparent (other than as tapping notions somewhat similar to Factor 1). Despite the fact that the Eigenvalue for Factor 2 was 1.34, suggesting it was somewhat unique statistically, it became reasonable to consider the possibility of accepting a one- rather than two-factor solution for conceptual reasons. A final part of deciding whether or not to treat the item group as one factor involved assessing changes in the Cronbach's alpha coefficient with and without the items that were loading strongly on Factor 2 (items 24, 84, 89). The Cronbach's coefficient for the larger version of the sub-scale (8 items that included 24, 84, and 89) was alpha = .59, with an average inter-item correlation of r = .16. The shorter version of the scale (5 items excluding 24, 84 and 89) had a Cronbach's alpha = .57 and an average inter-item correlation of r = .22. In the interests of maintaining as large a  Development and Evaluation of the PCERQ 62  group of theoretically-relevant items as possible, and given the lack of psychometric improvement in the smaller 5-item version (other than the marginal increase in inter-item correlation coefficient), the larger 8-item version of the Standards to be Non-Abusive (SNA) sub-scale was tentatively accepted pending the outcome of the larger obliquely-rotated factor analysis of items from all of the PCERQ sub-scales.  TV) Deriving the "Exonerative Rationalizations" Sub-Scale The main idea for this scale was rooted in Balshaw's (1994) finding that many previously-violent men refraining from aggression in close relationships had come to consciously adopt a strong resolve to live according to personally-essential non-abusive values at all times. She found these men reported an explicit and uncompromising resolve to be non-abusive toward their partner. Indeed, this outlook on their own behaviour was one of the most pronounced characteristics of men who had made the change from a violent to a non-violent way of relating in their intimate relationships; it appeared to be of foremost importance in how they conducted themselves in their intimate relationships. This sub-scale, labelled Exonerative Rationalizations (ER), will attempt to measure the strength of respondents' intentions to be non-abusive during conflict and not to excuse themselves from adhering to those intentions. The idea here comes from commentaries on domestic violence research (by various researchers themselves) that have expressed concern about the lack of research regarding abusers' accountability for becoming aggressive. In other words, despite all the aversive components of an intimate problem, why do some men elect to be violent and others refrain from it? This sub-scale is intended to highlight some of the mitigating evaluative processes that may lead a man to chose aggression over non-violence even when he knows it is wrong and would otherwise claim a preference to be non-violent. In sum, the scale is about living by one's non-violent intentions, regardless of how challenging it may be to do that.  Development and Evaluation of the PCERQ 63  The PCERQ item pool was examined for questions relating to the notion that, no matter what, remaining non-abusive was of primary importance to one's experience, options and motivations during intimate conflict. A group of 7 items (18, 28, 29, 60, 77, 78, 90) were noted to capture aspects of this concept. They were entered into a factor analysis; a twofactor solution was indicated by the criteria of Eigenvalues greater than 1.0 (Eigenvalues of 2.11 and 1.08 were observed) and a Scree plot. Factor 1 was the primary factor for the items that best-captured the 'unwavering non-abusive disposition/intentions' desired within the final scale (items 60 and 77 had -.77 and -.64 on Factor 1, -.05 and -.19 on Factor 2). Factor 2, anchored by item #90 (unrotated loading = .73), appeared to be about not feeling culpable for one's actions, yet it was unrelated to Factor 1 (unrotated loading = .00). Item 18 also loaded predominantly on Factor 2 (unrotated loading = .57) as opposed to Factor 1 (unrotated loading = -.26) and was omitted from subsequent analyses (also, re-appraisal of its content, positing abuse had already occurred, was felt to contaminate the self-regulatory tone of Factor 1). The second analysis, utilizing the remaining 5 items, resulted in a single-factor solution. Inspection of by-item properties in the solution showed that item #29 had a relatively weak unrotated loading (.41) relative to the other items (all greater than .60), perhaps due to its overly general wording (it was not as tailored specifically to intimate conflict as the other items). A final step in considering the possible exclusion of item #29 related to its effect on the sub-scale's internal consistency. Including item #29, Cronbach's alpha = .63 and the average inter-item correlation was r = .26. Excluding item #29, the Cronbach's alpha = .66 and the average inter-item correlation was .32. Importantly, item #29's correlations with other items were uniformly low and below the average. Given this indication of its feeble association with the Factor 1 items it was dropped. The interim E R sub-scale contained 4 items as it was entered into the four-factor obliquely-rotated factor analysis with other sub-scale items.  Development and Evaluation of the PCERQ 64  V. The Final Step of Sub-Scale Construction: An Obliquely-Rotated Factor Analysis. The preceding sections described how each of the four PCERQ sub-scales were developed. All sub-scales were developed in isolation relative to others, in that items were classified as to their apparent theoretical meaningfiilness and statistical properties of those item sub-groups were evaluated independently of the other sub-scales. The next stage of instrument development involved considering the statistical uniqueness of the subscale items relative to one another. This strategy involved entering all items from the final stage of each of the sub-scale processes outlined above into a factor analysis forcing a four-factor solution. An oblique rotation (positing intercorrelated factors) was used, consistent with hypotheses that that the factors would be related to one another. Optimally, it was hypothesized that each of the four factors generated in this solution would be comprised of items from those same groupings identified through the four processes above. While it was possible that items could load strongly on more than one factor (given the theoretical overlap among the concepts of interest), primary item loadings were anticipated to highly correspond with the sub-scale groupings condlucded from the previous stage of the research. Any items not loading substantially onto their theoretical factor of origin but loading primarily and strongly on a different factor were to be considered for elimination. In sum, this factor analysis was a final step in sub-scale construction, as it tested the notion that items from the various subscales would remain within those groupings in a factor analysis utilizing a larger and more conceptually diverse item pool. The items of the four sub-scales (CI, NP, SNA, and ER) generated from the analyses above (a total of 29 items) were entered into a Principle Components factor analysis. It was specified that the factor analysis would extract a four factor solution which would be subjected to an oblique rotation (the degree of obliqueness was set at the maximallyintercorrelated level by the default setting of the "Oblimin" setting in the SPSS-X for Windows Version 6.0 program). The correlation matrix was acceptable for the purposes of factor analysis, as indicated by various preliminary descriptive statistics requested.  Development and Evaluation of the PCERQ 65  Specifically, the determinant of the correlation matrix was .0000850, suggesting item multicollinearity was not a substantial problem (which would have been indicated by a value closer to .0000000). Also satisfactory were the Kaiser-Meyer-Olkin indicator sampling adequacy yielded a coefficient of .86 (values greater than the .70's are considered desirable) and Bartlett test of matrix sphericity (coefficient = 1963.69, p < .001). The obliquely-rotated pattern matrix of the solution appears alongside the item subscale content as reported in Tables 3 to 6. The most important overall outcome was that all but two items loaded primarily on the same factors as other items from their anticipated subscale. Essentially, the four sub-scales derived through the processes described earlier remained intact as distinctive factors following the oblique rotation of this factor analysis. The two weakest NP items (#49 and #55) loaded more strongly on the CI factor than the NP factor and were dropped from the NP scale. On the basis of this oblique factor analysis final sub-scale composition was determined. Due to the stability of sub-scales in the form of factors when all items were considered simultaneously, it was decided that all sub-scales (with the exception of NP) would have the same form as they did at the termination of their respective derivation processes as outlined earlier. The NP sub-scale had its two weakest items omitted, and its alpha coefficient dropped from .69 to .65.  Relationships Among the P C E R Q Sub-Scales Intercorrelations among the scales are presented in Table 7. Overall and as anticipated, there was a moderate degree of association between the sub-scales. While the (CI) sub-scale correlated strongly and in the anticipated manner with Exonerative Rationalizations [ER; r (221) = .40, p < .001], it showed a strong and unexpected relationship with N-Power [NP; r (221) = .36, p < .001). The NP sub-scale was also associated with E R [r (221) = .33, p <001]; as anticipated NP was unrelated to Standards to be Non-Abusive (SNA). The two responsibility-oriented scales, E R and SNA,  Development and Evaluation of the PCERQ  66  <  o o  o  vo o  o  vo o  o o  vo  o  0\  oo  r-  Ov  S  Ov  o  C/3 C N  vo  vo  00  v£> u-)  vb  II II  (11 *=>  o  a o o  1o  u o-  ccS  2  CO  e o  ca  Q.  6  S3  •a  •J3  s  P  o e £ c8  O  •a  i* 0)  tj o. .a  <3  o u  C8  JO  a  a.  !> cn  ts  )nfli  «  O  11  O  Pi  w u  «  ~ £ cn  s  00 VO  of  1)  cn ~~ & c3 3 J3  3  43  <2  T3 S3  tf 3  oo  •§  S  CL)  3 * cC <u H C  TO Q  8  •§  O  s-  ta> s  eg  e o  22 o  IS  ft a  y i—i ° I  [5 8  .8  €  CL)  6  •a CD  e  CD  o 0)  VO cn  <N  VO  VO  00  Development and Evaluation of the PCERQ  67  cu >  o  vo VO  < CO  o\ o  o  fN VO  VO  tN tN  o  tN  CN  CN O  VO O  o  o tN  O CN  O  VO O  VO O  VO  o  3  cu 3  a  cd  CH  1  s o o <u  *a  <U  JS U HO  -*-» CO  O  cu  o  u c o  -»-»  T3  O _S >< <L>  3  <u  1-  e CU  O JS  o  cci CD  o  .£  cs  S  o 3  GO  1o o  .ts £  T3  ti  U  CH  CH  43 GO  J2 "3  £ s  3  CO  O- co  Ki  o  2 .3  s 2 o £  CD  a  3  .3  S3 5  O GO  cu  2  O  £  o  PH  oo  JS  *  CU <+H  •S * h-H C H  2  ^  «J  co .TS C 3  g cr TO J Q  "3u +> cu o JS s  £ £  3 J3  s  o  t+H  o  .3 O 3  cu co  3 o  O O  <U  >>  o (U  1  CH  CH  CU  s o co O T3 cu  s  CO CO CCJ  u  CU  3  co T3  <U  o ccj  -4-*  3  •a  a a  CO I-.  ±3 <O *5  g  §  ti — s i c3  co =3  > >  GO '+J  CU  ?> -3  3  cu  CU  tJ  CH  3  C H ^  ^ 3  pi  w o  cu  3  -5 ts >,  co co  CU  CH  T3 5o  c3  cu  cd  a  -O  CA  <U  Is  I-H  4=  <U  O JS  .3  CO  o  00  3  -a  r--  <u  CO  3  CU  ca  •a  CH  S  3  "5 c  3 '-3 cu a  Development and Evaluation of the PCERQ 68  o  9  e  o  PH  T3  CN  s ii  vo CN  vo  00  CN O  CO  o  00  OV  VO  C/3  CN  •Sl-  -3-  00 ©  av ©  o  3  .2'  CN  o  O3  CN  CO  O  G O  b  O  T3 C  u c o  U  13 o  C/3 -O 3  (U _> '35  3 •  S3  o  u o at  e C/3  S  §w o  ro CN  00  ro  OV  00  CN  VO  Development and Evaluation of the PCERQ 69  tN  oo  Pi  <  VO  VO  CN o  VO CN  O  00  in  VO o  CN  VO CN O  CN  O CN  CZ3 o  OV CN  CN  cn CD  e o o  00  s  o o  X  s  e " x5  CD  G  xo  cn  cn CD  <H-H  O  cd CD  XI 00  cd  c  ts  o  x+-i> G  a. xo <u X  o <u S .15  II  -4-»  <t> X  t-i cd  cd  e u  C  s CD «  9  w u  cn CD  O T3  C4H  O 60  CD cn  cd  O  cn CD  So  "S +-> 3  s °  a, x c I—t "O cn CD — x a. HH  t-r--  00 CN  s  o  11 X  °  cn CD  s CD > CD  oo oo  2 ^ cd cn  00 B  CD  CD  X  CD  B o o  e CD  a  cn CD  O  cn  CJ  T3  o2 cn 3 o CD  1 3  •-O3 XS  o 9/ til PH  3  0 2  IH  o  'Z °< 3 H H  oo  G  cn • —•  o  vo  Development and Evaluation of the PCERQ 70  were uncorrelated with one another. The pattern of intercorrelation results was largely consistent with hypotheses, although the unanticipated correlation between CI and NP was the exception to such a general summary. In essence, the sub-scales' combination of overlap and uniqueness relative to one another is a desirable feature suggesting they measure different facets of the psychological experience of conflict.  Temporal Stability A sub-group of 33 UDS participants agreed to return for a second testing session (Time Two) three months after they initially (Time One) completed the questionnaire battery. At Time Two these men completed a smaller assortment of measures, including the PCERQ. Results from Time Two were analyzed relative to Time One to determine if the PCERQ subscales were stable over time. Hypotheses predicted that the new measures would be relatively invariable over time. The correlations between sub-scales at Time One and Time Two were as follows: (CI: r = .75), (NP: r = .84), (SNA: r = .59), and (ER: r = .46). T-tests of sub-scale means were conducted to probe for possible differences between scores at Times One and Two. The CI scale scores were stable across the 3-month interval (p > .50), whereas the three other scales showed signs of fluctuation over time. The results indicated the following trends: E R scores decreased slightly (p < .10), NP scores increased slightly (p < . 10) and SNA scores decreased (p < . 15). The possibility that the responses of 9 subjects no longer with their Time One girlfriend at Time Two could be affecting the outcome was considered by omitting those 9 men from the re-test data. The only change was the increased stability of the ER sub-scale (p > .40) when considering only subjects still with the same partner as at Time One. These analyses of the PCERQ's temporal stability found that, despite being correlated at Times One and Two, there were signs that the NP, E R and SNA sub-scales were subject to variability in responding over time. That these measures may be somewhat vulnerable to the  Development and Evaluation of the PCERQ  71 Table 7. Intercorrelations Among PCERQ Sub-Scales.  1  2  3  4  1 CI 2N-P  .35***  3 SNA  .00  .01  4 ER  40***  .33***  -.08  Note: CI = Conflict Ineffectiveness, N-P = Need for Power, SNA = Standards of Non-Abusiveness, ER = Exonerative Rationalizations. ***p<.001. (n = 221).  Development and Evaluation of the PCERQ  72  passage of time is most reasonably viewed as due to error, incomplete measurement of the construct of interest or changes in appraisals of self within the relationship; it cannot be concluded that the changes noted reflect trends within relationships over time. This finding of general temporal stability for the measures offers modest support for the hypothesis which posited that the variables of interest would be relatively unchanged without substantial personal or social challenge. It could be that such events occured for some men to the degree that temporal differences emerged.  Summary of Sub-Scale Derivation and Internal Properties. Against the theoretical backdrop of the psychology of intimate violence described earlier in the thesis, this chapter provided an account of how the PCERQ sub-scales were assembled. The sub-scales measure four different concepts of central importance to the research being conducted: Conflict Ineffectiveness (CI), N-Power (NP), Standards of NonAbusiveness (SNA) and Exonerative Rationalizations (ER). Each of the sub-scales were developed by isolating theoretically-relevant items from an initial 90-item pool and, from that point, by eliminating items of unsatisfactory theoretical and/or statistical properties. This led to the creation of four unidimensional sub-scales. A final test of the sub-scales' integrity was conducted in a factor analysis that extracted four factors and subjected them to an oblique rotation. Inspection of item-factor loadings in this analysis revealed a high degree of sub-scale integrity and distinctiveness; that is each of the sub-scale factors essentially retained the items of which they had been comprised prior to the factor analysis. Once the scales had been constructed, their various psychometric properties were investigated. The CI sub-scale was the most internally consistent of the sub-scales (alpha .87), while the other scales had coefficients in the .60 to .70 range. This pattern of internal consistency outcomes cautions against unqualified acceptance offindingsinvolving the NP, E R and SNA sub-scales (including interaction terms) that will presented in subsequent chapters. Investigation of test-retest sub-scale means indicated that the CI was also the most  Development and Evaluation of the PCERQ 73  temporally stable, whereas the other scales showed signs of change over the course of a three-month interval; correlations of Time One and Two sub-scale values were strongly significant for all sub-scales. The four PCERQ sub-scales showed an anticipated pattern of intercorrelations with one another, with the exception of the unexpected correlation between CI and NP, and between NP and ER.  Development and Evaluation of the PCERQ 74  Chapter V I  Concurrent Validity of the P C E R Q Sub-Scales  This chapter considers how the PCERQ sub-scales relate to other variables, so as to determine their degree of overlap and uniqueness relative to other concepts. These investigations of concurrent validity are an essential part of assessing the new scales' meaningfulness and utility. The analyses presented in this chapter discuss the data in terms of an aggregated sample combining the UDS, TGS and CS, rather than discussing each of the three samples separately for each analysis. The rationale for this is rooted in the essential similarity of correlational relationships between key variables across samples. In other words, despite mean differences between the samples on such variables as age, number of arrests, personality measures and levels of violence experienced and inflicted, the basic correlational patterns among core variables were nonetheless relatively similar within each of the samples. Where there are notable differences in correlational relationships as a function of sample, they will be addressed in the text. Description of concurrent validity and related hypotheses. The concurrent validity of the PCERQ sub-scales is a function of how much they statistically overlap with conceptually similar and dissimilar variables. As the PCERQ sub-scales are measures of primarily psychological features, they should be essentially unrelated to demographic features not linked to conflict management (e.g., age). On the other hand, the PCERQ sub-scales could correlate with features that are somewhat associated with the experience and interpersonal expression of emotions (e.g., the Big Five traits). Examining the strength of these relationships enables conclusions to be drawn regarding the PCERQ's uniqueness relative to existing concepts.  Development and Evaluation of the PCERQ 75  The P C E R Q in Relation to Demographic Variables. This passage considers how the PCERQ sub-scales relate to demographic influences such as age, marital status, cultural background socioeconomic status (see Appendix B-2). As hypothesized, age was uncorrelated with the PCERQ scales within samples. In terms of marital status, two sub-groups of the TGS and CS were formed on the basis of current marital status [currently involved (married or common-law) versus separated or divorced]. These two groups differed on a Hotellings T-Square test [T-Square = 14.89, F (4,67,) = 3.56, p < .05] that compared the four PCERQ sub-scales simultaneously. Post-hoc inspection of the individual sub-scales revealed that the overall significance was attributable to higher Conflict Ineffectiveness (CI) scores in the separated / divorced group [t (71) = 2.49, p < .05]. A tendency toward higher scores in the estranged group was also observed on the NP sub-scale [t (71) = 1.89, p < .10]. The finding that estranged participants had less efficacy for managing conflict makes conceptual sense, but this finding cannot serve as proof that that ineffectiveness is a causal source of estrangement; rather, ineffectiveness co-varies with estrangement. To consider the possible influence of culture on the PCERQ sub-scales, two groups of UDS subjects were formed on the basis of linguistic background to consider the influence of cultural background. Those with North American-born parents who spoke only English at home (n = 52) were contrasted with participants with Asian-born parents who spoke no English at home (n = 28) on the PCERQ sub-scales . These sub-groups did not differ on the 5  CI or NP sub-scales. Hypotheses regarding cultural differences on PCERQ dimensions were Results of a multivariate Hotelling's T-Square showed an overall difference between the two groups across the four measures [T-Square = 29.48, F (4, 74) = 7.08, p < .001]. Univariate ttests showed a significantly higher ER mean score among the Asian sub-group [t (78) = 3.95, p < .001], with SNA scores modestly higher among this group as well [t (78) = 2.11, p < .05]. These results indicate that, while the UDS participants from the Asian sub-group placed greater importance on acting non-abusively, they also had a cognitive set in place descriptive of inconsistently living in accordance with those standards. Non-Asian UDS subjects had lower standards for non-abusiveness, but were less accepting of aggressive violations of, or exceptions to, those standards. 5  Development and Evaluation of the PCERQ 76  not generated due to limitations of how culture was measured, as well as a dearth of research in the area of cultural variation in conflict management that prevented the formulation of especially meaningful or testable hypotheses. These culturally-oriented findings are presented primarily as descriptive information, and suggest ways that conflict may be researched in light of cultural variation in the future. Correlations between SES and the PCERQ scales were probed within each sample: no significant associations were noted. While the number of arrests was correlated with the NP sub-scale in the TGS [r (49) = .32, p < .05], the PCERQ sub-scale were otherwise uncorrelated with this variable within samples. Family of origin structure was considered by forming two groups on the basis of family type [two-parent (n=192) versus all others (n = 35)]. The Hotelling's multivariate test of the two sub-groups found mean differences [F (4, 222) = 3.77, p < .01]. Post hoc testing showed that those raised in two-parent families reported greater efficacy for managing conflict [CI sub-scale: t (43.97) = 2.80, p < .01] and higher standards to be non-abusive [SNA sub-scale: t (47.50) = 2.28, p < .05]. These results are presented for their descriptive value, as they were not anticipated by hypotheses from any particular theoretical or empirical foundations. In sum, with respect to demographics and the PCERQ scales, the main areas of association were: (1) estranged respondents in the TGS sample had higher scores on the CI sub-scale, (2) participants in the UDS of recent and strong Asian cultural identification had higher SNA and ER scores, and (3) participants raised in two-parent families reported more functional self-appraisals in the form of lower CI and higher SNA scores. Thefindingthat currently estranged men reported being less effective in conflict than those still with their partners makes conceptual sense. The outcome showing higher SNA and ER scores among the Asian sub-group of the UDS can be related to personality research suggesting that some Asian cultures may report lower levels of Achievement Striving and Dutifulness than those in  Development and Evaluation of the PCERQ  77  non-Asian cultures (McCrae, Yik, Trapnell, Bond & Paulhus, 1998). That those raised in two-parent families reported more effectiveness and higher non-abusive standards suggests ways that early family experiences could relate to one's view of conflict situations later in life. Otherwise, the PCERQ was essentially unrelated to distal influences on behaviour, especially age and socioeconomic status, which bodes well for it being considered principally as a measure of psychological phenomena. Some PCERQ sub-scales appear meaningfully associated with demographic features that interact with management of conflict (i.e., marital status) and interpersonal functioning or expectations (i.e., cultural and family background).  The PCERQ and the Family of Origin (Global Impressions and Abuse). The PCERQ sub-scales were considered in relation to a question that asked how pleasant participants' childhood years were (one = very wwpleasant, nine = very pleasant). Responses to this question correlated only with the CI scale [r (221) = -.26, p < .001], showing low efficacy for managing intimate conflict as an adult was associated with an unpleasant childhood. The new scales were also considered in relation to family of origin (FOO) abuse as measured by the CTS-FOOF. The CI sub-scale was the only PCERQ component associated with FOO abuse, as it correlated with being a victim of abuse from mother [verbal: r = .15, p < .05; physical: r = .22, p < .01] and with inflicting physical abuse toward mother [r = .21, p < .01]. The complete results are presented in Appendix B-2. Collectively, these analyses showed that the PCERQ sub-scales were mostly unrelated to early abuse, although the CI sub-scale was mildly but uniquely related abuse in the motherson relationship. The relatively weak associations reported in this area suggest some conflict pattern transference between one's close childhood and adulthood relationships with women. It had been hypothesized that the PCERQ sub-scales would be mostly unrelated to childhood experiences (and more relevant particularly to adult relationships), so the relative absence of strong associations between the PCERQ and early abuse experiences was supported.  Development and Evaluation of the PCERQ 78  The P C E R Q and Social Desirability. Correlations were conducted to determine if the PCERQ sub-scales were vulnerable to response bias as assessed on the Balanced Inventory of Desirable Responding (BIDR). The BIDR assesses two types of response bias: Impression Management (deliberate adjustment of responses to down-play one's faults) and Self-Deceptive Enhancement (a characterological tendency to minimize one's problems so as to defend against acknowledging them). These analyses indicated that, while Impression Management was not associated with the PCERQ sub-scales, Self-Deceptive Enhance was negatively correlated with both CI [r (217) = -. 17, p < .05] and NP [r (217) = -.25, p < .001]. This pattern of correlations was noted consistently within each of the three samples, with the exception that ER was highly correlated with Impression Management in the TGS only [r (49) = .42, p < .001]. See Appendix B-3 for a presentation of all relevant correlation coefficients. Overall, these analyses suggested that the PCERQ scales were modestly and differentially vulnerable to self-presentational biases in questionnaire responding. As a result, when the scales are considered in regression models predicting abusiveness in the next chapter, a more in-depth consideration of response bias will be presented.  The P C E R Q and the SPQ. The CI sub-scale was based on notions of self-control in conflict assessed by the SPQ, a measure pilot-tested by the author in an earlier project. In that study, the SPQ was found to correlate with self and partner reports of abuse. The three SPQ items are embedded within the CI sub-scale developed in this dissertation. An abridged version of the CI was formed by omitting those three items that comprise the SPQ. The correlation between the SPQ items and this abridged CI sub-scale was r = .77 (p < .001). This value of this coefficient will be integrated into a passage in the next chapter that considers the criterion validity of the CI sub-scale.  Development and Evaluation of the PCERQ 79  The PCERQ and the Abusive Personality. The PCERQ items were written to be considered in light of existing measures of perpetrator personality characteristics, particularly the Propensity for Abuse Scale (PAS; Dutton, 1995). Recall the PAS measures shame experiences from the parent-child relationship, stability of one's personal and relationship-oriented self-concepts, and traumarelated somatic symptoms of anxiety. As can be seen in Table 8, the PAS was significantly correlated with the following PCERQ sub-scales (all n's = 218, p's < .001): CI (r = .44), NP (r = .26), and ER (r = .22). It had been hypothesized, given their relatively narrow conceptual and temporal focus on the conflict situation per se, that the PCERQ sub-scales would be somewhat more statistically independent of the PAS than these relationships suggest. A reasonable and conservative interpretation of these findings is that the shame, self-concept and anxiety features of the Abusive Personalityfilterinto the psychological experience of conflict assessed on the PCERQ scales. Clearly, appraisals of self, other and situations during conflict connect with the more general and long-standing personality characteristics assessed on the PAS. The implications of the strong relationships between the PAS and these PCERQ subscales will be more completely investigated in the next chapter's regression analyses which examine their relative contributions to predictions of abusiveness.  The PCERQ and Personality and Attitude Measures. This section considers how the PCERQ scales relate to measures of the Big Five traits and relationship attitudes, so as to facilitate a clearer understanding of the degree to which they measure unique material not especially addressed by other instruments. Given the PCERQ scales were intended to be of relatively narrow application (i.e., describing the psychological functioning of men during intimate conflicts), it was hypothesized that they  Development and Evaluation of the PCERQ 80  would show only mild to moderate correlations with the variables discussed in the following passage. Table 8 includes the results of correlational analyses between the PCERQ scales and the Big Five traits. It can be seen that the CI sub-scale was strongly and equally related to both Agreeableness (inversely) and Neuroticism, with the coefficients absolute values of 1  approximately .40. The NP scale was a relative pure measure of low Agreeableness (r = -.43, p < .001). The SNA scale was modestly associated with pro-social tones reflected on the Extraversion (r = .26, p < .001), Agreeableness (r=.21,p<.01) and Conscientiousness (r = . 15, p < .05) measures. The ER sub-scale was most strongly associated inversely with Openness (r = -.31, p < .001) and Agreeableness (r = -.21, p < .01). These results empirically clarify the conceptual meaning of the PCERQ sub-scales. Respondents with high scores on the CI sub-scale approach conflict with prominent blends of emotional reactivity and uncooperativeness, along with closed-mindedness and sociallyisolating tendencies. Those high on the NP scale are unlikely to experience or express much interest in the desires or perspective of their partner (in the interests of pursuing their own vision of their partner's conduct). Elevations on the SNA scale are consistent with a desire for positive interactions with others and accomodating others' needs; abuse is counter-productive to such aims. Elevations on the ER scale were most associated with a narrow and rigid view of one's choices for alternative courses of action, and a resistance to working with others' perspectives; this closed-mindedness can be interpreted as buttressing pro-aggression cognitions about conflict 'as needed'. It had been hoped that the PCERQ sub-scales, in their attempts to narrowly focus on conflict, would be essentially unique relative to the more general descriptive properties of the Big Five traits (and hence represent a novel collection of interpersonal concepts). The strength of some of the relationships with the Big Five traits makes it somewhat more difficult to view the PCERQ scales as independent of general and ongoing patterns of interpersonal conduct. Comparisons of the relative contributions of the PCERQ and Big Five  Development and Evaluation of the PCERQ 81  Table 8. Correlations of PCERQ Sub-Scales with Personality Measures. PCERQ Sub-Scales CI  PAS  NP  SNA  ER  41***  26**  -.04  .22***  Openness  -.23***  .00  .10  -.31 ***  Conscientiousness  -.11  -.14*  .15*  -.08  Extraversion  -.22***  .00  .26***  -.01  Agreeableness  -.42***  -.43***  .21**  -.21**  Neuroticism  .41***  .12  -.09  .13  NEO-FFI  Note:  CI = Conflict Ineffectiveness. NP = N-Power. SNA = Standards of Non-Abusiveness. ER = ExonerativeRauonalizations. PAS = Propensity for Abuse Scale. N = 227. * p < .05. ** p < .01. *** p < .001.  Development and Evaluation of the PCERQ  82 traits in predicting abusiveness will be evaluated in the next chapter's regression equations. In the meantime, the pattern of these correlational results suggest that the conflict-oriented experiences assessed by the PCERQ are unlikely to be entirely independent of personality traits that have broader and more general association with behaviour across many domains of functioning. The fact that many correlation coefficients were relatively modest in absolute value (despite a high degree of statistical significance) is consistent with the notion that, despite the statistical significance of the relationships, the PCERQ may nonetheless represent something relatively unique that is not assessed within the Big Five traits (again, this possibility will be more fully evaluated in regression analyses reported in the next chapter). The PCERQ was correlated with the Marital Locus of Control Scale (MLOCS) to assess its measurement of conflict efficacy versus relationship efficacy more generally. These analyses showed that each of the PCERQ sub-scales were associated with external locus of control for relationship outcomes more generally. For example, the CI, NP and E R sub-scales were associated with viewing social context as more influential over intimate relationship outcomes than one's own personal contributions. The magnitude of these relationships was stronger than hypothesized at the outset, and suggested that slicing relationship interactions into particularly fine pieces about relationship conflict versus relationship tension more generally was not an especially strong suit of the PCERQ. The UDS participants completed the Rosenberg Self-Esteem Scale (RSEI), a measure of global feelings of self-worth. The CI and ER scales were found to correlate strongly and inversely with the RSEI (r's = -.44 and -.27, respectively; p's < .001). These results provide further support for a conclusion reached earlier that the PCERQ sub-scales relate to more general trait-related influences on behavioural functioning. While the inverse correlations with self-esteem noted were significant, once again the reader is reminded of the notwithstanding plausibility that the PCERQ scales represent something substantial beyond self-esteem.  Development and Evaluation of the PCERQ 83  The P C E R Q and Narrative Measures. The PCERQ sub-scales were evaluated in relation to coded data of subjects' written feelings and reactions to hypothetical conflict scenarios. Analyses of the coded text categories were intended to examine n-power, conflict efficacy and responsibility qualities through a method other than questionnaires. All participants offered narrative responses to hypothetical conflict, and the UDS also generated stories describing events in pictures. Further perspective about the narrative sub-scales' meanings is offered in Technical Note #1, which evaluates these narrative dimensions in light of the Big Five personality traits. Narrative responses to hypothetical conflict. It is important to remember the validitybased hypotheses being tested in this section. The goal was to evaluate the forced-choice PCERQ questionnaire against alternative methods for measuring the same phenomena. Of primary interest in these analyses were: the NP sub-scale in relation to Winter's n-power coding, the SNA and E R sub-scales in relation to Winter's responsibility coding categories, and the CI sub-scale relative to the conflict-oriented coding categories. Bearing these hypotheses in mind, the following results were observed: (a) the NP sub-scale was significantly correlated with n-power coding scores [r (220) = .18, p < 0 1 ] ; (b) the SNA and ER sub-scales were mostly uncorrelated with the five responsibility coding categories (with the exception of SNA being significantly related to 'Concern for Others', r(220) = .21, p<.01); (c) the CI sub-scale was correlated in the anticipated direction with each of the efficacy-oriented coding categories [number of negative attributions of partner: r (220) = 18, p < .01; number of positive attributions of partner: r (220) = -.15, p < .05; negative affective tone regarding conflict: r (220) = .15, p < .05; number of conflict solutions offered spontaneously: r (220) = -.26, p < .001]. These analyses of the PCERQ sub-scales relative to narrative data indicated a modest empirical overlap. The support was strongest for the CI sub-scale and effectiveness-oriented  Development and Evaluation of the PCERQ  84  narrative indicators, with a weaker relationship noted for the NP sub-scale and its coding cateogory, while no relationship was observed between the responsibility coding and the SNA and ER sub-scales. Taken together these results partially support the hypothesis that the PCERQ concepts, as measured in a questionnaire format, would be associated with similar constructs measured through alternative methods.  Summarizing the P C E R Q ' s Concurrent Validity. In general terms, understanding the meaning and value of new measures such as the PCERQ sub-scales is accomplished by delineating how they relate to existing relevant measures. The findings reported in this section have investigated the extent to which the PCERQ sub-scales are unique versus similar compared to a range of other variables relevant to conflict behaviour. Some relationships were noted between the PCERQ sub-scales and demographic variables (culture, marital status and family of origin structure), while only the CI sub-scale was related to violence in the family of origin (only in the mother-son relationship). The focus of this section has been more heavily weighted on how the PCERQ subscales fared relative to various individual differences variables. Self-Deceptive Enhancement response bias correlated negatively with the CI and NP sub-scales; in light of this, socially desirable responding will be examined once again in the criterion validity section that follows. At the outset of this study it was hypothesized that the PCERQ sub-scales would not be associated in a particularly strong manner with such long-standing behavioural influences as the Abusive Personality and the Big Five personality traits. These hypotheses met with variable support: despite the fact that several PCERQ sub-scales correlated with the Abusive Personality and many of the Big Five traits, the extent of covariation leaves room for the possibility that the new sub-scales capture unique variance relative to these trait-oriented variables.  Development and Evaluation of the PCERQ 85  Narrative coding was also examined in relation to the PCERQ sub-scales, with results showing modest relationships between the CI and NP scales' narrative coding categories, but no support for relationships between the ER and SNA their respective responsibility coding categories. In sum, it was found that the PCERQ sub-scales were more related to some personality indicators than had been anticipated, but this is not seen as discounting the potential utility of the new scales. The relative merits of the new scales and the personality trait indicators in the statistical context of abuse will be more fully probed as part of the following section.  Development and Evaluation of the PCERQ 86  Chapter VH  The Criterion Validity of the PCERQ  This chapter's focus is on analyses that examine the PCERQ's capacity to predict verbal and physical abuse beyond the capacity of other instruments. By way of a rationale for the presentation of results that follows, procedures used in controlling the influence of socially desirable response bias in the self-report abuse and predictor measures are described first. Following this, the capacity of the PCERQ sub-scales to uniquely predict abuse relative to other known predictors is examined. Next, the properties of the PCERQ scales on their own and in the context of a theoretical model incorporating the Abusive Personality to evaluate possible mechanisms of action (including interactions) in greater detail are evaluated. Additional considerations of criterion validity conclude the chapter.  A Comprehensive Consideration of Social Desirability in Relation to the PCERQAbuse Analyses. This research has relied on self-report questionnaire responses as the main source of data. Ideally, studies of this type use external criteria to evaluate predictor variables. In this field, partner reports of abusiveness or ratings by therapists or official criminal justice sources are examples of such external criteria. Without an independent criterion against which the PCERQ and other predictors can be evaluated, the next-best option with this data-set was to comprehensively incorporate response bias analyses into the process. These analyses of response bias utilize the Balanced Inventory of Desirable Responding (BIDR; Paulhus, 1984). The BIDR assesses two factors of social desirability: Impression Management (IM; an indicator of the degree to which one consciously attempts to present themself in socially desirable manner) and Self-Deceptive Enhancement (SDE; a scale assessing the tendency to  Development and Evaluation of the PCERQ  87  protect against personal shortcomings by describing oneself in an especially favourable manner). The BIDR will be incorporated into the PCERQ / self-reported abuse analyses in three different ways in order to most fully understand and control for subjects' response style with the various measures. First, the BIDR will be evaluated as a suppressor variable. This involves determining if the PCERQ sub-scales show stronger relationships with abuse after the response bias variance of BIDR is removed from the PCERQ/abuse relationship. Second, the BIDR will be examined as a moderator variable with the PCERQ sub-scales. This moderator possibility involves computing an array of interaction terms by multiplying standardized PCERQ and BIDR scores (i.e., z-score CI * z-score SDE) and testing these against the PCERQ main effects in regression equations predicting abuse to see if the interaction terms make unique contributions. Third, the BIDR's IM and SDE sub-scales will be used as main-effects entered as the first block of hierarchical regression analyses. This approach allows for social desirability variance contaminating both the dependent variable and the predictor variables to be removed before the relationships between dependent and independent variables are tested. Suppressor effect. A variable ("C") exerts a suppressor effect on the relationship between two other variables ("A" and "B") when the strength of the A-B relationship increases once the variance of C is partialled out of that A-B relationship. In such a case, the relationship between A and B is initially underestimated; the stronger "true" relationship between A and B emerges once the dampening influence of C is accounted for. While this AB-C example illustrates the principle of the suppressor effect, the analyses conducted in this dissertation were somewhat more complex. The first suppressor effect involved the PCERQ sub-scale terms ("A"), verbal abuse ("B"fromthe above example) and the BIDR sub-scales ("C"). The PCERQ terms were entered as the first block and accounted for significant variance [F (6, 213) = 24.21 p < .001; Adjusted R-Square = .39]. The second block was comprised of the BJDR sub-scales, which  Development and Evaluation of the PCERQ  88  made a substantial contribution above and beyond the PCERQ sub-scales [F (8, 211) = 20.32, p < .001; R-Square Change = .03, p < .01]. The average change in Beta of the PCERQ sub-scales in phase two of the equation was .01. The key observation from these analyses for the purposes of considering the possibility of a suppressor effect is that the PCERQ sub-scales did not show improved relationships with the criterion once the BIDR variables were entered. The second suppressor effect tested the PCERQ sub-scale and interaction terms in light of physical abusiveness and the BIDR sub-scales. The PCERQ sub-scales initially shared significant variance with physical abuse [F (6, 213) = 11.00, p < .001; Adjusted R-Square = .21]. The BIDR sub-scales were entered as the second block and did not account for signficant variance [F (8, 211) = 8.37, p < .001; R-Square Change = .00, p >.50]. The average change in the PCERQ sub-scales' Beta was .01. As with the verbal abuse analyses discussed in the previous paragraph, there were no indications of a suppressor effect caused by response style on the PCERQ-abuse relationship. Moderator effect. The idea behind a moderator effect is that the product of two variables (A and B) creates a new variable (C) that can relate to a criterion (D) in ways not anticipated by either A or B independently. In this thesis two sets of BIDR-PCERQ moderator variables were formed, as each PCERQ sub-scale was multiplied by each of the EVl and SDE sub-scales. In all, a total of 8 PCERQ/BIDR moderator terms were computed. These moderator terms were tested in relation to each of the two measures of abuse (verbal and physical) through two regression equations. In the first hierarchical regression equation testing moderator terms (with verbal abuse as the dependent variable), the PCERQ sub-scales were entered as main effects in the first block. The BIDR-PCERQ moderator terms were entered as the second block and, block-wise, did not improve on the variance explained by the PCERQ main effect terms (RSquare change = .03, p < .28). The IM*ER interaction term was the only signficant moderator variable in thefinalequation (B = .93, t = 2.14, p < .05). As a result of this  Development and Evaluation of the PCERQ  89  finding, the IM*ER moderating effect was entered in later analyses considering the PCERQ and other predictors of verbal abuse. The second hierarchical regression evaluated response bias moderator terms relative to physical abuse. As with the verbal abuse equation, the second block of moderator terms did not improve on the variance explained by the PCERQ main effects entered as the first block (R-Square change = .04, p < .30). The SDE*SNA moderator effect was the sole significant moderator term in the final equation (B = . 15, t = 2.16, p < .05). Hence, in subsequent analyses that examined the PCERQ and other predictors in relation to physical abuse, the SDE*SNA moderator term was used. Main effect. The third andfinalway that social desirability was statistically considered in relation to the PCERQ and abuse criteria was as a main effect. In this procedure, the BIDR sub-scales were entered as thefirstblock of variables in hierarchical regression models that tested the PCERQ sub-scales and other variables in subsequent steps. By considering the BK)R scales asfirst-blockmain effects, response bias variance contaminating both the criterion (self-reported abusiveness) and predictors (self-reported psychological qualities) was eliminated at thefirststep of multi-step regression models. This technique was used in each of the regression models that follow, and the results of considering the BIDR sub-scales as main effects in this way will be addressed with respect to each of those equations. Summary of social desirability considerations. The gist of this section has been that social desirability influenced both self-report abuse criteria and predictors used in this study. The BDDR questionnaire was used to determine the degree to which self-presentational style affected the data and relationships among variables. Response bias was not found to exert a suppressor effect on the abuse-PCERQ relationships. Two BIDR-PCERQ interaction terms were significant in regression equations predicting abuse (verbal abuse term = IM*ER, physical abuse term = SDE* SNA). The role of the BIDR sub-scales as main effects will be evaluated in ensuing parts of the thesis that test the PCERQ sub-scales in relation abuse and  Development and Evaluation of the PCERQ 90  different arrays of abuse predictors. These various procedures will allow for PCERQ/abuse relationships to be examined in a way asfreeas statistically possible of response bias.  Correlations Between the PCERQ and Abuse Across the Three Samples.  Having seen that the PCERQ scales indeed offered something novel relative to other tested predictors of abuse, additional investigation of the scales' relationships with abuse were undertaken. Correlation coefficients offered one such method for looking at the relationships between the PCERQ and abuse. These values are included in all of the subsequent tables of regression results, as well as in Appendix C-l. In general, the CI subscale performed most consistently in showing strong coefficients with verbal and physical abuse across samples (see Appendix C-l). The NP and ER scales were also associated with abusiveness across samples, but at a lower magnitude than CI.  Regression Analyses Predicting Abuse with Multiple Independent Measures  This section examines the PCERQ scales relative to other independent variables known to be associated with abusiveness. The purpose is to determine if, in the big picture of abuse predictors, the PCERQ sub-scales predict even after one uses various other useful measures. If the PCERQ sub-scales cannot improve on variance explained by other variables there is little point in examining them in much detail. The hypothesis underlying these analyses is that the PCERQ sub-scales and interaction terms bring novel and relevant concepts to the abuse area, and should therefore account for unique variance beyond that explained by prior blocks of variables in predictions of verbal and physical abuse. Before proceeding with more on these regression analyses, it is necessary to clarify the way in which the word "predict" is being used. True prediction occurs when variables collected at one point in time are able to anticipate the occurrence or magnitude of a criterion variable that occurs at some later point in time. In the paradigm of abuse research, true prediction is attempted when potential predictor variables are measured and then the  Development and Evaluation of the PCERQ  91  individuals' behaviours are recorded over time; true prediction occurs when the initiallyassessed variables are statistically related to the subsequent observed behaviour. In this dissertation the independent variables are being examined relative to self-reported behaviour from the individual's past; all data was collected at one point in time. While the long-range goal of this type of research is that variables can be isolated that will relate meaningfully to one's future conduct, reaching that conclusion is not possible based on the methodology used here. Rather, it would be a next step in a research program to determine if the PCERQ subscales predict subsequent behavioural events. A hierarchical regression method was selected because it allows for the evaluation of groups, or 'blocks', of variables to be entered into the equation in a sequence determined by the researcher. This method allows for an evaluation of "R-Square change", a statistic indicating the degree to which each block of variables accounts for unique variance in the dependent measure beyond the level explained by blocks already in the equation. This method also describes the unique contributions of each individual variable in the equation. This paragraph describes the rationale for how the variables were entered into the two regression equations (one predicting verbal abuse, the other predicting physical abuse); the same sequence of variables was used in each equation. As discussed in the previous section, response bias was a variable that could have affected both the dependent and independent measures. The IM and SDE sub-scales of the BIDR were entered as main effects in the first block in the equation. This allowed for social desirability variance in measures entered in later blocks to be accounted for upfront,allowing those subsequent variables to be evaluated relative to the criterion without being contaminated by response bias. After this response bias block, the variable entry sequence was considered in light of: (a) relative temporal stability (i.e., long-standing and enduring versus situational) and, (b) range of functional domains they influence (i.e., general behavioural implications versus specificity to intimate conflict). Hence, appraisal of childhood and physical abuse by parents were entered as Block 2. The Big Five personality traits, PAS and current satisfaction with the relationship comprised Block 3. The  Development and Evaluation of the PCERQ 92  fourth block consisted of the PCERQ sub-scales and the fifth was made up of PCERQ interaction terms (NP*SNA, NP*ER, and CPNP). Before reporting on the regression equations for the whole sample with verbal and physical abuse as dependent variables, mention will be made on the role of socioeconomic status (SES) as an independent variable. The confounding of SES and abuse was an artifact of data collection in this study, as the TGS was markedly lower on SES than the other two samples. Nonetheless, SES was considered relative to abusiveness on a within-samples basis and found nil results in the CS and TGS (all p's > .35), but a trend toward inverse significance in the UDS [verbal abuse: r (145) = -.16, p < .06; physical abuse: r (146) = -.19, p < .05]. It was determined that because SES was confounded with abuse due to the sampling procedure, as well as the fact that the other independent variables were more psychologically-oriented in nature, it would not be included in the regression analyses. Verbal abuse. Results of the equation predicting verbal abuse showed that all blocks except the PCERQ interaction terms (Block 5) explained unique variance beyond previous blocks of variables (see Appendix C-2). The final equation with all variables showed that the PCERQ scales performed strongly in relation to the explanatory power of other psychologically-oriented variables (e.g., personality measures, abuse history). The PCERQ scales proved to be the most substantial block in the equation, bolstered especially by the CI sub-scale. As emphasized earlier in this section, the key issue with this analysis is the general outcome, in this case that the PCERQ sub-scales accounted for unique variance in verbal abuse after other variables had made their contributions in the equation. It is not especially worthwhile to speculate on the 'meaningfiilness' of the constellation of predictors and the PCERQ's place within them, as it is something of an arbitrary and atheoretical array of predictors. Due to that line of reasoning, the regression table is presented in an appendix rather than a table; thefinedetails are not essential to the thesis. More detailed discussions of  Development and Evaluation of the PCERQ 93  results from later analyses will be offered that examine the PCERQ sub-scales' performance in more theoretically meaningful regression models which are offered in tables. Physical abuse. This section considers results of an equation wherein the PCERQ subscales and interaction terms were entered as thefinalblocks relative to the other predictors of physical abuse as described above. The summary table appears in Appendix C-3. The social desirability block did not contribute, but all subsequent blocks of variables made unique contributions relative to the preceding ones. As with the verbal abuse equation, the CI subscale appeared to be the strongest PCERQ contributor. Again, as highlighted above, the main issue being investigated here is the capacity of the PCERQ sub-scales to predict physical abusiveness beyond the contributions of an array of other variables. It is worth emphasizing, however, that the PCERQ scales explained variance in abuse beyond that accounted for by more broad and 'established' personality variables of the Big Five domain. In other words, the PCERQ appears to be offering something new and substantial, based on the results of these analyses. The hypothesis was supported by this analysis, warranting the further testing of the PCERQ in more theoretically-meaningful regression models. The PCERQ Sub-Scales in Isolation; Main Effects and Interaction Terms. This section describes the outcomes of regression procedures wherein the PCERQ scales and interaction terms were entered as dependent variables in regression equations using verbal and physical abuse as the dependent variables. These analyses were conducted to evaluate the capacity of the PCERQ, independent of other predictors, to predict abuse, and will speak to the potential research and clinical utililty of the scales on their own. Preliminary considerations. Recallfromthe earlier section describing the role of social desirability in relation to the PCERQ and abuse that response bias was to be considered light of main and moderator effects. Analyses conducted at that time found that the BIDR's IM scale interacted with the PCERQ's ER sub-scale in regression models predicting verbal abuse, and the BIDR's SDE sub-scale interacted with the PCERQ's SNA scale in an equation  Development and Evaluation of the PCERQ  94  predicting physical abuse. Those two interaction terms will be used in this section, along with the IM and SDE scales as main effects. Given the moderating effect of the particpants' language of origin (degree to which English was spoken in the home while growing up) in relation to verbal abuse, it was entered in the first block along with the BIDR terms of the verbal abuse regression equation. Interaction terms. The regression models constructed in this section evaluated the PCERQ sub-scales as main effects, but also tested three theoretically-important interaction terms. The first of these was the CI*NP term, of interest in relation to earlier studies that had found relatively low efficacy, but variable degrees of power-seeking, among abusers (e.g., Prince & Arias, 1994). The second and third interaction terms evaluated the n-power by responsibility interaction posited by Winter through NP*SNA and NP*ER terms. Recall Winter's finding that n-power and responsibility are relatively independent and interactive (e.g., Winter & Barenbaum, 1984) with respect to how they relate to anti-social behaviour. In the following regression analyses the BIDR main and moderator effects were entered as the first block of a hierarchical equation, followed by the PCERQ main effects in the second block, then three interaction terms as the third block. Verbal abuse. Table 9 presents the summary of the regression of sub-scales on verbal abuse. It can be seen that the response bias and language main effects, and the IM*ER interaction terms all made significant contributions in block one. The second block, consisting of PCERQ main effects, revealed a strong main effect for the CI sub-scale; none of the other PCERQ sub- scales made significant contributions. Thefinalblock of PCERQ interactions did not make an overall significant contribution to the prediction of verbal abuse, but the CI*NP interaction term was marginally significant (Beta = .12, t = 2.07, p < .05). These results demonstrated the capacity of the PCERQ's CI sub-scale to predict verbal abusiveness when controlling for the influence of social desirability and language. Also apparent was the weak association of the other PCERQ sub-scales and interaction terms, including the npower by responsibility interaction researched by Winter, with verbal abuse.  Development and Evaluation of the PCERQ 95  Table 9. Summary of Regression Predicting Self-Reported Verbal Abuse with BIDR and PCERQ. Block Predictor Number Variable 1.  r  B  SEB  Beta  IM  .05  .16  .07  .12*  SDE  -.20**  -.16  .05  -.16**  IM*ER  .21**  .91  .38  .13*  17**  1.52  .52  .15**  CI  .63***  .57  .06  59***  NP  .15*  -.24  .13  -.11  SNA  .00  .11  .10  .06  ER  25***  .18  .15  .07  CI*NP  .07  .83  .40  .12*  -.26  .37  -.04  -.15  .36  -.02  Language 2.  3.  NP*SNA -.07 NP*ER  .00  R-Sq Chg.  22***  .35***  .01  Note: Multiple R = .69, Adjusted R-Square = .45, F (11, 208) = 17.36 (p < .001). IM = BIDR Impression Management, SDE = BIDR Self-Deceptive Enhancement. Language=Language spoken at home growing up.CI = Conflict Ineffectiveness, NP = NPower, SNA = Standards of Non-Abusiveness, ER = Exonerative Rationalizations. * p < .05, * p < .01, *** p < .001.  Development and Evaluation of the PCERQ 96  Physical abuse. Table 10 presents the results of the regression equation predicting physical abuse with the BIDR and PCERQ sub-scales and interaction terms. The response bias indicators exerted less influence relative to the PCERQ scales in this equation than in the previous verbal abuse equation. It was also apparent, consistent with the verbal abuse prediction, that CI was the only PCERQ sub-scale predictive of physical abuse as a main effect. In the physical abuse prediction, however, the NP*ER interaction term added significantly to the equation (Beta = .19, t = 2.84, p<.01). This interaction term is graphically depicted in Figure 1. The NP*ER interaction in predicting physical abuse supports Winter's contention that n-power contributes to aggressive conduct in a problematic manner when combined with an irresponsible approach to relationships (in this case, as assessed on the ER sub-scale). While the PCERQ's NP and ER scales are not particularly valuable independently, their mutual interaction is substantial, both empirically and theoretically.  On the Generation and Control of Aggressive Impulses: Predicting Abusive Behaviour bv Integrating the PCERQ and PAS.  At the outset of this thesis it was outlined how the Abusive Personality presents a theory of intimate abuse that is rooted in notions of impulsivity. The Abusive Personality integrates various psychological elements (sequelae of early shame experiences, ongoing trauma-anxiety symptoms, and instability of personal and relationship identity) into a construct descriptive of those prone to experiencing high levels of emotional distress in the context of their intimate relationship. This turmoil occurs along with cognitions of blame, insecurity and suspicion, which coalesce into aggressive impulses toward one's partner. Recall as well that, while the empirical support for the existence and influence of the Abusive Personality is strong, there is little research describing how such impulses can be managed or restrained during intimate conflicts. The analyses that follow have the aim of describing how impulse-generating personality characteristics may be enhanced or diffused by other psychological components of the conflict experience measured by the PCERQ sub-scales and  Development and Evaluation of the PCERQ 97  Table 10. Summary of Regression Predicting Self-Reported Physical Abuse with BIDR and PCERQ.  Predictor Variable  r  B  SE.B  Beta  1. IM  -.07  -.03  .02  -.09  2. SDE  -.04  .01  .02  .04  3. SDE*SNA  .15*  .24  .11  .14*  4. CI  42***  .11  .02  41 * * *  5. NP  .16*  .03  .04  .06  6. SNA  .05  .05  .03  .10  7. ER  .15*  -.03  .04  -.05  8. CI*NP  .15*  .21  .13  .11  9. NP*SNA  -.06  -.13  .12  -.07  10. NP*ER  .21**  .32  .11  1 g**  R-Sq Ch.  .03  17***  .06***  Note: Multiple R = .51, Adjusted R-Square = .23, F (10, 209) = 7.59 (p < .001). IM = BIDR Impression Management, SDE = BIDR Self-Deceptive Enhancement. CI = Conflict Ineffectiveness, NP = N-Power, SNA = Standards of Non-Abusiveness, ER = Exonerative Rationalizations. * p < .05, * p < .01, *** p < .001.  Development and Evaluation of the PCERQ  98  Figure 1 The Interactive Effect of the Need for Power by Exonerative Rationalizations Sub-Scales in Predicting Self-Reported Physical Abuse.  LoNP/LoER  HINP/Lo ER  LoNP/HiER  H1NP/HIER  Note: NP and ER denote Need for Power and Exonerative Rationalizations sub-scales of the PCERQ, respectively. Multiple-R = .51, Adjusted R-Square = .23. NP*ER interaction Beta = .19 (p < .01). Y-axis is self-reported You-to Partner Physical Abuse scores from the Conflict Tactics Scale (CTS).  Development and Evaluation of the PCERQ 99  interaction terms. To accomplish the goal of statistically integrating the PAS and PCERQ in a theoretically meaningful way, hierarchical regressions were conducted. The model of abusiveness being tested in this section evaluates independent (main effect) and interaction (moderator) terms from the PCERQ sub-scales and the PAS. It is important to address several different issues about how the PCERQ and PAS could combine in relation to abusiveness. First, it is necessary to determine if the impulse-generating Abusive Personality traits measured by the PAS remain significant after the situation-oriented PCERQ components are considered. Second, it is necessary to determine if the NP*ER interaction remains significant when considered as part of this abuse model: the preservation of significance would indicate that it is a unique predictor variable in relation to the impulsegenerating content of the PAS. Third, it is worthwhile to consider interactions between the PAS and PCERQ scales. If such terms make contributions to predictions of abusiveness it would show the importance of integrating impulse-generation and impulse-management in single measurement indicators. From these points, then, the hierarchical regressions predicting verbal and physical abuse used afive-blockapproach: (1) BIDR response bias terms (main effects and moderators), (2) PAS main effect, (3) PCERQ main effects, (4) PCERQ interaction terms, and (5) PAS-PCERQ interaction terms. Regression analyses predicting verbal abuse. The summaryfindingsfromthe prediction of verbal abuse are presented in Table 11. It can be seen that the response bias and moderator terms of thefirstblock remained significantfromtheir initial entry through to the final result. While the PAS was significant at its point of entry, it did not contribute significantly to thefinaloutcome as a main effect once subsequent blocks were entered. The PCERQ block was significant at its point of entry, bolstered by the CI and NP variables that remained significant into thefinalequation. Unexpectedly, the NP sub-scale was inversely predictive of verbal abuse, suggesting that submissive men were the most verbally abusive. The fourth block of PCERQ interaction terms was non-significant in its contributions to the equation. While thefifthblock, comprised of PAS-PCERQ interaction terms as a whole, was  Development and Evaluation of the PCERQ 100  Table 11. Summary of Regression Predicting Self-Reported Verbal Abuse with PAS and PCERQ. r .06  B .22  SE.B .07  Beta  SDE  -.21**  -.16  .05  -.16**  IM*ER  .21**  1.45  .41  .21**  Language  .16*  1.23  .52  .12*  IQ***  22***  Block # / Predictor 1. IM  17**  2.  PAS  27***  .05  .03  .11  3.  CI  54***  .54  .06  55***  NP  .15*  -.34  .13  -.16**  SNA  .00  .16  .10  .08  ER  25**  .20  .15  .08  CI*NP  .07  .79  .45  .11  NP*SNA  -.07  -.34  .41  -.05  NP*ER  .00  -.85  .42  -.13  PAS*CI  .17*  -.16  .41  -.02  PAS*NP  .06  .24  .47  .03  PAS* SNA  -.06  .47  .41  .07  PAS*ER  .07  1.11  .53  .16*  4.  5.  R Sq Ch.  25***  .01  .02  Note: Multiple R = .71, Adjusted R-Square= .48, F (16, 199) = 13.68 (p< .001). R Sq Ch = R Square Change of block at point of entry (prior to entry of subsequent blocks). IM = BIDR Impression Management, SDE = BIDR Self-Deceptive Enhancement. Language = Language spoken at home growing up. CI = Conflict Ineffectiveness, NP = N-Power, SNA = Standards of Non-Abusiveness, ER = Exonerative Rationalizations. * p < .05, * p < .01, *** p < .001.  Development and Evaluation of the PCERQ 101  not signficant, the PAS*ER interaction term remained modestly significant in the final equation. Regression analyses predicting physical abuse. The summaryfindingsof the hierarchical prediction of physical abuse are presented in Table 12. Contrary to the verbal abuse predictive model above, the response bias block was unrelated to physical abusiveness throughout the equation. As with the verbal abuse equation, however, the PAS main effect, which contributed significantly when entered initially, was eroded to a non-significant status following the entry of subsequent blocks. The CI sub-scale of the PCERQ remained significant as a main effect in the final equation even after the two subsequent blocks of variables were entered. Of the PCERQ interaction variables in the fourth block, the NP*ER interaction remained significant into thefinalsolution, as it had in the model without the PAS described in a previous section. One of the most powerful predictive terms in the model was the PAS*CI interaction term, which is graphically represented in Figure 2. From viewing thisfigureit is apparent that feeling ineffective in conflict is an essential component giving rise to abusiveness (both HighCI groups are dramatically more abusive than the Low-CI groups). The Abusive Personality's main connection to physical abuse is in its interaction with a prominent sense of conflict ineffectiveness, as it appears that those with both a level of Abusive Personality traits and strong feelings of ineffectiveness in their ability to manage the conflict experience have the highest levels of self-reported physical abuse. Summary of the PAS-PCERO analyses. The regression analyses of this section have tested a new theoretical model of abusiveness. The procedures used have examined the degree of support for viewing abuse as the product of two independent and interactive psychological domains: ( 1 ) intimacy-specific emotional distress due to long-standing personality characteristics (the Abusive Personality), and ( 2 ) the appraisals, motivations, and experience of intimate conflict per se (as measured by the PCERQ sub-scales). The outcomes  Development and Evaluation of the PCERQ  102  Table 12. Summary of Regression Predicting Self-Reported Physical Abuse with PAS and PCERQ.  IM  r -.06  B -.02  SE.B -.02  Beta -.08  SDE  -.04  .01  .02  .05  SDE* SNA  .15*  .21  .11  .12  2.  PAS  24***  .01  .01  .08  3.  CI  42***  .09  .02  .35***  NP  .15*  .04  .04  .06  SNA  .04  .05  .03  .11  ER  .15*  -.02  .04  -.04  CI*NP  .15*  .02  .14  .01  NP*SNA  -.06  -.15  .13  -.08  NP*ER  .20**  .29  .13  17***  PAS*CI  .33***  .45  .13  25***  PAS*NP  .16*  .07  .15  .04  PAS* SNA  .00  .07  .13  .04  PAS*ER  .18**  -.09  .16  -.05  1.  4.  5.  R Sq Ch.  .02 i j***  20***  .05**  .05**  Note: Multiple R = .57, Adjusted R-Square = .28, F (15, 200) = 6.61 (p < .001). R Sq Ch = R Square Change of block at point of entry (prior to entry of subsequent blocks). EVI = BIDR Impression Management, SDE = BIDR Self-Deceptive Enhancement. CI = Conflict Ineffectiveness, NP = N-Power, SNA = Standards of Non-Abusiveness, ER = Exonerative Rationalizations. *p<.05, *p<.01, ***p<.001.  Development and Evaluation of the PCERQ 103  Figure 2 The Interaction of the Propensity for Abuse Scale by Conflict Ineffectiveness in Predicting Self-Reported Physical Abuse.  LoPAS/LoCI  HiPAS/LoCI  LoPAS/HiCI  HiPAS/HICI  Note: CI denotes Conflict Ineffectiveness Sub-Scale of the PCERQ, PAS is the Propensity for Abuse Scale. Multiple-R = .57, Adjusted R-Square = .28. PAS*CI interaction Beta = .25 (p < .001).  Development and Evaluation of the PCERQ 104  with verbal and physical abuse were different from one another, which leads to the evaluation of each behaviour separately. With respect to verbal abuse, response bias indicators exerted a strong influence on the self-report predictors and criterion. Beyond this in the most general of terms, the PAS main effect, though initially significant, was usurped by the CI sub-scale of the PCERQ. The predictive model was not enhanced by considering PCERQ interaction terms or the PASPCERQ interaction terms. These results suggest that verbal abuse may be best predicted from proximal indicators related to the conflict experience, as opposed to personality traits of greater importance to the long-term experience of negative affect in close relationships. The physical abuse predictions with this model had a more complicated pattern of results than those observed with the verbal abuse analyses. No response bias terms contributed significantly to the final predictive equation, but an array of main effects and interactions were independently associated with physical abuse. As was the case with the verbal abuse analyses, the PAS was initially significant but dropped off in influence once subsequent blocks of variables were entered. The CI sub-scale of the PCERQ retained significance through to the final equation, again similar to the verbal abuse results. The results with the PAS and CI main effects are where the similarities end, however, as two interaction terms made significant contributions to the physical abuse model. The NP*ER interaction term accounted for significant unique variance and supported Winter's notions of high npower only being of particular concern when coupled with an irresponsible outlook toward one's conduct in relationships. The PAS*CI interaction was also significant and showed the explosive consequences of having both a predisposition to feel emotional distress in one's intimate relationship as well as feelings of ineffectiveness for managing conflict situations. Those two psychological phenomena were shown to be somewhat independent (i.e., distressescalating traits are not redundant with conflict ineffectiveness) and their combination is clearly dangerous. Thisfindingis likely one of the main applications uncovered in this dissertation.  Development and Evaluation of the PCERQ 105  Other Criterion Validity Considerations with the CI Sub-Scale.  The various correlational and regression analyses presented to this point have shown strong relationships between the PCERQ scales (and its interactions with other measures) and self-reported abusiveness. The CI sub-scale has emerged as an especially notable new predictive measure from this research. The analyses presented afford one form of criterion validity, but their meaningfulness is questionable in light of the lack of an externally-generated criterion measure of abuse. Fortunately, the capacity of the CI to meaningfully relate to abuse-related criteria is not entirely bound by these self-report results. What follows is a description of additional information that supports the general tone of the findings with the CI sub-scale. A simple way of considering criterion validity was by testing the CI sub-scale's mean between samples. Recall that the TGS was comprised of men who had been court-mandated to attend treatment on the basis of evidence meeting a legal standard (beyond a reasonable doubt) that they had committed assault against their partner. Differences between mean CI scores as a function of sample, with scores by the TGS men more facilitative of abuse, would be anticipated. A univariate ANOVA showed that the TGS was significantly higher than the the other two samples on the CI sub-scale (p < .001; see Table 13). A second additionalfindingrelevant to criterion validity was noted during the concurrent validity analyses of the previous chapter in relation to participants' relationship status. Recall that scores on the CI sub-scale were signicantly higher among estranged men in the TGS than those still with their partner. Similarly, in the test-retest component of the study, a Hotelling's T-Square analysis that contrasted Time One PCERQ sub-scales of those "still with" versus "no longer with" the same girlfriend at Time Two was conducted. The results indicated that participants "no longer with" the same girlfriend at Time Two had significantly higher Time One scores on CI [t (13.78) = 2.63, p < .05]. The fact that the CI scale was relevant in distinguishing between men who encountered terminal problems in their  Development and Evaluation of the PCERQ 106  Table 13. Mean Scores of PCERQ Sub-Scales as a Function of Sample.  Sub-samples Overall  TGS  CS  UDS  PCERQ Subscale  (n = 221)  (n = 50)  (n = 27)  Conflict Ineffectiveness  27.15  33.08  24.33  25.67  N-Power  14.72  14.38  13.14  15.20  SNA  29.94  30.14  30.44  29.72  ER  10.55  11.16  10.04  10.48  (n = 144)  Note: Overall Wilk's Lamda F (8, 430) = 7.92, p < .001. Conflict Ineffectiveness (CI): F (2, 218) = 21.36, p < .001: TGS > both UDS & CS. N-Power (NP): F (2, 218) = 4.42, p < .05: UDS > CS only. SNA & ER showed no significant differences between samples.  Development and Evaluation of the PCERQ 107  relationship within a three-month periodfromthose who did not speaks to its utility in assessing material relevant to the commission of destructive behaviour in relationships. The final piece of support for the CI's criterion validity is in its association with the SPQ, a similar measure developed by the researcher that was tested in a pilot study. In that pilot study, the SPQ consisted of three itemsfromthe CI sub-scale that correlated with partner reports of verbal [r (48) = .25, p < . 10] and physical [r (48) = .26, p < . 10] abuse, as well as se//"-reported verbal [r (48) = .36, p < .01] and physical [r (48) = .15, p = .30] abuse. With the present data, those same three SPQ items correlated strongly with self-reported verbal [r (225) = .62, p < .001] and physical [r (225) = .31, p < .001] abuse. An abbreviated CI scale, formed by omitting the three SPQ items, retained the strong relationship with selfreported verbal [r (225) = .54, p < .001] and physical [r (225) = .28, p < .001] abuse. In other words, it seems unlikely that the three items of the CI that form the SPQ were the only ones associated with abusiveness. This abbreviated CI scale was highly correlated with the SPQ [r (225) = .77, p<.001]. The purpose of this section has been to describe how the PCERQ's most effective sub-scale relates to several relevant criteria, other than self-reported abusiveness, in substantial ways. Men court-referred to wife assault groups score markedly higher on this scale, as do estranged men relative to men who remain in relationships. It was also noted that the CI has a great deal in common with a brief measure, the SPQ, that was found to correlate with partner reports of verbal and physical abuse in a pilot study. These results, it is being advocated, help solidify the foundation of the CI sub-scale in particular, improving the firmness with which one can interpret the abuse-related conclusions with that sub-scale in particular as observed in this chapter.  Development and Evaluation of the PCERQ 108  Chapter VTH  Discussion  This chapter begins by integrating the results reported in the preceding chapters into a summary that highlights the major findings and degree of support for the hypotheses that directed the research. A concise overview of scales and interaction terms, in the form of Figure 3, is intended to anchor this overview of outcomes. Limitations of the study are touched on, and the PCERQ's contributions relative to research and clinical issues in the wife assault literature area are discussed. Suggestions for future research with the PCERQ are made toward the end of the chapter, which concludes with a synopsis of the project.  Evaluation of Hypotheses and Interpretation of Major Findings  Internal Properties At this project's outset there were two clusters of hypotheses concerning the internal properties of the PCERQ sub-scales. The first cluster related to the reliability properties of the measures, and forecast them to be internally consistent and temporally stable. The second cluster addressed plausible interrelationships between its CI, NP, SNA and E R sub-scales. Sub-scale reliability. The CI sub-scale had the most satisfactory Cronbach's alpha coefficient (alpha = .87). The other three scales had more marginally acceptable coefficients of Cronbach's alpha (low- to mid-.60's). Temporal stability was tested by having a subsample of the UDS (n = 33) return three months after their initial testing session to complete a small questionnaire battery. The results of the test-retest analyses indicated that the CI subscale was the most stable over time (robustly correlated and no change in mean scores), while the other three sub-scales were somewhat more variable in their temporal stability  Development and Evaluation of the PCERQ 109  Figure 3 Overview Summary of the PCERQ Sub-Scales' Properties. PCERQ Scales  Conceptual Meaning  Alpha  Concurrent Validity  Criterion Validity (selfreported abusiveness)  Low Agreeableness, high Neuroticism; lack solutions in conflict narratives Low Agreeableness; negative emotional tone and hostile attributions in conflict narratives  Strong main effect, beyond various other predictors, in regression models predicting verbal and physical abuse Correlates with verbal & physical abuse, but does not maintain these relationships relative to other predictors tested in regression models predicting abusiveness Uncorrelated with verbal or physical abusiveness; nil contriubution to regression models Correlates modestly with both verbal and physical abuse; loss of significance relative to other variables within regression models.  CI: Conflict Ineffectiveness  Descriptions of oneself as unable to cope with intimate conflict; feelings of being overwhelmed by conflicts  .87  NP: N-Power  Pursuit of, and concerns with, influence and control over partner and relationship  .65  SNA: Standards of NonAbusiveness  Having values advocating a non-abusive way of managing intimate conflict  .59  ER: Exonerative Rationalizations.  Intentions to be non-abusive in one's personal conduct, no matter how difficult the situation  .66  High Agreeableness & Extraversion, low Neuroticism Inversely correlated with both Openness and Agreeableness  Interaction Terms NP*SNA: N-Power by Standards of Non-Abusiveness  Focuses on how n-power interacts with one's 'responsible' values in relation to abusiveness; tests Winter's npower by responsibility interaction concept  Unrelated, statistically, to abusiveness in correlational or regression analyses  NP*ER: N-Power by Exonerative Rationalizations  Focuses on how n-power interacts with one's strength of resolve for being non-abusive; tests Winter's n-power by responsibility interaction concept.  PAS*CI: Propensity for Abusiveness by Conflict Ineffectiveness  Focuses on how personalitybased aggressive impulses interact with a poor capacity to effectively manage oneself during situational conflict  Strongly associated with physical abusiveness in both correlational and hierarchical regression analyses; high NP and high ER combine to produce markedly high levels of abuse Strongly predictive of physical abusiveness; high PAS & high CI combine to result in markedly high levels of abusiveness  Development and Evaluation of the PCERQ 110  (manifested in somewhat weaker correlations and trends toward changes in the means over time). Interrelationships among the PCERQ sub-scales. An examination of intercorrelations among the sub-scales showed that, contrary to predictions, the dominance-seeking captured in the NP sub-scale was associated with feelings of conflict ineffectiveness on the CI subscale. Past research with assaultive men was mixed in conclusions about the relationship between efficacy and dominance-seeking; this outcome is consistent with the inverse covariation noted by Dutton and Strachan (1987) and not in keeping with the independence reported by Prince and Arias (1994). The NP measure was also correlated with the ER subscale, which assesses for cognitions condoning aggression as a tolerable option. The CI subscale also correlated with ER. The SNA sub-scale was uncorrected with the other new measures; it had been hypothesized that it would correlate with the ER and CI sub-scales. In sum, the PCERQ sub-scales inter-related in a manner largely in keeping with predictions.  Evaluation of Convergent and Divergent Validity. This section provides a summary of how the new scales relate to other indicators of similar or comparable constructs and, thereby, the extent to which they assess unique psychological phenomena untapped by other instruments. Provided in this section are overviews of how the PCERQ sub-scales were found to correspond to other measures of npower, efficacy and responsibility. Also, relationships with demographic, personality disturbance, and prior family violence indicators will also be described. Concurrent Validity of Core Sub-Scale Concepts. The CI sub-scale was evaluated against a variety of other questionnaires that assessed features relevant to managing conflict in one's intimate relationship: (1) the Marital Locus of Control Scale (MLOCS), (2) the Rosenberg Self Esteem Scale, (3) the Conflict Tactics Scales (Family of Origin Form), and (4) coding of narratives describing hypothetical responses to conflict. The CI sub-scale correlated with dimensions of low efficacy for resolving relationship problems as assessed by  Development and Evaluation of the PCERQ 111  the MLOCS and it was also associated with low self esteem. Exposure to childhood abuse was modestly associated with the CI sub-scale, which was also correlated with resourceless responses in narrative responses to hypothetical conflicts. These results offered strong support for the conclusion that conflict efficacy was meaningfully measured by the PCERQ's CI sub-scale. The NP sub-scale was evaluated against Winter's n-power coding scheme and the NEO-FFI's Agreeableness scale. The NP sub-scale correlated significantly but, in absolute terms, relatively modestly with each of those measures. The PCERQ's ER and SNA sub-scales were evaluated against the Conscientiousness scale of the NEO-FFI and Winter's Responsibility coding scheme for narrative text content. These two PCERQ sub-scales were intended to be the prime measures of responsibility in this dissertation, but showed weak associations with these measures. It was unexpectedly observed, however, that Winter's coding scheme was more related to Agreeableness, Openness and Extraversion than Conscientiousness. The SNA sub-scale correlated faintly with Conscientiousness, while the ER sub-scale was uncorrelated with this Big Five trait. It is possible that the cognitively-regulated processes inhibiting violence that were ideally to be measured by the SNA and ER sub-scales, but that proved unassociated with Conscientiousness, share variance with traits other than Conscientiousness. For example, in their discussion of psychopathy and the Big Five model of personal traits, Ffarpur, Hart and Hare (1994) noted that several traits relate to different aspects of irresponsibility and impulsivity: Neuroticism and Extraversion have strong relationships with impulsivity and sensation seeking, while the more cognitive aspects of self-discipline fall under Conscientiousness. Harpur, et al. indicate that the Conscientiousness trait is made up of different components which are not all implicated in antisocial behaviour. It could be that the notion of "restraintfromabusiveness", which the SNA and ER sub-scales were intended to measure, corresponds to trait space in the Big Five interpersonal domain that is not exclusively or primarily linked to Conscientiousness. The internal consistency of the SNA and  Development and Evaluation of the PCERQ 112  ER measures is another source of difficulty in evaluating this connection to Conscientiousness and the Big Five traits generally. In this light, the lack of strong correspondence between the SNA and ER sub-scales does not invalidate, but rather suggests a cautiously informed application of, their use as measures related to the "responsible" abuseinhibition processes underlying this research. Demographic variables. In order to be interpreted primarily as a measure of psychological processes, it was considered desirable for the PCERQ to be relatively independent of such variables as socioeconomic status, age, and cultural background. Generally, the PCERQ sub-scales were unrelated to age or socioeconomic status. However, among the TGS and CS men, estranged participants had higher CI scores than those currently involved in an intimate relationship. Also, in terms of cultural background, UDS participants of primarily Asian background reported higher SNA and ER scores than those of an Englishonly background. These outcomes supported the notion that the PCERQ sub-scales can be interpreted as essentially measuring psychological phenomena, but that in a few key areas (cultural background and intimate relationship status) they may be related to a respondents' background or life situation. Relationship satisfaction. Analyses with the PCERQ scales indicated that desiring power over one's partner (NP) and feeling inadequate about conflict management (CI) were each associated with relationship dissatisfaction. A conclusionfromthese results is that regulating one's desire for influence and feelings of ineffectiveness are part of relationship satisfaction; those least satisfied in their relationships reported such problems particularly related to conflict and self-management during conflict. Social Desirability. Self-Deceptive Enhancement was the BYDR scale most associated with the PCERQ scales. The CI and NP sub-scales correlated inversely with that measure. That participants who scored low CI and NP would have relatively high scores on SelfDeceptive Enhancement is consistent with underreporting vulnerability or dominance in the relationship so as to protect oneselffromthe emotional consequences of dissonance between  Development and Evaluation of the PCERQ 113  one's desired values and the experience of one's actions as inconsistent with them. It is considered encouraging that the PCERQ was unrelated to the Impression Management scale of the BIDR, suggesting that the new scales do not tap self-descriptions that respondents see as unbecoming to the point where desirability biases disclosures. Personality disturbance. A point of interest in this research was the measuring of psychological phenomena important to abusiveness by men without signs of a personality disturbance. It was hypothesized that the PCERQ sub-scales would not correlate especially strongly with indicators of personality dysfunction. The Propensity for Abuse Scale (PAS) and NEO-FFI were used in considering the PCERQ sub-scales' relationships with personality difficulties; it is first helpful to consider how the PAS related to the Big Five traits. The PAS was highly correlated with Neuroticism and Agreeableness (inversely), and somewhat less so (inversely) with Extraversion and Conscientiousness. By this account, the PAS appraises a personality style characterized by the experience of fervent negative moods, abrasive interpersonal characteristics, social withdrawal and lack of diligence. Varous PCERQ sub-scales were correlated with the PAS and most traits of the Big Five. The CI sub-scale was significantly correlated with the PAS, and the traits of Agreeableness (inversely), Neuroticism and Conscientiousness (inversely). Collectively, these results were somewhat at odds with hypotheses that had anticipated weak relationships between situation-oriented conflict management and personality dysfunction. While the PCERQ scales measure some psychological phenomena that co-vary with personality disturbance, they are not rooted (empirically or theoretically) within that realm as the PAS or even the NEO-FFI measures. It would seem that the dilemma of trying to study interpersonal or psychological 'problems' is that they can generally be translated back to a personality conceptualization: the fact that many psychological problems have negative personal and interpersonal consequences is sufficient for this pattern to be fulfilled. From this perspective, it was perhaps inevitable that the conceptual and content scores of the PCERQ would be correlated with  Development and Evaluation of the PCERQ 114  various features of personality dysfunction (especially in the emotion-laden domain of interpersonal violence).  Evaluation of Hypotheses Regarding Criterion Validity.  The ideal of this dissertation was to develop scales that could make unique and strong contributions to the study of abuse by effectively isolating pre-abuse conflict phenomena. The extent to which this was achieved was evaluated through three sets of regression analyses that used self-report verbal and physical abuse as criteria. The first step involved testing the PCERQ sub-scales against a wide array of other variables known to predict abuse, with the goal of determining if the new measures could contribute beyond these predictors. The second test of the PCERQ sub-scales in relation to abuse examined them in isolation with respect to other predictors, so as to examine the PCERQ main effects and interactions without the competing influence of other variables. The third and final step, the most theoretically substantial component of the dissertation, involved assessing the PCERQ subscales along with the PAS (a measure of the Abusive Personality) within a hierarchical regression model. This final step was engineered to evaluate the quantitative support for an interactive model of impulse management in relation to abuse. The PCERQ sub-scales' criterion validity was considered in light of socially desirable response biases by evaluating how Self Deceptive Enhancement (SDE) and Impression Management (IM) affected the PCERQ/abuse relationships. First, examination of suppressor effects in the PCERQ/abuse relationships caused by SDE and IM revealed no such effects. Second, the computation and testing of a pool of PCERQ*BIDR interaction terms in relation to verbal and physical abuse found two moderator terms that were entered into subsequent criterion analyses (verbal abuse: EVI*ER; physical abuse: SDE* SNA). Third, the SDE and IM sub-scales were used as main effects in thefirststep of hierarchical regression equations so as to control for response bias in the criterion and predictor variables as comprehensively as possible.  Development and Evaluation of the PCERQ 115  i. )PCERQ scales and abusiveness in light o f other independent variables. I n basic terms, the PCERQ could only be considered useful or worthwhile i f it added something new to the understanding o f abuse beyond the contributions made by its other known predictors. The regression analyses undertaken to address this issue entered the PCERQ sub-scales and interaction terms after response bias, childhood abuse victimization, traits o f the B i g Five and Abusive Personality theories and relationship satisfaction. The hypothesis that the PCERQ blocks would make unique contributions above and beyond these other variables was supported in equations predicting both verbal and physical abuse. This finding warranted the further inspection o f how these new scales could perform in analyses governed by theoretical considerations. ii. ) PCERQ scales and abusiveness in relative isolation. Given the PCERQ scales made unique contributions relative to some commonly-used measures in the area o f perpetrator characteristics, it was appropriate to look more closely at how the new scales corresponded to abuse on their own (while controlling for the influence o f response bias as outlined in the previous paragraph). While there were subtle differences between the verbal and physical abuse equations in terms o f the exact variables entered in the initial response bias block, the second and third blocks represented PCERQ main effects and interaction terms respectively. The theoretically-meaningful PCERQ interaction terms tested were C I * N P , N P * S N A and NP*ER. The verbal abuse analyses indicated that the C I sub-scale was the most robust predictor, and the C I * N P interaction also contributed modestly (those with high C I scores were especially abusive). N o other PCERQ main effects or interaction terms explained unique variance in the final step o f the equation. The SDE and I M response bias main effects, as well as the I M * E R moderating term, were all significantly and modestly associated with verbal abuse. The lack o f signficance among the n-power*responsibility interaction terms in relation to verbal abuse, as well as the salient role o f response bias variables, will be considered shortly.  Development and Evaluation o f the P C E R Q  116  With the physical abuse predictive model, the CI sub-scale was again the most prominent predictor, while the NP*ER interaction term was moderately significant in the hypothesized direction (high scores on both NP and ER were endorsed by the men who were the most physically violent in the study). It should be emphasized that the NP, SNA and ER scales were not anticipated to be prominent main effects in predicting abuse; rather, the interactive combinations of NP*ER and NP*SNA were hypothesized to be of significance. The low internal consistency of the NP, ER and SNA sub-scales had the effect of suppressing their respective correlational relationships with the physical abuse criterion. For example, when the procedure for correcting attenuation caused by the use of unreliable instruments as outlined by Hopkins and Stanley (1981) is employed, the NP/You-to-Partner physical abuse correlation jumpsfromr (221) = .15, p < .05 to r (221) = .21, p < .001 . Response bias terms 6  did not explain significant variance in relation to physical abusiveness. The pattern of results observed here will be explored in greater detail in the context of the next andfinalset of analyses with the PAS, as the same relationships are preserved there as well. Evaluating a model of impulsivity with the PCERQ and PAS. Thefinalaspect of criterion validity work tested a theoretical account of how impulses may be generated and managed during intimate conflict. In these analyses social desirability variables were first partialled out, followed by entry of the PAS (a measure of the Abusive Personality), PCERQ main effects, PCERQ interaction terms and PAS*PCERQ interaction terms. Again, equations predicting both verbal and physical abuse were constructed. The verbal abuse equation using both the PCERQ scales and the PAS found that the PAS, while it was initially significant prior to entry of the PCERQ scales, it was not  The formula correcting for attenuation caused by using instruments of known unreliability yields a "true" correlation coefficient that would be obtained if the measures were perfectly reliable. To obtain this true coefficient, the observed correlation coefficient is divided by the square root of the product of the two reliability (Cronbach's alpha) coefficients of the measures being used. In this case, .15 was divided by the square root of (alpha NP * alpha You-to-Partner physical abuse), which was the root of (.65*.83). 6  Development and Evaluation of the PCERQ 117  significant in thefinalequation. This appears mostly due to the influence of the CI sub-scale. No PAS/PCERQ interaction terms were significant in the verbal abuse equation. The outcomes from the physical abuse equation were somewhat similar to the verbal abuse results, as the PAS was signficant at its point of entry but contributionsfromthe PCERQ variables and interaction terms eroded that significance in subsequent blocks. While the CI sub-scale was still dominant in the equation, the NP*ER and PAS*CI interaction terms explained significant levels of variance as well. Inspection of graphical representations of the latter term suggested that respondents who reported high levels of conflict ineffectiveness were the most abusive, especially if they showed prominent traits of Abusive Personality. The NP*ER interaction The NP*ER interaction term contributed significantly to predictions of physical abuse, providing empirical support for the notion that resolving to be consistently non-abusive is a key part of achieving that aim (regardless of one's desire to exert influence over one's social environment). The independence of power-seeking and responsible volition, in terms of their influence on abuse as shown in this interaction term, supports the theoretical prediction of Winter and Barenbaum (1984). It argues for a more complex consideration of "power" within the abuse realm, as seeking power does not appear to be an inherently problematic or physical abuse-engendering trait: its combination with minimizing the value of non-abusive intentions is the danger. This resultfitswith research that seeks to identify the role of deliberation as an important factor in understanding the self-regulation of behaviour (e.g., Baumeister, Heatherton & Tice, 1994; Koestner, Bernieri, & Zuckerman, 1992). No response bias terms were found to explain significant quantities of variance in relation to physical abusiveness. It is also worthwhile to note that the SNA sub-scale was largely unrelated to the abuse criteria as a main effect or as part of interaction terms, whereas the ER sub-scale interacted in the expected manner with NP. This outcome suggests simply "wanting to do the non-violent thing" is an insufficient element of non-violence; intending to adhere to that belief no matter what, with no exceptions, seems to be a necessary contributor to non-violence in relation to how one strives for influence in situations of intimate duress.  Development and Evaluation of the PCERQ 118  Other criterion validity considerations with the CI sub-scale. Several other considerations were raised in relation to supporting the CI scale's criterion validity. First, a between-samples difference was noted such that the T G S men reported significantly higher CI scores than those in the U D S or CS. The legal status of the T G S men (e.g., guilty o f assault beyond a reasonable doubt in a court o f law) acted as an external criterion verifying that these were men whose violent behaviour was o f such concern that it required legal and therapeutic intervention. Second, it was noted that Time One scores on C I distinguished those still with same partner at Time 2 from those who were not. O n a related note, estranged men  had higher C I scores than those with a partner at the time of testing in the T G S sample.  This scale offers a window onto the experiences o f the individual who has a more difficult time sustaining the relationship, perhaps in part due to an tendency to resort to abusive tactics during conflict. A final criterion validity consideration involved the SPQ, a short scale comprised o f items from the CI scale that correlated with partner- and self-reports o f male abusiveness in a pilot study. The SPQ was computed for the participants in this thesis, and was highly signficant in its correlations with both self-reported abuse and the C I sub-scale.  Methodological Issues and Limitations of the Current Research Sampling issues. Throughout this dissertation consistency and variation in data and results as a function o f sample have been highlighted. It has been apparent that, despite some similarities, each o f the three samples had some relatively unique characteristics. In the most general view, cultural composition was relatively unique to each sample, the T G S was the most violent and least socioeconomically privileged, and its members also had more criminal justice contact than the two other samples. The U D S was younger and had less relationship experience (duration and degree of involvement) than the C S and T G S . Various analyses were run on a within-samples basis to determine the potential roles o f S E S and arrest in  Development and Evaluation of the PCERQ 119  relation to abusiveness and the PCERQ, and showed nil results. With a few exceptions of minimal theoretical importance, the analyses typically indicated that the background dimensions on which samples differed (e.g., culture) were not particularly imperative to the PCERQ's development and its links to abusiveness that were of central importance to the thesis. It should also be recalled that the respondents in each of the three samples in this study were self-selected. The offer to participate in the study was declined by the majority of men in treatment groups, and the initiative to follow up on a posted advertisement among UDS and CS men is not likely typical of the majority of men in their situations either. While some TGS men did not wish to participate on the basis of poor reading skills or had insufficient time to complete the measures, it was the impression of the researcher that many men simply did not want to provide answers to questions about themselves or their relationship problems in such a 'voluntary' context to someone of a relatively elitist and/or authoritarian position (e.g., someonefroma university collaborating with a facet of the justice system). Without the responses of the men who refused to participate, the data and results likely best map onto that portion of men most interested in considering their psychological and relationship functioning; the men who completed questionnaires may be relatively less guarded in reporting about their situation than men who opted not to participate. It would not be unreasonable to think that some of the men who refused to be involved have higher levels of personality pathology (e.g., Borderline, Avoidant or Antisocial personality disorders) which could lead them toframeparticipation in this study in a rather aversive light. If this is true, the data may actually be a more accurate reflection of the  Development and Evaluation of the PCERQ 120  responsesfromthe population of interest: men in heterosexual relationships whose abusiveness is unlikely associated with features of severe personality dysfunction. The Treatment Group Sample was divided in half based on whether respondents were above or below the mean of the PAS measure of personality disturbance. The fact that each group had strong correlations between CI and self-reported physical abusiveness suggests that the CI construct is a useful one in assessing the aggressiveness of most men . 7  A final word on sampling is reserved for the question of gender and abusiveness. The PCERQ item pool was developed by considering the wife assault literature's findings of individual and relationship factors that contribute to aggression by men in intimate heterosexual relationships. While some of the content on the PCERQ may relate to abuse in other types of close relationships (e.g., by women in heterosexual or lesbian relationships, or by men in gay relationships), it should not be assumed at this point that it would perform as well in other abuse contexts. Measurement and measurement selection. This thesis relied extensively on self-reports of abusive conflict tactics that participants had witnessed, been subjected to and used in various close relationships. While this is not an uncommon practice in research in this area, it misses the perspective on a man's use of such tactics as experienced by his partner (or other family members, in the case of family of origin abuse reports) or even therapists or probation officers. Fortunately, the items used in the PCERQ-90 included items that had been pilot tested in a couples study and were strongly correlated with partner- and self- reported abuse. Examination of responses to those items (i.e., the SPQ scale) among the samples in the The PAS mean in the Treatment Group Sample was 70.4. The High-PAS group (n = 21, PAS mean = 90.4) had a CI/You-to-Partner r = .37 (p < . 10). The Low-PAS group (n = 29, 7  Development and Evaluation of the PCERQ 121 current study found results similar to the SPQ / self-reported abuse findings from the pilot study. Those items were also robustly correlated with the conceptual and factor-derived subscales. Hence, suggestions are that the observed relationships with the self-reported abuse criterion (particularly for the CI sub-scale) are likely capturing some part of a pattern that would also be noted with corroborative criteria data. A main concern with the lack of corroborative reports of abuse in this study involves the potential minimization of abusiveness, perhaps due to response biases. A recent meta-analytic study found only mild to moderate influence of social desirability in reports of abuse, although this influence is typically observed more in perpetrator than victim reports of abuse (Sugarman & Hotaling, 1997). The most optimal strategies to control this influence in the present study were the B I D R sub-scales, along with assurances for T G S men that their responses would remain anonymous and not be shared with their treatment or probation staff. N o suppressor or moderator effects o f response style in relation to the P C E R Q scales and physical abuse were found in the larger hierarchical regressions of this study. On the other hand, when the B I D R scales were considered as main effects entered first in hierarchical regressions predicting verbal abuse, Self-Deception maintained a significant contributions through until the final equation. While it would have been optimal to use measures of abuse more free from response style contamination, social desirability was considered as comprehensively as the methodology would allow. It is also relevant to consider the implications of using only one questionnaire indicator of abuse. The C T S is a widely-used and accepted measure of abuse, but it is not without its detractors or criticism. For example, Marshall (1992) highlighted the absence of  P A S mean = 56.0) had a CI/You-to-Partner r = .43 (p < .05).  Development and Evaluation of the PCERQ 122  an 'intent' component, the unknown correspondence to subjective experiences of victimization and the lack of finer-grained distinctions (e.g., making threats of mild versus serious violence) as weaknesses of the CTS. While such concerns have led to a greater interest in developing more focal measures of abusiveness, the CTS nonetheless remains a cornerstone of research in the area. Indeed, it continues to be updated, revised and reevaluated (Straus, Hamby, Boney-McCoy & Sugarman, 1996). The pattern of violence differences observed between samples, namely the higher levels of abusiveness on the CTS relative to the UDS and CS, served as a basic manipulation check of the CTS's capacity to detect differences in abusiveness between groups known to be different. While the results with the CTS may not allow for as rich an appraisal of nuances of abusive behaviour as some of the newer scales, it was clearly adequate for the basic purposes of this research. Attempts were made in this study to statistically control for a variety of influences that could potentially confound the reporting and interpretation of results regarding the psychological phenomena on the PCERQ (i.e., cultural background, personality qualities). Efforts were made to select measures of these features that could fit within a questionnaire package that needed to be brief but comprehensive. Thefirstof two problems with the culture questions on the demographic screening page was the aggregation of too many countries/cultures within some categories (e.g., 'Asian', the predominant non-North American culture experienced by UDS participants). Also, participants were not asked about their subjective 'cultural identification', or the degree to which they felt a part of the dominant North American culture. In short, had more information regarding participants' culture been obtained, the role of culture in the interpretation of the PCERQ could have been clearer. Only coarse distinctions between participants on the basis of culture were possible given its manner  Development and Evaluation of the PCERQ 123  of assessment. As a result, variation on psychological dimensions as a function of culture (e.g., Campbell, Trapnell, Heine, Katz, Lavallee & Lehman, 1996) could not be considered as closely as desired. Item writing. The degree to which some PCERQ material may be redundant with abuse warrants some reflection. The scales were formulated on the supposition that studying perceptoins and motivations during intimate conflict would offer a basis for identifying why some men decide to implement abusive tactics. In other words, the PCERQ material was intended to isolate differences in the experience of conflict between those who become abusive and those who do not. How can one tell when that conflict process turns from discord to demeaning conduct? The new sub-scales, especially the CI sub-scale, can be seen as describing more distress in conflict among men who go on to become abusive. Factor analysis. With factor analysis it is optimal to use very large samples, as well as to test the replicability of a factor structure in a second independent sample before presenting findings with confidence. In terms of the factor analytic procedure itself, one of the weaker parts of the study is its lack of a particularly large derivation sample, and the absence of a separate cross-validation sample to examine the stability of the sub-scales' factor structures. On the other hand, the factor analysis of the itemsfromall four sub-scales at once, which found that almost all items remained on their anticipated sub-scale factor, suggests that the sub-scales could reasonably be expected to remain cohesive and, quite possibly, replicable in other samples.  Development and Evaluation of the PCERQ 124  The Findings in Light of the Existing Wife Assault Literature.  The introductory chapter of this thesis presented a perspective on the wife assault literature that touched on contributionsfromsocial learning theory, personality research, sociocultural influences and relationship interactionfindings.Mapping the study's results onto those various areas may help the reader place this dissertation'sfindingswithin the wife assault literature more broadly. The social learning perspective deals with the acquisition of behaviour through modeling, and also pertains to emotional and behavioural regulation through cognitive processing. The resultsfromthis study have suggested that physical abuse inflicted by both parties in the mother-son relationship was uniquely related to feelings of conflict ineffectiveness in relation to one's adult female partner. It would not be unreasonable to consider that subjective conflict inadequacy in close relationships with women relates uniquely to earlier life experiences in that area for some men, though further research in this area is warranted to evaluate such a possibility more completely. It may also be worthwhile to consider an area of research bridging responsibility and self-efficacy is the topic of self-regulation. Baumeister, Heatherton and Tice (1994) have written on the process of competing internal processes (based on opposing priorities or motivations) that guide behaviour. They argue that a de-stabilization of behaviour-directing cognitive and emotional processes can lead people into lapses of such self-destructive behaviours as substance abuse, sexual impulsivity and aggression. Internal conflicts between competing desires, they advance, detractsfromself-control. The CI and NP*ER variables of the PCERQ offer a window on competing drives salient to the abusive man: desire for power within the relationship, the perception that he doesn't have it or know how to respectfully work toward it, and that it's too hard tofightfair all the time. That last part is especially pertinent to self-regulation as outlined by Baumeister, et ai, as it provides an escape hatch to get awayfromthe aversive experience of the internal struggle. The self-regulation concept  Development and Evaluation of the PCERQ 125  has yet to be systematically applied in the area of domestic violence, but the bridging potentials sketched in this paragraph show that it has some potentially interesting application to intimate violence. The PCERQ scales can also be considered relative to personality traits in the wife assault area. Resultsfromthis study indicate that, while there were some strong relationships between the PCERQ sub-scales and the PAS, the latter was more highly associated with Neuroticism, low Extraversion and family of origin abuse. These differences suggest the PAS is more related to general interpersonal anxiety and distress than the PCERQ scales. In keeping with emerging indications that some portion of abusive men do not exhibit symptoms of personality disorder (i.e., Holtzworth-Monroe & Stuart, 1994), the PCERQ sub-scales appear to describe problematic emotional processing related to intimate conflict and intimacy more generally. This level of distress, however, does not appear so strong as to be deemed a sign of major personality disturbances that affect a wider range of relationships in more pervasive and insidious ways. Some researchers have found that personality traits can be helpful in understanding the distinctions between types of violence, namely that Neuroticism seems linked to reactive aggression, whereas low Agreeableness is associated with instrumental aggression. To the extent that those two traits correspond with motivations for aggression, the findings of this thesis would suggest that both may be common in the domain of wife assault. That is, both the CI and PAS were related to both high Neuroticism and low Agreeableness. Sociocultural theories have suggested that social power imbalances between men and women are integral to male abusiveness. Resultsfromthis thesis did notfindsupport for this notion. It was found that those PCERQ sub-scales of the greatest predictive value were unrelated to various socioeconomic and social power measures. Thisfindingcontinues to support the emphasis on psychological and interpersonal dynamics underlying violence that transcend socioeconomic boundaries. Another interpretation of thisfindingis that various socioeconomic conditions may contribute to similar problematic experiences and expectations  Development and Evaluation of the PCERQ 126  in intimate relationships. Put somewhat differently, all sociocultural contexts may provide bases for facilitating abusiveness through similar psychological experiences of intimacy and conflict. Relationship interaction approaches have yielded many meaningful contributions to the wife assault literature in recent years (e.g., Coan, Gottman, Babcock & Jacobson, 1997). The interactionist research has found that physical abusiveness is closely associated with male resistance to the demands and expectations of their partner (Coan, Gottman, Babcock & Jacobson, 1997). The power aspects of this study, particularly its interaction with responsible cognitions (ER sub-scale), offer an explanation for the intrapsychic sources of that resistance. In other words, the reluctance or inability of men to work with, let alone submit to, their partner's wishes has contributionsfrompower and responsibility that appear to have been meaningfully measured for thefirsttime in this study.  Avenues of Clinical Application.  Assessment. The pre-treatment evaluation of clients and candidates for wife assault treatment groups tends to be done through relatively unstructured clinical interviews, although there is a move toward risk assessment with such guides as the Spousal Assault Risk Assessment Guide (SARA; Kropp, Hart, Webster & Eaves, 1995). The SARA requires raters to consider twenty facets of the wife assaulter which can be categorized as: (1) criminal history, (2) psychosocial adjustment and (3) spousal assault history. While this instrument has gained a positive reputation for its consideration of a relatively broad range of information about the individual, it does not enter the offender's subjective experience of the relationship and his behaviour. Other measures that are used for clinical assessment in this area, such as the PAS or even broader indicators of personality pathology, fail to query the relationshipspecific dimensions of the PCERQ.  Development and Evaluation of the PCERQ 127  While the PCERQ is not yet ready for clinical use, its uniqueness and strength in this study warrant continued investigation of how it could be best refined and used in a clinical setting. Assessing pre-treatment functioning for areas of particular difficulty requiring focused interventions, as well as considering post-test readings as potential indicators of initial treatment gain and risk for recidivism are all potential application areas for the PCERQ. The recommendations for future research that follow need to be investigated for the PCERQ to eventually be used as a clinical tool. Treatment. It would seem that some of the themes uncovered in the results of this study warrant more consideration in light of treatment as well as assessment. Insofar as the PCERQ identifies the phenomenology brought by many abusive men into a treatment setting, it speaks to what may or may not be helpful in setting useful clinical goals. Various researchers and clinicians have written on the components of what makes a 'good' treatment program (e.g., Dutton, 1995b; Pence & Paymar, 1993), but there remain many variations on approaches and emphases for group programming. These differences reflect not only in the content of the didactic material imparted, but underlying philosophies about the process and locus of therapeutic change. The interested reader is directed to End Note #2 for a thorough discussion of particular treatment issues related to the findings of the research. In brief, that note emphasizes the importance of focusing on efficacy enhancement, as opposed to just 'behavioural skill aquisition', during the treatment process. Ways in which conflict efficacy may be enhanced in therapeutic settings are also presented. An overriding goal of most (forensic) treatment is helping the client internalize selfawareness and self-regulation skills. To the extent that the PCERQ sub-scales have causal roles in psychological mechanisms of abusiveness, they may provide helpful ways for abusive men to gain some control over their behaviour. In other words, by helping men gain faith in their capacity to understand and control some the pre-abuse conflict experiences, they may be more likely to develop and implement non-violent alternatives to relationship conflicts.  Development and Evaluation of the PCERQ 128  Suggestions for Future Research  The instrument construction and evaluation purposes of this thesis have shown that the CI sub-scale, along with the NP*ER and PAS*CI interactions, were particularly useful in the prediction of self-reported physical abuse frequencies. The suggestions that follow consider how these scales can be evaluated further. What are the next steps with the PCERQ? This research was undertaken to build a measure that could eventually be useful in a clinical setting with abusive men. The outcomes supporting the utility of the new measure rely on results involving self-reported abuse. A study examining how the PCERQ concepts fare relative to corroborative reports of abuse (e.g., from past or present partners, police reports, etc.) stands out as a necessary next step. While at this point the PCERQ scales may help therapists target areas that could benefitfromconsideration in treatment, the extent to which change in these features may decrease abuse has not been investigated. To this end, a treatment study evaluating the PCERQ at pre-group, post-group and various follow-up intervals would help gauge the utility of the measure. Examining how malleability the PCERQ concepts are in treatment also relates to the issue of whether the new sub-scales assess entrenched characterological features relatively resistant to brief interventions, or if it is measuring circumscribed psychological dimensions that are relatively specific to intimate abuse. Conducting a study of that sort, that could examine the PCERQ sub-scales in relation to the subsequent behavioural indicators, could also speak to the issue of whether the PCERQ constructs represent causal psychological phenomena or not. At this point, with the available research, it is not possible to conclude that the PCERQ variables cause one to become abusive. Indeed, they could be variables that are shaped by one's aggression and, indeed, may facilitate subsequent aggression. The causal role of these variables, then, requires further experimental consideration.  Development and Evaluation of the PCERQ 129  The results achieved in this dissertation hint at questions about the Conflict Ineffectiveness construct that suggest a need for it to be explored in more detail. Clarifying the roles of predispositional contributors, such as physiological arousability, partner qualities, and communication skills, would help build an understanding of how situational and distal influences could intersect within this construct. Examining the degree to which the CI subscale interacts with other measures of personality disturbance, such as interview-based ratingsfromclinical assessment interviews, would help address the arousability question, while indicators of partner psychological functioning and the communicative abilities of each party would round out such a study. Moreover, a longitudinal component that could track individuals over time could determine how the CI sub-scale corresponds to changes in relationship functioning and status (and, therefore, the degree to which the scale represents situational versus persistent qualities). Investigating the plausibility that high CI scores could 8  be picking up situational experiences with some men, and long-standing interaction style in others, thereby suggesting sub-groups of men scoring high on the scale likely with different prognoses, also relates to this situation/trait research stream. The test-retest aspect of this study offered results which attested to the temporal stability of the PCERQ scales, but it also showed that the CI sub-scale was predictive of relationship dissolution independent of relationship satisfaction and abuse. Further research in this area could more accurately assess if and how variables assessed on the PCERQ scales relate to the longitudinal course of relationships. Acquisition of datafromboth parties in the relationship could also lead to a clearer impression of how women experience being in a relationship with men scoring in problematic ranges on PCERQ scales. When in the relationship do the conflict patterns assessed by the PCERQ begin to exert negative effects? The likelihood that the CI sub-scale captures substantial variance attributable to the relationship conflict interaction, independent of both trait and relationship satisfaction elements, is indicated by the partial correlation (controlling for PAS and joy in current relationship) between CI and You-to-Partner physical abuse r (213) = .27, p < .001. Without controlling for those variables the correlation is r (221) = .42, p < .001.  8  Development and Evaluation of the PCERQ 130  Are there points in the course of the relationship when these discordant psychological processes are most optimally addressed? Such a study could also clarify the causal role of the PCERQ concepts with respect to abuse. Observational studies of couples' interactions, along with partner reports or narratives, may reveal where in the conflict interaction these processes appear and rigidly set in. There was also some ambiguity as to the role of power-seeking in this study due to the correlation between the NP and CI sub-scales. The results cannot definitively say if the desire for dominance over one's partner stemsfroma wide-ranging compensation to inefficacy for conflict management, or if it is a personality orientation that is a barrier to the development of mutually beneficial conflict management strategies. Further research is needed to more completely understand this power/efficacy association in relation to abusiveness. It is likely that returning to considerations of what allows responsibility to develop and remain a prominent part of one's interpersonal orientation would be helpful as part of such investigations. Bolstering the item content of the PCERQ sub-scales with low-end reliability coefficients, such as the NP and ER sub-scales, is also an aim worthy of additional effort. This research also incorporated some non-questionnaire methodology to search for useful trends and threads within spontaneous reactions to conflict. This data likely contains much more of interest than what was reported in this dissertation. Further examination of the narrative data, particularly as it relates to interaction processes (resistance to influence and intrapsychic barriers to that), would likely complement the questionnaire and observation methods typically used in wife assault research.  Synopsis: What Has Been Accomplished Here? This research set out to expand on a model of abusiveness that emphasized the importance of long-standing personality features by evaluating how the conflict situation is perceived and managed. Thefirststep in working toward this goal was the construction of a  Development and Evaluation of the PCERQ 131  new set of questionnaire scales to measure the conflict features of interest. Once this measure, called the PCERQ, was developed, the characteristics of its four sub-scales [Conflict Ineffectiveness (CI), Need for Power (NP), Standards of Non-Abusiveness (SNA) and Exonerative Rationalizations (ER)] were examined. Analyses indicated that the new scales were, for the most part, psychometrically stable and satisfactory, and measured what they purported to. It was found that the PCERQ sub-scales were uniquely associated with physical abuse when examined in a regression model that included measures of distal (Abusive Personality) and proximal (PCERQ) influences on abuse. Particularly, feelings of ineffectiveness during conflict, as well as the interaction need for power over others by excuse-making, contributed to statistical explanations of abuse beyond variance explained by other known predictors. Moreover, the interaction of Abusive Personality traits by feelings of conflict ineffectiveness showed that high levels of ineffectiveness were necessary for the highest levels of abuse to occur. This effect all the more potent among individuals with prominent instability traits of the Abusive Personality.  Development and Evaluation of the PCERQ 132  End Notes  1.  This note considers how the narrative coding measures related to questionnaire  measures of Big Five personality traits. This may help with an understanding of the PCERQ/narrative results which are presented in the text of Chapter 6. None of Winter's responsibility subscales were correlated with Conscientiousness (all n's = 220); the strongest relationships in this category were with Obligations (r = -.ll,p = .ll) and Self-Judgment (r = -.10, p = .14). Stronger relationships were noted on Agreeableness (with Concern for Others: r = .21, p < .01; and Self-Judgment: r = .16, p < .05), Openness (with Concern for Others r = . 17, p < .05) and Extraversion (with Concern for Others: r = . 18, p < .01). From these results, it is apparent that Winter's responsibility measure captured a broad band of Big Five traits conducive to pro-social interactions, with an insignificant presence of Conscientiousness. Winter's N-Power correlated most strongly with Agreeableness (r = -.15, p < .05). In terms of conflict efficacy coding, "number of negative attributions" correlated with Agreeableness (r = -.26, p < .001), while "number of positive attributions of partner" also correlated with Agreeableness (r = .17, p<.01). Number of solutions generated correlated with Neuroticism (r = -.26, p < .001), Extraversion (r = .23, p < .001) and Openness (r =.19, P<01).  2.  This note offers thoughts on treatment implications and applications generated from  the results of the dissertation. Didactic components of treatment programs: Communication skills and perspectivetaking. Most treatment groups typically focus, at least in part, on the acquisition and/or improvement of communication skills. This has typically beenframedas helping men redress the communication deficiencies abusive men often show as part of their relationship difficulties (e.g., Dutton & Strachan, 1987). Working with men on their communication skills  Development and Evaluation of the PCERQ 133  has been challenged recently by researchersfrominteractionist orientations who have advocated that such endeavours do not address abusers' problem of resistance to partner influence. The results of this study, in the context of other research suggesting perspective taking is a buffer against negative relationship outcomes (Leith & Baumeister, 1998), may re-vitalize and re-orient the focus on communication skills issues in groups. Particularly, developing a sense of competency for managing oneself during conflict, perhaps enhanced through the development and practice of such skills, may serve an important role in reconstructing one's conflict effectiveness. The capacity to take the perspective of another has been demonstrated to be a key factor in refraining from aggression toward that individual (e.g., Richardson, Hammock, Smith, Gardner & Signo, 1994). It has also been shown that acute emotional distress of the sort experienced during a shame reaction can be debilitating to evaluative cognitive processing of the immediate situation (Leith & Baumeister, 1998). The research in this dissertation suggests that blocked dominance motives and perceptions of conflict inadequacy may also contribute to a state of internal conflict detractingfromthe perspective taking. Communication skills can be a way for men to practice perspective taking in the group, and to experience how other-focus is impaired and skewed during inner distress, undermining one's capacity for empathy. A secondary value for empathy and communication practice in groups could be re-working the bases for the defensive and antagonistic appraisals of one's partner (a siege mentality) that characterize the CI and NP scale elevations. Feeling attacked, humiliated, and resentful (and viewing one's partner as the source of it) do little to foster openness to influence or empathic attunedness with one's partner. Therapists could also be aware of the potential for such practice and subjective improvement to make a difference in the important CI domain uncovered in this thesis. Emphasizing interpersonal process in groups. So, if it seems necessary to back up prior to 'empathy' and prior to 'openness to influence' in trying to better understand the roots of abusiveness, where does one go? What are the pre-requisites to empathy and openness to  Development and Evaluation of the PCERQ 134  influence in relationship that appear particularly important to non-abusiveness? Those questions followfromsome of the findings in this research. They also dovetail with the other ideal of therapy outcome with assaultive men: gaining insight and self-awareness about internal psychological processing within the relationship. The therapist should not expect this to be an easy route either. To the degree that there is merit in Kohut's (1977) psychodynamic concept that the relationship with the "other" is a mirror of the relationship with oneself, the abuser may unveil a relationship with himself that is marked by some of the same characteristics tapped in the CI sub-scale: inefficacy, uncontrollability, unmet expectations and unpredictability. It may be that gaining some self-knowledge and efficacy for coping with these potentially troubling personal qualities is a foundation for diminishing the intensity of threat and antagonism that colour the partner perceptions of many assaultive men. From such a stand-point the assaultive man would appear in a better position to feel more psychologically stable, intact and predictable to themselves, which would seem to be helpful in adopting a more functional stance in relationship: namely, being empathic and open to partner input and suggestions. Working with these sorts of concepts within a group setting clearly calls for a certain skill base on the part of the therapist. It becomes important for the clinician to be able to identify, elicit, contain and work through these psychological processes as they come up within a group between group members and in relation to the therapist. If one holds any value for the notion that the work of therapy takes place through relationship transference, clinicians need to be aware of how to recognize and work with aspects of transference that may arise relatively uniquely in relationships among, and with, assaultive men. Such feelings as inefficacy within relationships (be they with group members or the therapist) could be examined relative to inefficacy experiences with one's spouse. By the same token, feelings of dominance or disgruntled submissiveness noted by the therapist could also be used as therapeutic raw material.  Development and Evaluation of the PCERQ 135  The degree to which these various concepts have merit bears on such issues as the timing of confrontation by the therapist, and the focus on psychoeducational versus psychological or interpersonal material within the groups themselves. In other words, strategies that adopt an approach designed to 're-educate' men through confrontation without heeding the deeper psychological and experiential processes of the assaultive male may altogether miss intervening substantially on the levels required to facilitate 'empathy', 'communication skills' or 'just say yes'. Murphy and Baxter (1997) have also emphasized the risks of failing to consider the psychological state of the assaultive man in treatment, and the risks of naive confrontation. Given the PCERQ's co-variation with Self-Deception and low Openness to experience, the therapist is advised that making gains in this area is unlikely to come from adopting a psychoeducational approach that does not identify, work with or elicit some of the more interpersonal and dynamic underpinnings of inefficacy, dominance and oppositionality that characterize abusive men. A clinical perspective to working on these issues, that brings immediate emotional reactions and problematic self-awareness to the experiential forefront for re-working in the client/therapist/group interaction, would seem to have more potential in leading to more substantial and lasting growth and change.  Development and Evaluation of the PCERQ 136  References American Psychiatric Association (1994). Diagnostic and statistical manual of the mental disorders (4th edition). Washington, D.C: American Psychiatric Association. Andrews, D.A. & Bonta, J. (1994). The psychology of criminal conduct. Cincinnati, OH: Anderson Publishing Co. Babcock, J.C, Waltz, J., Jacobson, N.S., & Gottman, J.M. (1993). Power and violence: The relation between communication patterns, power discrepancies, and domestic violence. Journal of Consulting and Clinical Psychology, 61, 40-50.  Balshaw, H.B. (1993). Living intentionally: A study of nonviolent men who witnessed spousal abuse in childhood. Unpublished Ph.D. Dissertation (Educational Psychology), University of Calgary. Bandura, A. (1977). Self-efficacy. Towards a unified theory of behaviour change. Psychological Review, 84, 191-215.  Bandura, A. (1979). The social learning perspective: Mechanisms of aggression. In H. Toch (Ed.), Psychology of crime and the criminal justice system (pp. 298-336). New York:  Holt, Rinehart, and Winton. Bandura, A. (1992). Exercise of personal agency through the self-efficacy mechanism. In R. Schwarzer (Ed.), Self-efficacy: Thought control of action, (pp. 3-38). Washington, DC: Hemisphere Publishing Corporation. Bartlett, M.S. (1954). A note on the multiplying factors for various chi square approximations. Journal of the Royal Statistical Society, 16 (Series B), 296-298.  Barnett, O. & Hamberger, L. (1992). The assessment of maritally violent men on the California Psychological Inventory. Violence and Victims, 7, 15-28. Baumeister, R.F, Heatherton, T.F. & Tice, D.M. (1994). Losing control: How and why people fail at self-regulation. San Diego: Academic Press, Inc.  Development and Evaluation of the PCERQ 137  Baumeister, R.F., Smart, L., & Boden, J.M. (1996). Relation of threatened egotism to violence and aggression: The dark side of high self-esteem. Psychological Review, 103, 5-33. Blau, P.M. (1964). Exchange and power in social life. New York: Wiley. Bograd, M., & Yllo, K. (1988). Feminist perspectives on wife abuse. New York:  Sage Publications. Bugental, D.B., Blue, J., Cortez, V., Fleck, K., Kopeikin, H., Lewis, J.C, & Lyon, J. (1993). Social cognitions as organizers of autonomic and affective responses to social challenge. Journal of Personality and Social Psychology, 64, 94-103. Burman, B., Margolin, G., & John, R.S. (1993). America's angriest home videos: Behavioral contingencies observed in home reenactments of marital conflict. Journal of Consulting and Clinical Psychology, 61, 28-39.  Bushman, B.J. & Baumeister, R.F. (1998). Threatened egotism, narcissism, selfesteem, and direct and displace aggression: Does self-love or self-hate lead to violence? Journal ofPersonality and Social Psychology, 75, 219-229.  Caesar, P.L. (1986). Men who batter: A heterogeneous group. Paper presented at American Psychological Association, Washington, DC. Campbell, J.D., Trapnell, P.D., Heine, SI, Katz, I.M., Lavallee, L.F. & Lehman, D.R. (1996). Self-Concept Clarity: Measurement, personality correlates, and cultural boundaries. Journal of Personality and Social Psychology, 70, 141-156. Caprara, G.V., Barbaranelli, C , Pastorelli, C , & Perugini, M. (1994). Individual differences in the study of human aggression. Aggressive Behavior, 20, 291-303. Cather, P. (1995). Violence and personal responsibility. (Letters to the Editor). American Journal of Public Health, 85, 413-414.  Cattell, R.B. (1966). The scree test for the number of factors. Multivariate Behaviour Research, 1, 245-276.  Development and Evaluation of the PCERQ 138  Coan, J., Gottman, J.M., Babcock, J. & Jacobson, N. (1997). Battering and the male rejection of influence from women. Aggressive Behavior, 23, 375-388. Cordova, J.V., Jacobson, N.S., Gottman, J.M., Rushe, R. & Cox, G. (1993). Negative reciprocity and communication in couples with a violent husband. Journal of Abnormal Psychology, 102, 559-564.  Costa Jr., P.T. & Widiger, T.A. (1994). Introduction: Personality disorders and the five-factor model of personality. In P.T. Costa, Jr. & T.A. Widiger (Eds.) Personality disorders and the five-factor model ofpersonality. Washington, D.C.: American  Psychological Association. Cox, K. (December 13, 1995). Nova Scotia minister orders probe into judge's court comments: Man released on probation charged with murder two weeks later. The Globe and Mail: Toronto, Canada Dobash, R.E., & Dobash, R.P. (1979). Violence against wives: A case against the  patriarchy. New York: Free Press. Dutton, D.G. (1994a). Behavioral and affective correlates of borderline personality organization in wife assaulters. International Journal of Law and Psychiatry, 17, 265-277. Dutton, D.G. (1994b). The origin and structure of the abusive personality. Journal of Personality Disorders, 8, 181-191.  Dutton, D.G. (1994c). Patriarchy and wife assault: The ecological fallacy. Violence and Victims, 9, 167-182. Dutton, D.G. (1995a). The Propensity for Abuse Scale. Journal of Family Violence, 10, 203-221. Dutton, D.G. (1995b). The domestic assault of women: Psychological and criminal  justice perspectives. Vancouver, BC: University of British Columbia Press.  Development and Evaluation of the PCERQ 139  Dutton, D.G. & Browning, J.J. (1988). Concern for power, fear of intimacy, and aversive stimuli for wife assault. In G.T. Hotaling, D. Finkelhor, J.T. Kirkpatrick, & M.A. Straus (eds), Family abuse and its consequences: New directions in research, (pp. 163175). Newbury Park, CA: Sage Publications. Dutton, D.G. & Hemphill, K.J. (1992). Patterns of socially desirable responding among perpetrators and victims of wife assault. Violence and Victims, 7, 29-39. Dutton, D.G., & Starzomski, A.J. (1993). Borderline Personality Organization in perpetrators of psychological and physical abuse. Violence and Victims, 8, 327-338. Dutton, D.G., & Starzomski, A.J. (1994). Psychological differences between courtreferred and self-referred wife assaulters. Criminal Justice and Behaviour: An International Journal, 21, 203-222. Dutton, D.G, & Starzomski, A.J. (1997). Personality predictors of the Minnesota Power and Control Wheel. Journal ofInterpersonal Violence, 12, 70-82. Dutton, D.G, Saunders, K., Starzomski, A.J., & Bartholomew, K. (1994). Intimacyanger and insecure attachment as precursors of abuse in intimate relationships. Journal of Applied Social Psychology, 24, 1367-86. Dutton, D.G. & Strachan, CE. (1987). Motivational needs for power and spousespecific assertiveness in assaultive and non-assaultive men. Violence and Victims, 2, 145-155. Dutton, D.G., Van Ginkle, C , & Starzomski, A.J. (1995). The role of shame and guilt in the intergenerational transmission of abusiveness. Violence and Victims, 10, 121-131. Elbow, M. (1977). Theoretical considerations of violent marriages. Social Casework, 58, 515-26. Flournoy, P.S., & Wilson, G.L. (1991). Assessment of MMPI profiles of male batterers. Violence and Victims, 6, 309-320. Gelles, R.J. & Straus, M.A. (1988). Intimate violence. New York: Simon and Schuster.  Development and Evaluation of the PCERQ 140  Gondolf, E. W. (1985). Anger and oppression in men who batter: Empiricist and feminist perspectives and their implications for research. Victimology, JO, 311-324. Gortner, E.T., Gollan, J.K., & Jacobson, N.S. (1997). Psychological aspects of perpetrators of domestic violence and their relationships with the victims. The Psychiatric Clinics of North America, 20, 337-352. Gottman, J.M. (1982). Temporal form: Toward a new language for describing relationships. Journal ofMarriage and the Family, 44, 943-962. Gottman, J., Jacobson, N.S., Rushe, R.H., Shortt, J.W., Babcock, J.C., LaTaillade, J.J. & Waltz, J. (1995). The relationship between heart rate reactivity, emotionally aggressive behavior, and general violence in batterers. Journal of Family Psychology, 9, 227-248. Gough, H.G. & Bradley, P. (1992). Delinquent and criminal behaviour as assessed by the revised California Psychological Inventory. Journal of Clinical Psychology, 48, 298-308. Green, L. W. (1970). Manual for scoring socioeconomic status for research on health behavior. Public Health Reports, 85, 815-827. Gunderson, J.G. (1984). Borderline personality organization. Washington, D.C: American Psychiatric Press. Hale, G., Zimostrad, S., Duckworth, J., & Nicholas, D. (1988). Abusive partners: MMPI profiles of male batterers. Journal ofMental Health Counseling, JO, 214-224. Hamberger, L.K., & Hastings, J.E. (1986). Personality characteristics of men who abuse their partners: A cross-validation study. Journal of Family Violence, J, 323-341. Hamberger, L.K., & Hastings, J.E. (1988). Characteristics of spouse abusers: A controlled comparison. Violence and Victims, 3, 31-48.  Hart, S.D., Dutton, D.G., & Newlove, T. (1993). The prevalence of personality disorder among wife assaulters. Journal of Personality Disorders, 7, 329-341.  Development and Evaluation of the PCERQ 141  Harpur, T.J., Hart, S.D., & Hare, R.D. (1994). Personality of the psychopath. In P.T. Costa, Jr. & T.A. Widiger (Eds.) Personality disorders and thefive-factormodel of personality. Washington, D.C: American Psychological Association. Haslam, N. & Beck, A. T. (1994). Subtyping major depression: Ataxometric analysis. Journal of Abnormal Psychology, 103, 686-692. Hilton, N.Z., Harris, G.T., & Rice, ME. (1998). On the validity of self-reported rates of interpersonal violence. Journal of Interpersonal Violence, 13, 58-72. Holtzworth-Monroe, A. & Smutzler, N. (1996). Comparing the emotional reactions and behavioral intentions of violent and nonviolent husbands to aggressive, distressed, and other wife behaviors. Violence and Victims, 11, 319-339. Holtzworth-Monroe, A., & Stuart, G.L. (1994). Typologies of male batterers: Three subtypes and the differences among them. Psychological Bulletin, 116, 476-497. Holtzworth-Monroe, A., & Anglin, K. (1991). The competency of responses given by maritally violent versus nonviolent men to problematic marital situations. Violence and Victims, 6,  257-269.  Hopkins, K.D. & Stanley, J.C. (1981). Educational and psychological measurement and evaluation (6 Edition). Englewood th  Cliffs, NJ: Prentice-Hall, Inc.  Hornung,C.A., McCullough, BC, & Sugimoto, T. (1981). Status relationships in marriage: Risk factors in spouse abuse. Journal ofMarriage and the Family, 43, 675-692. Hotaling, G.T., & Sugarman, D.B. (1986). An analysis of risk markers in husband to wife violence: The current state of knowledge. Violence and Victims, 1, 101-124. Howard, J.A., Blumstein, P., & Schwartz, P. (1986). Sex, power and influence tactics in intimate relationships. Journal of Personality and Social Psychology, 51, 102-109.  Development and Evaluation of the PCERQ 142  Jacobson, N.S., Gottman, J.M., Waltz, J., Rushe, R., Babcock, J., and HoltzworthMonroe, A. (1994). Affect, verbal conceptual, and psychophysiology in the arguments of couples with a violent husband. Journal of Consulting and Clinical Psychology, 62, 982988. Jacobson, N.S., Gottman, J.M., Gortner, E., Berns, S., & Shortt, J.W. (1996). Psychological factors in the longitudinal course of battering: When do the couples split up? When does the abuse decrease? Violence and Victims, 11, 371. Jouriles, E.N & O'Leary, K.D. (1985). Inter-spousal reliability of reports of marital violence. Journal of Consulting and Clinical Psychology, 53, 419-421. Kalmuss, D.S. & Seltzer, J.A. (1986). Continuity of marital behaviour in remarriage: The case of spouse abuse. Journal ofMarriage and the Family, 46, 11-19. Kasian, M., & Painter, S. (1992). Frequency and severity of psychological abuse in a dating population. Journal of Interpersonal Violence, 7, 350-364. Kernberg, O. (1977). The structural diagnosis of borderline personality organization. In P. Hartocollis (Ed.), Borderline personality disorders: The concept, the syndrome, the patient, (pp. 87-121). New York: International Universities Press. Kohut, H. (1977). The restoration of the self. New York: International Universities Press. Koestner, R., Bernieri, F, & Zuckerman, M. (1992). Self-regulation and consistency between attitudes, traits and behaviors. Personality and Social Psychology Bulletin, 18, 5259. Kropp, P.R., Hart, S.D., Webster, CD., & Eaves, D. (1995/ Manualfor the Spousal Assault Risk Assessment Guide  (2nd Edition). Vancouver BC: The British Columbia Institute  on Family Violence. Lamb, S. (1991). Acts without agents: An analysis of linguistic avoidance in journal articles on men who batter women. American Journal of Orthopsychiatry, 61, 250-257.  Development and Evaluation of the PCERQ 143  Langhinrichsen-Rohling, J. & Vivian, D. (1994). The correlates of spouses' incongruent reports of marital aggression. Journal of Family Violence, 9, 265-283. Leith, K.P. & Baumeister, R.F. (1998). Empathy, shame, guilt and narratives of interpersonal conflicts: Guilt-prone people are better at perspective-taking. Journal of Personality, 66, 1-37.  Margolin, G , Burman, B., & John, R.S. (1989). Home observations of married couples reenacting naturalistic conflicts. Behavioral Assessment, 11, 101-118. Margolin, G., Gordis, E.B., Oliver, P.H. & Raine, A. (1995). A physiologically based typology of batterers - Promising but preliminary: Comment on Gottman et al. (1995/ Journal of Family Psychology, 9, 249-252. Marengo, J. (1994). Classifying the courses of schizophrenia. Schizophrenia Bulletin, 20, 519-535. Mason, A. & Blankenship, V. (1987). Power and affiliation motivation, stress, and abuse in intimate relationships. Journal ofPersonality and Social Psychology, 52, 203-210. McClelland, D.C. (1975). Power: The inner experience. New York: Irvington Publishers. McCrae, R.R., Yik, S.M., Trapnell, P.D., Bond, MH. & Paulhus, D.L. (1998). Interpreting personality profiles across cultures: Bilingual, acculturation, and peer rating studies of Chinese undergraduates. Journal of Personality and Social Psychology, 74, 10411055. Miller, P.C., Lefcourt, H.M., Holmes, J.G., Ware, E.E., & Saleh, W.E. (1986). Marital locus of control and marital problem solving. Journal of Personality and Social Psychology, 51, 161-169.  Murphy, CM. & Baxter, V.A. (1997). Motivating batterers to change in the treatment context. Journal of Interpersonal Violence, 12, 607-619.  Development and Evaluation of the PCERQ 144  Novaco, R. (1976). The functions and regulation of the arousal of anger. American Journal of Psychiatry, 133,  1124-1128.  Paulhus, D.L. (1984). Two-component models of socially desirable responding. Journal of Personality & Social Psychology, 46, 598-609. Paulhus, D.L. (1998). Interpersonal and intrapsychic adaptiveness of traits selfenhancement: A mixed blessing? Journal of Personality and Social Psychology, 74, 11971208. Pence, E. & Paymar, M. (1993). Education groups for men who batter: The Duluth model.  New York: Springer.  Petrik, N.D, Olson, R.E.P., Subotnik, L.S. (1994). Powerlessness and the need to control: The male abuser's dilemma. Journal ofInterpersonal Violence, 9, 278-285. Prince, J.E. & Arias, I. (1994). The role of perceived control and the desirability of control among abusive and non-abusive husbands. The American Journal of Family Therapy, 22, 126-134. Renzetti, C. (1994). On dancing with a bear: Reflections on some of the current debates among domestic violence theorists. Violence and Victims, 9, 195-200. Richardson, D.R., Hammock, G.S., Smith, S.M., Gardner, W. & Signo, M. (1994). Empathy as a cognitive inhibitor of interpersonal aggression. Aggressive Behavior, 20, 275289. Riggs, D.S. & O'Leary, K.D. (1996). Aggression between heterosexual dating partners: An examination of a causal model of courtship aggression. Journal of Interpersonal Violence, 11,  519-540.  Rosenberg, M. (1965). Society and the adolescent self-image. Princeton, NJ: Princeton University Press. Saunders, D.G., (1991). Procedures for adjusting self-reports of violence for social desirability bias. Journal ofInterpersonal Violence, 6, 336-344.  Development and Evaluation of the PCERQ 145  Saunders, D.G. (1992). A typology of men who batter women: Three types derived from cluster analysis. American Journal of Orthopsychiatry, 62, 264-275. Saunders, D.G., Lynch, A.B., Grayson, M., Linz, D. (1987). The Inventory of Beliefs about Wife Beating. Construction and initial validation of a measure of beliefs and attitudes. Violence and Victims, 2, 39-55.  Schwarzer, R. (1992). Self-efficacy: Thought control of action. Washington DC: Hemisphere Publishing Corporation. Smith, J.P., & Williams, J.G. (1992). From abusive household to dating violence. Journal ofFamily Violence, 7, 153-165. Snell, J.E., Rosenwald, P.J., & Robey, A. (1964). The wifebeater's wife. Archives of General Psychiatry, 2, 107-113.  Spanier, G. (1976). Measuring dyadic adjustment: New scales for assessing the quality of marriage and similar dyads. Journal ofMarriage and the Family, 3, 15-28. Starzomski, A.J. (1993). Attachment style, affect, and construal of interpersonal conflict.  Unpublished Master's Thesis. Department of Psychology, University of British  Columbia. Stewart, A. & Rubin, Z. (1976). The power motive in the dating couple. Journal of Personality and Social Psychology, 34,  305-309.  Straus, M.A. (1979). Measuring family conflict and violence: The Conflict Tactics Scale. Journal ofMarriage and the Family, 41, 75-88. Straus, M.A. (1990). New scoring methods for violence and new norms for the Conflict Tactics Scales. In M.A. Straus and R.J. Gelles (Eds) Physical violence in American families: Risk factors and adaptations to violence in 8,145 families. New Brunswick,  Transaction Publishers.  NJ:  Development and Evaluation of the PCERQ 146  Straus, M.A., Gelles, R.J., & Steinmetz, S. (1980). Behind closed doors: Violence in the American family.  Garden City, N.J.: Anchor/Doubleday.  Straus, M.A., Hamby, S.L., Boney-McCoy, S. & Sugarman, DB. (1996). The Revised Conflict Tactics Scales (CTS2): Development and preliminary psychometric data Journal of Family Issues, 17, 283-316. Suedfeld, P, Hakstian, A.R., Rank, D.S., & Ballard, E.J. (1985). Ascription of responsibility as a personality variable. Journal ofApplied Social Psychology, 15, 285-311. Sugarman, J.B. & Hotaling, D.B. (1997). Intimate violence and social desirability: A meta-analytic review. Journal ofInterpersonal Violence, 12, 275-290. Tabachnick, B.G. & Fidell, L.S. (1989). Using multivariate statistics (2ndEdition). Harper & Row, Publishers: New York. Tolman, R.M., & Bennett, L.W. (1990). A review of quantitative research on men who batter. Journal of Interpersonal Violence, 5, 87-118.  Tolman, R.M., Edleson, J.L. & Fendrich, M. (1996). The applicability of the Theory of Planned Behavior to abusive men's cessation of violent behavior. Violence and Victims, 11, 341-354. Widom, C. (1989). Does violence beget violence? Acritical examination of the literature. Psychological Bulletin, 106, 13-28. Winter, D. G. (1973). The power motive. New York: The Free Press. Winter, D.G. (1989). Scoring manual for Responsibility. Unpublished manual, University of Michigan. Winter, D.G. & Barenbaum, N. B.(1984). Responsibility and the power motive in women and men. Special Issue: Conceptualizing gender in personality theory and research. Journal of Personality, 53,  335-355.  Development and Evaluation of the PCERQ 147  Yllo, K. & Straus, M. (1990). Patriarchy and violence against wives: The impact of structural and normative factors. In M. Straus & R. Gelles (edsj, Physical violence in American families (pp. 383-99). New Brunswick, NJ: Transaction.  Development and Evaluation of the PCERQ 148  Appendix A-l. Demographic Features by Sample. Sample  Age Type of family of origin  UDS (n= 147) Mean (SD) or n(%)  TGS (n = 48) Mean (SD) or n (%)  CS (n = 27) Mean (SD) or n (%)  19.81 (2.0)  38.25 (8.6)  37.56(10.4)  n (%)  n m  n m  Two-parent  128 (88%)  34 (69%)  23 (85%)  Single-parent  10 (5%)  8 (16%)  3 (11%)  Step-family  3 (2%)  5 (7%)  1 (4%)  Other  5 (3%)  2 (4%)  0 (0%)  8.01 (8.1)  9.24 (7.3)  3.82(4.5)  .19 (.8)  n(%)  n m  Length of relationship (yrs) 1.39 (1.1) Number of arrests Current marital status  .07 (.3) n(%)  Single  143 (97%)  6 ( 12%)  3 (12%)  Married  1 (1%)  10 (20%)  16 (62%)  Separated/divorced Common-law  0 (0%) 2 (2%)  18 (37%) 15 (31%)  1 (4°/ 6 (23%)  Note: UDS denotes Undergraduate Dating Sample, TGS denotes Treatment Group Sample andCS denotes Community Sample, (n) denotes number of participants. (%) denotes percentage of sample.  Development and Evaluation of the PCERQ 149  Appendix A-2. Cultural Background Features by Sample. UDS (n = 147) n (%)  TGS (n = 48) n(%)  CS (n = 27) n (%)  Subject's birthplace Canada/USA  109(74%)  43 (86%)  17(63%)  Asia  32 (22%)  2(4%)  2(6.5%)  Europe  3(2%)  4(8%)  6(22%)  3(2%)  1(2%)  2(6.5%)  Canada/USA  55 (38%)  31(74%)  15(55%)  Asia  62(42%)  3(6%)  3(11%)  Europe  21(14%)  9(18%)  8(30%)  9 (6%)  1 (2%)  1 (4%)  Canada/USA  54(37%)  37(74%)  15(55%)  Asia  63 (43%)  2(4%)  3(11%)  Europe  23(16%)  10(20%)  8(30%)  7 (5%)  1 (2%)  1 (4%)  93 (63%)  42(86%)  19(70%)  Somewhat  25(17%)  2(4%)  2(7%)  No  29(20%)  5(10%)  6(23%)  Other Father's birthplace  Other Mother's birthplace  Other English growing up? Yes  Note: UDS, TGS and CS denote Undergraduate Dating Sample, Treatment Group Sample and Community Sample, respectively, (n) denotes number of participants. (%) denotes percentage of sample.  Development and Evaluation of the PCERQ 150  OO  '—i-H  ON  H H  CN CN  cn  O  o  CN CN CN  ^-^ «-)  •Ti  00  00  o o"  © ©  ©  ©  © cn  © cn  ©  ©  m -*  00  >n  cn  Tt  >/->  CN  VO  Tt  —'  o  o vd  r-  ON  Tt  cn"  o  o  VO  vd  o Tt  y  Tt Tt  Tt  © ©  Tt  > 00 00 Tt  '  ls ids2  p O CO  vd  O  O  CN  •O cn  1  Tt  ON  O o"  o  00  'o ' ©'  cn  vd  —'  © ©  cn  •- '  © ©  m  vd  —-  © CN  I  CD  T3 CO  U  00 0) 60  3 CN"  i °  p cn  © cn  ©  ©  Tt  Tt  Cu)  •3 2  Tt  C/3  0) 60  co  .5  II  co  vd CN cn  vd  vd  vd  *t  Ti-  vd  ir>  CN  en  00  t  r-  </->  Tt  Tt  OS Tt  00 Tt  vi  ON cn vi  fa  H H  cj  cn  CN CN  o  q  O  o"  3  '3  i-H  © ©  vs  cn  © ©  © ©  3  a  cN i ^  s° CO  ^ _  1  -G  g _CJ  o  o  Tt"  Tt"  © CN  © cn  ©  o  Tt  v>  Tt  U  •B 00  3  ON  $  3 °  an  CN  Cu)  p cj  (2.  .  I  3  S xi uM -  «  8 "S CS CN ON  t~-  Tt'  tt, s  </->  00  Tt  o\  ''  VO © Tt  VO  CN  Tt  Tt  Tt  Tt  >o  Os © vi  CN CO vi  ^-^ 00 Tt  I  £3  P  co CJ  -HH  •e  cn •  3  <u (X  o 3  c/3 O  a x  w  -a Q  HH  o PH  O  O0  2 fa  00 P  Q  p  o  to  Q  CJ  o  2  £  Liu PH  II  PH IH  Jj 2 a  H H  P CJ  1>> 5  i l  Development and Evaluation of the PCERQ 151  CD  CD  -r) S II o  VO  00  CN  VO  <n  VO  00  00  o  o  o  o  o  o  ©  o  o  o  o  o  §  "§ £  m  € C  CD  ,9 .2 <a  C/3 C N  Q  U  m  CO  c  cd  CN"  q  cn  cn  O  CN  I-H'  00 o  o  m  CN  O  o  O  o  CN  O  O  VO  r--  i—i  in  CN  « g »cn  00  <D +->  C  1 - 1  CN  o —  O  CD T 3  cn  "° S CO  O  -rl  ^  C  CN C l  -  "  ^  1  O X> <U I d  &-§ E m -2 » rt  «  o cd  m ^—'  o  o  O  o  o  o  ON  VO  ON  —'  VO  00 —  1  o  *—' ©  o  d  d  y  ©  l ^ . s  O 00 ^ CN  CD co  §  CO  >n in  CN  CN  ON  00  in  m'  cn  d  cn  vo CN CN  cn  cd CD  00  H  CN  CN CN  00  cn  CN  VO ©  cn cn d  3  CO CD  CD  CO  O  £ cd  e  '3 5+1  o  co  PH C  <u co  CD  >  C/3 CD  00  5  m  cn VO  VO  CN  cd CO  c  00  o  o  o  O  o  o  d  d  d  d  d  d  c/i 00, t o 'PH  Q  o  vq  c  cd CD  in cn  CN  CN  00  VO ON  CN r-  CN  d  i—i  O t"d  VO  ^H  CN  — i l  vq  3  2 C/3  CN  m cn  CN  cn  d  CD  X  •5  PH  o  c  CD  T3  CM  CD  O  C CD  ki  ucd  CO  <  "a,  CO  S cd  £ T  "ed O '53  O  CD  ^  •tj  Q  CO  00  COH  CD  «>  CO  5  PH  * D H <D  in  "cD  co cd  t-  cd  Ii  ON  O  H e  OO  >. X  vo o •4-*  •8 co X  S3  TD  CD  cd  o  -3  CD  X  o  CO  op « cd  c  o,  PH  2 O  cd  PH  CO  PH  5 o  PH  Note: U  9  O  Sto  ex  CO  Sto  a  u. 3 CO  >  /  »H  CD C N OH  CL>  r.  cn O  Development and Evaluation of the PCERQ 152  IT)  Q  IT)  00  o  CO  o  ON  o  CN  cu _> %3  O  CO  <u  U  CH CO  O CN  00 CN  00  o  (U JU  e  cd  co '3  >>  3 3  > \£  § » o  o CO  vs  co  *>-> cs  CN  CN  U  cu  3. co  .  "O CU 3 ^  r  (D P H  IS  IT)  o  co *g 9-  m  oo O  CN  r-  CO  (U  3  o  Bcd  3  O <U • rs  £ c  o-  cu „ o cu cu -1  8 £  cd cu CO $  SP-o .2 C3 cd 00 CN  Q  Q S CN O  CN  CO  ON  o  CO  cd  ^2 2  oo o  o  O  cd  S  •2  <5 G  H->  CO  CU  o  ^  cd  *  2  s  ^  cl  ^  s S  CO  1-8 CO <L> co 3  <L> 3  -e cd  CH  •e <U  >  3  O  3  <U co  O >s  3  <u  o  3  ts  o  cd  CH  3  3  t: cd  PH  In  O PH  1  3  O >% O VH  CU 3  ts  cd  OH  Q  co"  rt  Development and Evaluation of the PCERQ 153  Appendix A-6 Differences on Violence Measures in the TGS Based on Inclusion and Exclusion of Four Most-Violent Respondents. Whole TGS Sample Mean (SD)  TGS Sub-sample Mean (S)  Verbal Abuse Father to Mother  6.68 (6.6)  6.20 (6.3)  Mother to Father  6.18 (6.0)  6.12 (6.1)  Father to You  5.98 (6.7)  5.33 (6.0)  Mother to You  5.04 (5.7)  5.11 (5.8)  You to Father  5.47(4.5)  5.38 (4.4)  You to Mother  5.57(4.7)  5.46 (4.7)  You to Partner  16.28 (9.0)  15.60 (8.8)  Partner to You  18.55(10.3)  18.18(10.3)  Father to Mother  3.14 (5.6)  2.76 (5.1)  Mother to Father  2.22 (4.2)  2.19 (4.3)  Father to You  4.38 (5.6)  3.93 (5.2)  Mother to You  2.47 (4.0)  2.50 (4.2)  You to Father  0.65(1.5)  0.61(1.4)  You to Mother  0.35 (0.8)  0.33 (0.8)  You to Partner  3.50 (4.3)  2.54 (2.6)  Physical Abuse  Partner to You. 5.47(5.4) 5.20 (5.31 Note: Family of Origin (full sample: n = 44-46; sub-sample n = 40-42). You-Partner (full sample: n = 54; sub-sample n = 50).  Development and Evaluation of the PCERQ 154  Appendix B-l PCERQ Sub-Scales in Relation to Demographic Features: Correlations and Mean Differences. CI  NP  SNA  ER  Age: TGS CS UDS  .09 .19 -.19*  .02 .35 .02  .09 -.16 .07  .00 .26 -.15  SES: TGS CS UDS  .00 -.32 -.13  -.15 .21 .05  .26 -.02 .11  -.12 -.10 -.14  Relationship Duration: TGS CS UDS  -.06 .31 .04  -.05 .22 .04  .00 -.22 .00  .11 .26 -.06  Number of Arrests: TGS CS UDS  .05 .21 .07  .32* -.17 .06  .06 .05 -.14  -.05 .11 -.05  Mean Differences Between Sub-Groups Varying on Demographic Features: UDS Culture Groups: Asian (n = 28) Non-Asian (n = 52)  25.96 24.90  CS/TGS Marital Status: Presently Involved (n = 45) 28.28 a Presently Estranged (n = 33) 32.44 6  14.75 15.71  30.71 a 28.81 6  11.96 c 9.44 d  13.33 14.81  29.60 31.16  10.84 10.66  Note: UDS = Undergraduate Dating Sample, TGS = Treatment Group Sample, CS = Community Sample. CI = Conflict Ineffectiveness, NP = Need for Power, SNA = Standards of Non-Abusiveness, ER = Exonerative Rationalizations. For mean differences section in the bottom of this Appendix, a & b denote means different at p < .05, c & d denote means different at p < .001(no other means differed significantly).  Development and Evaluation of the PCERQ 155  Appendix B-2. Correlations of Family of Origin Abuse and PCERQ Sub-Scales. Dyad  CI  NP  SNA  ER  Father to Mother Mother to Father  .02 .06  -.07 -.04  .06 .12  -.05 -.05  Father to You You to Father  .06 .06  .00 .08  .07 .08  .01 .00  Mother to You You to Mother  .15* .14  .01 .06  .12 .11  .03 .04  Father to Mother Mother to Father  .07 .17*  -.01 .13  .12 .05  .09 .12  Father to You You to Father  .15* .09  .02 .11  .07 .01  .05 .11  Mother to You You to Mother  .22** .21**  .07 .11  .03 -.03  .07 .10  Verbal Abuse:  Physical Abuse:  Note: CI = Conflict Ineffectiveness, NP = N-Power, SNA = Standards of NonAbusiveness, ER = Exonerative Rationalizations. * p < .05, ** p < .01, *** p < .001.  Development and Evaluation of the PCERQ 156  Appendix B-3 Correlations of Social Desirability with PCERQ Sub-Scales. Impression Management  Self-Deceptive Enhancement  Overall fn = 217): CI NP SNA ER  -.07 -.03 -.05 .04  -.17* -.25*** .03 -.05  TGS (n = 49): CI NP SNA ER  -.05 -.03 .00 .42**  -.34* -.35* .25 .00  CS (n = 27): CI NP SNA ER  -.14 -.03 -.01 .21  .07 -.20 -.07 .09  UDS (n= 141): CI NP SNA ER  -.16 .00 -.09 -.15  -.23** -.18* -.04 -.10  Note: UDS, TGS and CS denote undergraduate dating, treatment group and community group samples respectively. CI = Conflict Ineffectiveness, NP = N-Power, SNA = Standards of Non-Abusiveness, ER = Exonerative Rationalizations. * p < .05. ** p < .01. *** p < .001.  Development and Evaluation of the PCERQ 157  Appendix B-4. Correlations of Narrative Coding Conflict Categories with PCERQ Sub-Scales. CI  NP  SNA  ER  N-Power  .10  .18**  jp**  .09  Standards  -.03  .16*  .00  -.08  Obligations  .07  .01  -.09  .01  Others  -.07  .01  2i**  -.06  Conseq  .04  .10  .08  -.06  Slf-Judgmt.  -.03  -.04  .09  -.07  Neg.Tone  .15*  .23**  -.09  .14*  '+' Attrb.  -.15*  -.13  .08  -.15*  '-' Attrb.  .18**  .22**  .05  -.03  Solutions  -.26***  -.02  .14*  -.11  Conflict Narratives  Note: CI = Conflict Ineffectiveness, NP = N-Power, SNA = Standards of NonAbusiveness, ER = Exonerative Rationalizations. *p<.05, **p<.01, ***p<.001.  Development and Evaluation of the PCERQ 158  Appendix C-l Correlations of Self-Reported Relationship Abusiveness with PCERQ Sub-Scales: Overall and By- Sample. Sample  CI  NP  SNA  ER  Verbal  .63***  .14*  .00  25***  Physical  .42***  .15*  .05  .15*  Verbal  .53***  .22**  -.04  .23**  Physical  .23**  .19*  .03  .19*  1 6  03  .20  Overall (n = 221)  UDS (n = 144)  TGS (n = 50)  Verbal  57***  Physical  .44 ***  .23  .08  .00  .60***  .01  .06  .39*  CS (n = 27)  Verbal  Note: CI = Conflict Ineffectiveness, NP = N-Power, SNA = Standards of NonAbusiveness, ER = Exonerative Rationalizations. * p< .05, ** p< .01, *** p< .001.  Development and Evaluation of the PCERQ 159  Appendix C - 2 Summary of Hierarchical Regression Predicting Self-Reported Verbal Abuse with PCERQ Scales and Other Abuse Predictors. Step / Variable  r  B  SEB  Beta  1.  SDE  -.16*  -.16  .06  -.16*  IM  .01  .11  .08  .09  Lang  .15*  1.13  .57  .12*  Chdjoy  - 25***  -.29  .29  -.07  PAM-Y  .18**  .04  .14  .02  PA D-Y  .17*  -.01  .14  -.01  Open  -.17*  -.02  .07  -.02  Consci  -.04  .03  .08  .03  Agree  _ 35***  -.11  .10  -.08  Extra  -.17*  -.03  .07  -.03  Neurot  27***  -.06  .07  -.07  PAS  39***  .06  .04  .13  Joy now  - 27***  -.19  .26  -.04  CI  .63***  .51  .07  NP  .15*  -.27  .14  -.13  SNA  -.01  .10  .10  .06  ER  .26***  .15  .16  .06  CI*NP  .04  1.10  .44  .16*  NP*SNA  -.08  .00  .41  .00  2.  3.  4.  5  Note:  R-Sa Che  .05*  .08***  14***  54***  19***  .02 -.08 .38 -.48 -.01 NP*ER Mult.R = .70, Adj. R-Sq. = .43. F (20, 180) = 8.47 (p < .001). R-Sq Chg = R-Square Change of block at point of entry (prior to entry of subsequent blocks). SDE = Self Deceptive Enhancement, IM = Impression Management, Lang=Language spoken at home growing up, Chdjoy = joy in childhood, PA M-Y = physical abuse by mother, PA D-Y = physical by dad; Block 3 = Openness, Conscientiousness, Agreeableness, Extraversion, Neuroticism; PAS = Propensity for Abuse Scale; CI = Conflict Ineffectiveness, NP = N-Power, SNA = Standards of Non-abusiveness, ER = Exonerative Rationalizations. * p < .05. ** p < .01. *** p < .001.  Development and Evaluation of the PCERQ 160  Appendix C-3 Summary of Hierarchical Regression Predicting Physical Abuse with PCERQ Scales and Other Abuse Predictors. Step / Variable  r  B  SEB  Beta  1.  SDE  .01  .01  .02  .06  IM  -.10  -.03  .02  -.08  Chdjoy  - 24***  -.03  .09  -.02  PAM-Y  .38***  .14  .04  .23**  PAD-Y  35***  .05  .04  .09  Open  - 23***  -.05  .02  -.17*  Consci  -.04  .01  .02  .04  Agree  -.26***  .00  .03  -.01  Extra  -.18**  -.03  .02  -.10  Neurot  .20**  -.04  .02  -.15  37***  .01  .01  .11  .23***  -.12  .08  -.10  41 * * *  .08  .02  29***  2.  3.  PAS Joy now 4.  5  Note:  CI NP  .14*  .04  .04  .07  SNA  .07  .06  .03  .12  ER  .15*  -.06  .05  -.09  CI*NP  .13  .24  .13  .12  NP*SNA  -.06  -.06  .12  -.03  NP*ER  .20**  .21  .11  .13  R-Sa Che.  .01  ig***  IQ***  .06**  .04*  Mult.-R = .63, Adj. R-Sq. = .33, F (20,180) = 6.14 (p < .001). R-Sq Chg = R-Square Change of block at point of entry (prior to entry of subsequent blocks). SDE = Self Deceptive Enhancement, EVI = Impression Management, Chdjoy = joy in childhood, PA M-Y = physical abuse by mother PA D-Y = physical abuse by dad; Block 3 = Openness, Conscientiousness, Agreeableness, Extraversion, Neuroticism; PAS = Propensity for Abuse Scale; CI = Conflict Ineffectiveness, NP = N-Power, SNA = Standards of Non-Abusiveness, ER = Exonerative Rationalizations. * p < .05. ** p < .01. *** p < .001.  Development and Evaluation of the PCERQ  161 Appendix D-l Power. Conflict Efficacy and Responsibility Questionnaire: 90-Items (PCERO-90). PCERQ The following questions refer to your attitudes and behavior in your current intimate relationship. Please indicate the extent to which the following items describe aspects of your experience by circling the appropriate number. strongly disagree disagree 1 2  neutral 3  agree 4  strongly agree 5 1 2 3 4 5  1. )  In my relationship I like to be the one who guides the decisions.  2. )  Sometimes I get so upset, confused or frustrated with my partner 1 2 3 4 5 that I don't know what to do next.  3. )  During conflicts with my partner we sometimes stay angry with 1 2 3 4 5 each other for quite a while.  4. )  I learned about handling relationship conflict in a fair manner from 1 2 3 4 5 someone close to me.  5. )  For me, there are right and wrong ways to conduct myself during 1 2 3 4 5 relationship conflicts.  6. )  Although I may feel an urge to say or do something mean-spirited to my 1 2 3 4 5 partner, I have found better ways to deal with those feelings.  7. )  I don't mind if my partner suggests I try doing something differently.  1 2 3 4 5  8. )  When my partner and I have a disagreement things sometimes speed up and get out of control.  1 2 3 4 5  9. )  My partner and I have long talks where we express a lot of anger, sadness, or disappointment with one another.  1 2 3 4 5  10. )  I learned about conflictfromsomeone who was constructive and 1 2 3 4 5 fair in relationship disputes.  11. )  I would choose to put right and wrong aside in order to avoid conflict.  1 2 3 4 5  Development and Evaluation of the PCERQ  162  Appendix D-l PCERO-90. strongly disagree disagree 1 2  neutral 3  agree 4  strongly agree 5  12. )  Sometimes you have to just let it all out in a disagreement with your partner.  13. )  I speak my mind with my partner when I think something he/she 1 2 3 4 5 is doing could be done better.  14. )  The unhappy times of our relationship just happen regardless of what I am doing.  1 2 3 4 5  15. )  It can be hard for me to let go of feeling aggravated with my partner  1 2 3 4 5  16. )  During disagreements I am aware that how I act could determine if my partner will become more upset or even leave the relationship. 1 2 3 4 5  17. )  I've done quite a few things my partner disapproves of and did not tell him/her about it to avoid any further problems. 1 2 3 4 5  18. )  If I were to get abusive in a dispute, it would be because I had not tried hard enough to prevent it.  19.)  Td rather not give advice to my partner unless it is asked for.  20. )  How well I get along with my partner depends very much on how he/she is feeling that day.  1 2 3 4 5  1 2 3 4 5 1 2 3 4 5  1 2 3 4 5  21. )  There are times when my partner believes he/she is superior to me. 1 2 3 4 5  22. )  My parents were hard on me at times, but I don't hold it against them. 1 2 3 4 5  23. )  I've promised myself that under no circumstances will I be mean or insulting with my partner.  24. )  1 2 3 4 5  It is my deliberate effort that keeps me from getting too intense in 1 2 3 4 5 disputes with my partner.  Development and Evaluation of the PCERQ  163 Appendix D-l PCERO-90. strongly disagree disagree 1 2  neutral 3  agree 4  strongly agree 5  I often think about how we make decisions as a couple.  1 2 3 4 5  When discussions with my partner leave me feeling upset, I have ways of dealing constructively with those emotions and situations. 1 2 3 4 5 When my partner and I have an argument, we still show that we care for one another. 1 2 3 4 5 I believe that the fear or intimidation that can sometimes come up in conflicts goes away with time.  1 2 3 4 5  Trying to conduct your intimate relationship according to personal or societal ideals is unrealistic. 1 2 3 4 5 My partner and I can get along well in spite of the most trying circumstances if we decide to. I impress my partner by what I say and do.  1 2 3 4 5 1 2 3 4 5  I feel confident that I can talk with my partner in ways that can 1 2 3 4 5 lead to good resolutions to our problems. When wefightwe sometimes get stuck in uncomfortable feelings that last long time. 1 2 3 4 5 I can tell if my partner is going to compromise on issues we disagree about. 1 2 3 4 5 I become overly preoccupied about how my partner is doing in times when he/she is under a lot of stress. 1 2 3 4 5 My partner can do or say things which make me so irritated I fly off the handle. I debate with my partner and try to convince him/her into my point of view.  1 2 3 4 5 1 2 3 4 5  Development and Evaluation of the PCERQ 164 Appendix D-l PCERO-90 strongly disagree disagree 1 2  neutral 3  agree 4  strongly agree 5  38. )  Maintaining a smooth functioning relationship is a skill; things like luck don't come into it. 1 2 3 4 5  39. )  When I communicate that I have a problem with something in the relationship, my partner tends to get angry. 1 2 3 4 5  40. )  It really bothers me that how I deal with emotions in my relationship 1 2 3 4 5 problems is similar to how one or both of my parents did.  41. )  I don't always try to understand my partner's problems because 1 2 3 4 5 then we both get upset.  42. )  I sometimes think I have the potential to be hard on my partner 1 2 3 4 5 and keep this in mind during conflicts.  43. )  I like to see my partner's emotional reactions to what I say and do.  1 2 3 4 5  44. )  Good communication between partners is simply a matter of learning and applying the skills.  1 2 3 4 5  45. )  When relationship problems come up it crosses my mind that my partner may not be quite as smart or in touch with the situation as I am. 1 2 3 4 5  46. )  I learned from family and/orfriendshow to deal with disagreements in a comfortable way while growing up. 1 2 3 4 5  47. )  I sometimes find it hard to be completely sympathetic for my partner's difficult situations because my own life can be so challenging. 1 2 3 4 5  48. )  I am pretty accurate when it comes to keeping track of how emotional I'm getting during conflicts with my partner.  1 2 3 4 5  49. )  My partner tells me that I expect too muchfromhim / her.  1 2 3 4 5  50. )  There are things I can do willbetter. help to end an argument with 1 2 3 4 5 my partner that leave us that feeling  Development and Evaluation of the PCERQ  165 Appendix D-l PCERO-90 strongly disagree disagree 1 2  neutral 3  agree 4  strongly agree 5  I try to show some positive feelings along with the more difficult ones when we have a conflict. 1 2 3 4 5 Things can be said during a dispute that make matters worse for a long time afterwards.  1 2 3 4 5  I try to figure out if there is anything I can do to make a situation 1 2 3 4 5 easier for my partner. A desire to be direct and non-abusive determines how I act in conflicts with my partner.  1 2 3 4 5  It irritates me when my partner asks me to do something, or do it 1 2 3 4 5 differently than I am. As time goes on IfindI have a better understanding about how my partner and I can best cope with our disagreements. 1 2 3 4 5 If my partner dishes out anger at me, I just do the same.  1 2 3 4 5  I really worry sometimes that I may make some of the same mistakes with my family that my parents did. 1 2 3 4 5 Sometimes I'm unable to take my partner completely seriously when it comes to believing how bad things are for him/her.  1 2 3 4 5  It is too difficult to "fight fair" every time some kind of problem comes up in the relationship. I value commanding the respect of my partner  1 2 3 4 5 1 2 3 4 5  There seems to be a split-second emotional reaction that I can't control in some conflicts with my partner. 1 2 3 4 5 We go through some uncomfortable times in this relationship but they don't last for very long. 1 2 3 4 5  Development and Evaluation of the PCERQ 166  Appendix D-l PCERO-90 strongly disagree disagree 1 2  neutral 3  agree 4  strongly agree 5  64. )  I have shared with my parents my feelings about some things they did during my childhood or adolescence that were hurtful to me. 1 2 3 4 5  65. )  I don't really have any personal guidelines about what is OK to do during relationship conflict. 1 2 3 4 5  66. )  If I can tell we're going to have a big argument orfight,I know I'll have to keep myself in check and not get out of hand. 1 2 3 4 5  67. )  I try to calm my partner down if he / she is upset about something.  1 2 3 4 5  68. )  Conflicts with my partner tend to build and build and become hard to control. 1 2 3 4 5  69. )  Even if we have some strong words for one another, we still express care when we have relationship problems. 1 2 3 4 5  70. )  There have been people in my life who have shown a real interest in helping me get along better with others. 1 2 3 4 5  71. )  Though it can be hard, I am usually able to see my partner's view in arguments we have.  1 2 3 4 5  72. )  If my partner does something that gets me really angry or upset I end up feeling like I have no good ways of dealing with that situation. 1 2 3 4 5  73. )  Seduction is an important part of how I get together with a new partner.  1 2 3 4 5  74. )  I feel good about how I react and respond in relationship conflicts. 1 2 3 4 5  75. )  If we express anger to one another the other person gets angry back. My understanding of how conflicts work in my relationship has undergone a great deal of growth.  76.)  1 2 3 4 5 1 2 3 4 5  Development and Evaluation of the PCERQ 167  Appendix D-l PCERO-90 77. )  It doesn't really matter whether both partners "fight fair" when disputes arise.  1 2 3 4 5  78. )  People's expectations are too high when it comes to always being respectful of your partner during conflict. 1 2 3 4 5  79. )  My partner is attracted to me partly because of my status and/or potential for status.  1 2 3 4 5  80. )  My partner and I manage to sort out problems in an effective way. 1 2 3 4 5  81. )  It is OK for us to feel upset or angry with one another about something because we are able to work it out and feel better. 1 2 3 4 5 Even though I'm older, my relationship with my parents is still affected by unpleasant things about how they raised me. 1 2 3 4 5  82. ) 83. )  I've made choices to change how I act and what I say when I have fights with my partner. 1 2 3 4 5  84. )  It is up to me to see to it that no harm comes to my partner during fights or disagreements we may have.  1 2 3 4 5  85. )  I teach or guide my partner in how to do things better.  86. )  I sometimes feel at a real loss for what to do next in a relationship conflict.  1 2 3 4 5  1 2 3 4 5  87. )  I value my partner's input, concerns and feelings during conflict.  88. )  Nobody has helped me learn how to "fight fair" in close relationships. 1 2 3 4 5 I have made decisions to keep myselffromsaying or doing just anything during relationship conflict. 1 2 3 4 5 My partner ought to know that if he/she drives me crazy about something I can't always be responsible for my reactions. 1 2 3 4 5  89. ) 90. )  1 2 3 4 5  Development and Evaluation of the PCERQ 168  Appendix D-2 Conflict Tactics Scale: Family of Origin Form. We would now like you to try to remember how the members of your family dealt with conflict when you were growing up. We are interested in the following relationships: 1. your father and your mother (or male and female guardians) 2. your father (or male guardian) and you 3. your mother (or female guardian) and you 4. the sibling with whom you ad the most conflict and you please check sex male female For the relationships listed below, please rate how often each person used each conflict tactic by placing the appropriate number (0-4) in the space provided. Please leave blanks for any relationship that you did not have during your childhood [e.g., if you are an only child]. 0 never  1 at least once  2 occasionally Dad Mom  a. Discussed issue calmly b. Got information to back up your/their side c. Brought in someone to help settle things d. Argued heatedly but short of yelling. e. Insulted, yelled or swore at the other one. f. Sulked and/or refused to talk about it. g. Stomped out of the house or room. h. Cried. i. Did or said something to spite the other one. j. Threatened to hit or throw something at the other one. k. Threw or smashed or hit or kicked something. 1. Threw something at the other one. m. Pushed, grabbed or shoved the other one. n. Slapped the other one. o. Kicked, bit or hit with a fist. p. Hit or tried to hit with something. q. Beat up the other one r. Threatened with a knife or gun. s. Used a knife or gun.  3 frequently Dad You  4 weekly Mom You  Sib You  Development and Evaluation of the PCERQ  169 Appendix D-3 Conflict Tactics Scale - Marital Form (CTS-MF) No matter how well two people get along, there are times when they disagree on major decisions, get annoyed about something the other person does, or just have spats or fights because they are in a bad mood or tired, or for some other reason. They also use different ways of trying to settle their differences. Below are listed a number of behaviours that people use to settle their differences. Please indicate the number that best represents how often in the past year you and your partner have used these behaviours when dealing with each other. 0 never  1 once  2 twice  3 3-5 times  4 6-10 times  YOU- in past year a. Discussed issue calmly b. Got information to back up your/their side c. Brought in someone to help settle things d. Argued heatedly but short of yelling. e. Insulted, yelled or swore at the other one. f. Sulked and/or refused to talk about it. g. Stomped out of the house or room. h. Cried. i. Did or said something to spite the other one. j. Threatened to hit or throw something at the other one. k. Threw or smashed or hit or kicked something. 1. Threw something at the other one. m. Pushed, grabbed or shoved the other one. n. Slapped the other one. o. Kicked, bit or hit with a fist. p. Hit or tried to hit with something. q. Beat up the other one r. Threatened with a knife or gun. s. Used a knife or gun.  5 11-20 times  6 more than 20 times  PARTNER - in past year  Development and Evaluation of the PCERQ  170 Appendix D-4 Propensity for Abuse Scale (PAS). Part 1 For each of the statements below, please circle the number to the right of the statement that most accurately describes how it applies to you, from 1 (completely undescriptive of you) to 5 (completely descriptive of you). 1 completely undescriptive 1.  3.  2 3 mostly partly undescriptive undescriptive  4 mostly descriptive  5 completely descriptive  I can make myself angry about something in the past just by thinking about it  1 2 3 4 5  I get so angry, I feel that I might lose control.  1 2 3 4 5  If I let people see the way I feel, I'd be considered a hard person to get along with.  1 2 3 4 5  Part 2 For each of the statements below, please indicate how true it is about you by circling the appropriate number. 1 never true  2 seldom true  3 sometimes  4 often true  5 always true  true 1 2 3 4 5  4.  I see myself in totally different ways at different times.  5.  I feel empty inside.  6.  I tend to feel things in a somewhat extreme way, experiencing either great joy or intense despair.  1 2 3 4 5  It is hard for me to be sure about what others think of me, even people who have known me very well.  1 2 3 4 5  I feel people don't give me the respect I deserve unless I put pressure on them.  1 2 3 4 5  7.  8.  1 2 3 4 5  Development and Evaluation of the PCERQ 171 Appendix D-4 PAS Somehow, I never know quite how to conduct myself with people.  1 2 3 4 5  Part 3 Please read each of the following statements and rate the extent to which it describes your feelings about romantic relationships by circling the appropriate number. Think about all of your romantic relationships, past and present, and respond in terms of how you generally feel in these relationships. Not at all like me 1  Somewhat like me 3  Very much like me 5  4  10.  I find it difficult to depend on other people.  1 2 3 4 5  11.  I worry that I will be hurt if I allow myself to become too close to others.  1 2 3 4 5  12.  I am somewhat uncomfortable being close to others  1 2 3 4 5  Part 4 How often have you experienced each of the following in the last two months? Please circle the appropriate number. 0 never  1 2 occasionally  fairly often  3 very often  13.  Insomnia (trouble getting to sleep)  0  2  3  14.  Restless sleep  0  2  3  15.  Nightmares  0  2  3  16.  Anxiety attacks  0  2  3  17.  Fear of women  0  2  3  18.  Feeling tense all the time  Development and Evaluation of the P C E R Q 172  Appendix D-4 PAS 19.  Having trouble breathing  0  1 2  3  Part 5 Beside each statement, please write the number of the response listed below that best describes how often the experience happened to you with your mother (or female guardian) and father (or male guardian) when you were growing up. If you had more than one mother/father figure, please answer for the persons who you feel played the most important role in your upbringing. 1 never  2 occasionally  3 often  4 always Father  20.  My parent punished me for even small offenses.  21.  As a child I was physically punished or scolded in the presence of others.  22.  My parent gave me more corporal (physical) punishment than I deserved.  23.  I felt my parent thought it was my fault when he/she was unhappy  24.  I think my parent was mean and grudging toward me.  25.  I was punished by my parent without having done anything.  26.  My parent criticized me and told me how lazy and useless I was in front of others.  27.  My parent would punish me hard, even for trifles.  28.  My parent treated me in such a way that I felt ashamed.  29.  I was beaten by my parent.  Mother  Development and Evaluation of the PCERQ 173  Appendix D-5 Balanced Inventory of Desirable Responding (BIDR). Please rate the extent to which the following items are true or not for you. Not True 1  2  3  4  5  6  True 7  1.  My first impressions of people usually turn out to be right.  12 3 4 56 7  2.  I don't care to know what other people really think of me.  12 34 56 7  3.  When my emotions are aroused, it biases my thinking.  12 3 4 56 7  4.  I am fully in control of my own fate.  12 3 4 56 7  5.  I never regret my decisions.  12 3 4 56 7  6.  My parents were not always fair when they punished me.  12 34 56 7  7.  I am very confident with my judgments.  12 34 56 7  8.  It's all right with me if some people happen to dislike me.  12 34 56 7  9.  I sometimes tell lies if I have to.  12 3 4 56 7  10.  I never cover up my mistakes.  12 3 4 5 6 7  11.  There have been occasions when I have taken advantage of someone.  12 34 56 7  12.  I always obey laws, even if I'm unlikely to get caught.  12 34 5 6 7  13.  When I hear people talking privately, I avoid listening.  12 34 56 7  14.  When I was young I sometimes stole things.  12 3 4 5 6 7  15.  I never read sexy books or magazines.  12 3 4 56 7  16.  I have taken sick-leavefromwork or school even though I wasn't really sick.  12 3 4 5 6 7  Development and Evaluation of the PCERQ 174 Appendix D-6 Marital Locus of Control Scale (MLOCS). MLOC Please indicate the extent to which you agree or disagree with the following statements as descriptive of you and your present relationship. 1 Strongly Agree  2  3  4  5  6 Strongly Disagree  1.  I am often at a loss as what to say or do when I'm in disagreement with my spouse. 1 2 3 4 5 6  2.  Sexual incompatibility is something of a mystery to me; it is something that just happens. 1 2 3 4 5 6  3.  Putting effort into the relationship will practically guarantee a successful marriage. 1 2 3 4 5 6  4.  Difficulties with my spouse often start with chance remarks.  5.  If my marriage were to end in divorce, I would suspect that I had not tried enough to make it work. 1 2 3 4 5 6  6.  The unhappy times in our marriage just seem to happen regardless of what I am doing. 1 2 3 4 5 6  7.  Circumstances play a very limited role in causing marital satisfaction; it is largely a matter of effort. 1 2 3 4 5 6  8.  When things begin to go rough in my marriage I can see that I had a part to play in it. 1 2 3 4 5 6  9.  Raising children effectively is really just a matter of trying one's best; chance has absolutely nothing to do with it. 1 2 3 4 5 6  10.  Ifindthat external circumstances like day-to-day events can have considerable influence on how my spouse and I get along. 1 2 3 4 5 6  11.  I can always bring about a reconciliation when my spouse and I have an argument. 1 2 3 4 5 6  1 2 3 4 5 6  Development and Evaluation of the PCERQ  175 Appendix D-6 MLOCS 1 Strongly Agree  2  6 Strongly Disagree  12.  When I want my spouse to do something she/he hadn't planned on, it's often difficult to bring her/him to my way of thinking. 1 2 3 4 5 6  13.  Misunderstandings between my spouse and myself are generally purely circumstantial. 1 2 3 4 5 6 Some effort on my part is all that is required in order bring about pleasant experiences in our marriage. 1 2 3 4 5 6  14. 15.  Having a satisfactory sexual relationship with one's spouse is partly a matter ofluck. 1 2 3 4 5 6  16.  When we have unpleasant experiences in our marriage I can always see how I have helped to bring them about. 1 2 3 4 5 6  17.  Circumstances of one sort or another play a major role in determining whether my marriage functions smoothly. 1 2 3 4 5 6  18.  My spouse and I can get along happily in spite of the most trying circumstances. 1 2 3 4 5 6  19.  If parents discipline their children conscientiously they are sure to be well behaved. 12 3 4 5 6  20.  If my spouse and I were to experience sexual difficulties we would certainly be able to overcome them. 1 2 3 4 5 6  21.  Successful child-rearing is a result of some good hick along the way. 1 2 3 4 5 6  22.  If my marriage were a long, happy one I'd say that I must just be very lucky.  23.  Even with the most loving couples a mutually satisfying emotional relationship doesntjust happen, it is the result ofthe couple workingatit. 1 2 3 4 5 6  1 2 3 4 5 6  Development and Evaluation of the PCERQ  176 Appendix D-6 MLOCS 1 Strongly Agree  2  3  4  5  6 Strongly Disagree  24.  At times, there doesn't seem to be any way out of a disagreement with my spouse. 1 2 3 4 5 6  25.  It seems to me that maintaining a smooth functioning marriage is simply a skill; things like luck don't come into it. 1 2 3 4 5 6  26.  Good communication between spouses is simply a matter of learning and applying the skills; nothing can really interfere with good communication  1 2 3 4 5 6  27.  It's more often up to my spouse to make an argument end peaceably.  28.  How well your kids grow up depends very much on external factors like what kind of neighborhood you live in. 1 2 3 4 5 6 If my sexual relationship with my spouse was not entirely satisfactory, I would say that I wasn't putting enough effort into the relationship. 1 2 3 4 5 6  29.  1 2 3 4 5 6  30.  When I look over the course of my marriage I can't help but wonder if it wasn't destined that way. 1 2 3 4 5 6  31.  Good clear communication between spouses doesn't depend on things like compatibility and personality but on constant practice. 1 2 3 4 5 6  32.  Couples who run don't run into any marital conflict at some point in their marriage simply have been very lucky. 1 2 3 4 5 6  33.  A little planning can prevent most of the conflicts that occur between spouses over child-rearing. 1 2 3 4 5 6  34.  Problems in our marriage never seem to sort themselves out over time; we usually end up having to do something about them. 1 2 3 4 5 6  35.  I seem to have relatively little influence over when the intimate moments in our marriage will occur, they seem to happen oftheir own accord. 1 2 3 4 5 6  36.  My spouse and I get along well because we have the interpersonal skills; not because of things like luck or temperament. 1 2 3 4 5 6  Development and Evaluation of the P C E R Q 177  Appendix D-6 MLOCS 1 Strongly Agree  2  3  4  5  6 Strongly Disagree  37.  My spouse's moods are often mysterious to me in that I have little idea as to what may have set them off. 1 2 3 4 5 6  38.  There are always things I can do that will help to end an argument with my spouse that leave us feeling better. 1 2 3 4 5 6  39.  Some kids are unmanageable in spite of their parents' best efforts at discipline. 1 2 3 4 5 6  40.  Couples who have a satisfying emotional relationship are constantly trying to improve their relationship; a good relationship doesn't just develop spontaneously. 1 2 3 4 5 6  41.  When my spouse and I are communicating effectively we aren't doing anything in particular to make it happen. 1 2 3 4 5 6  42.  How well I get along with my spouse depends very much on how he/she is feeling that day. 1 2 3 4 5 6  43.  Happy times in our marriage don't just happen by chance; planning is usually required. 1 2 3 4 5 6  44.  Something more than a couple's intentions and abilities is needed to bring about a mutually satisfying emotional relationship; it's really a kind of special magic that is there or isn't. 1 2 3 4 5 6  Development and Evaluation of the PCERQ 178  Appendix D-7 Conflict Narrative Form. Narrative Please answer the following questions in 3-4 sentences. Describe any aspects of your experience (emotions, thoughts, behaviours, wishes) that seem relevant. 1. H o w would you respond or react i f your partner says there are some major problems in your relationship and he/she needed some time along to figure things out?  2. H o w would you respond or react i f you want to spend an evening with your same-sex friends but your partner wants to spend time with you?  3. H o w would you react or respond if, without talking to you, your partner makes decisions about how the two of you should spend your time?  4. H o w would you react or respond if you felt like your partner takes out the bad stuff about his/her day on you?  5. Please give an example of how you have reacted and responded to a recent conflict in your relationship. Briefly describe the problem issue and what happened.  Development and Evaluation of the PCERQ  179 Appendix D-8 Coding Scheme for Conflict Narratives. The conflict narratives ask respondents to write short descriptions of how they would react and respond to hypothetical and actual conflicts in their relationship. The following thematic areas were examined in coding these narrative responses. 1. TONE (emotional tone): This category examines the emotional expression aspects of the narratives. The nature of how conflict is perceived and experienced by the writer is assessed. This item is scored with the following scheme: 0=at least moderately strong positive affect in response to conflict (e.g., "I don't mind when we argue") l=neutral or mixed positive and negative affect (e.g., "It didn't bother me too much when she came home late") 2=moderately to strong negative affect (e.g., "I can't take it when she doesn't show up on time -1 didn't speak to her for the rest of the evening") 3=lack of awareness or no mention of affect as part of conflict process, (e.g., "I came in late and she went to bed." 2. NEGATIVE ATTRIBUTIONS: For this category each attribution of resentment, blame, or hostility towards partner is recorded. Each instance of such an attribution in a given narrative will receive one point. Viewing one's partner as controlling, capable of or intending to do harm, being naive or stupid and being inconsiderate all count in this category, (e.g., "She is always doing that to get on my nerves"; "She doesn't care"; "She doesn't listen, no matter how often I tell her") 3. POSITIVE ATTRIBUTIONS: For this category each attribution of appreciation, care, respect, or faith in one's partner is recorded. Each instance of such an attribution will receive one point. Viewing partners as open, cooperative, helpful, honest, and pleasant would all count in this category, (e.g., "I know she'll tell me what is on her mind"; "She wouldn't do anything to hurt me"; "She keeps our best interests at heart") 4. PROBLEM SOLVING: In this category the number of problem-solving behavior, ideas, or attempts is recorded. Listening to the opinions of one's partner is included as an instance of problem-solving behavior. Any action initiated so as to take the couple out of  Development and Evaluation of the PCERQ 180  a static, unproductive dynamic counts for one point. Low scores reflect lack of agency, Appendix D-8 (con't) Coding Scheme for Conflict Narratives. passivity or complacency, whereas higher scores reflect a more agentic approach to conflict management and resolution, (e.g., "I would listen to herfirstwithout interrupting (counts for 1) and probably take a walk to think it over (counts for 1)"; "I never know what to do when this happens (counts for 0)") 6. NEED FOR POWER: same as the Winter schema. 7. RESPONSIBILITY: same as the Winter schema.  Development and Evaluation of the PCERQ  181 Appendix D-9 Rosenberg Self-Esteem Scale RSE Please circle the number that best describes how the statement applies to you. 1 completely undescriptive 1.  2 3 4 mostly partly mostly undescriptive undescriptive descriptive  5 completely descriptive  I feel that I'm a person of worth, at least on an equal basis with others.  1 2 3 4 5  2.  I feel that I have a number of good qualities.  1 2 3 4 5  3.  All in all, I am inclined to feel that I am a failure.  1 2 3 4 5  4.  I am able to do things as well as most people.  1 2 3 4 5  5.  I feel I do not have much to be proud of.  1 2 3 4 5  6.  I take a positive attitude toward myself.  1 2 3 4 5  7.  On the whole, I am satisfied with myself.  1 2 3 4 5  8.  I wish I could have more respect for myself.  1 2 3 4 5  9.  I certainly feel useless at times.  1 2 3 4 5  10.  At times I think I'm no good at all.  1 2 3 4 5  Development and Evaluation of the P C E R Q 182  Appendix D-10 The SPQ (ItemsfromPCERO-90 Previously Aggregated in a Pilot Study of Couples. All of Which Appear on the CI Sub-Scale) 2.  Sometimes I get so upset, confused or frustrated with my partner that I don't know what to do next.  8.  When my partner and I have disagreements things sometimes speed up and get out of control.  36.  My partner can do or say things which make me so irritated I fly off the handle.  Development and Evaluation of the PCERQ 183 Appendix D-11 Demographic Form 1.  Age  Family of Origin: 2.  Where were you born? Canada or the USA Asia Europe Africa Central or South America Middle East Other: _  3.  Where were your birth parents born? Use M for Mother and D for Father Canada or the USA Asia Europe Africa Central or South America Middle East Other:  4.  Was English the language spoken in your household as a child? yes some no  5.  Did you have brothers and sisters growing up? yes no  6.  Number of sisters  7.  Were you:  8.  Please circle how much responsibility you had in looking after younger siblings during your childhood years: none some a lot  9.  Generally, how happy or pleasant were your childhood years? not enjoyable most enjoyable 1 2 3 4 5 6 7 8 9  Number of brothers  the youngest child  a middle child  the oldest child  Development and Evaluation of the PCERQ  184 Appendix D-11 Demographics Form 10.  Please check the situation that best describes the family you grew up in two-parent family one-parent family step-family other:  11.  Mother's occupation:  12. Mother's highest education:  13.  Father's occupation:  14. Father's highest education:  Current Life and Relationship: 17.  Marital Status: Single, never married Married and living together Separated or divorced Widowed Common-law (living together at least 6 months) Other:  If married or in a common-law relationship: 18.  Your partner's occupation:  19.  Partner's highest education:  20.  For how many years have you been married/common-law with your partner?  21.  Generally speaking, how happy or pleasant is your current relationship? not enjoyable most enjoyable 1 2 3 4 5 6 7 8 9  22.  Which of you do feel is more physically attractive? You Equal 1 2 3 4 5 6  7  Partner 8 9  23.  Who is more committed to the relationship, you or your partner? You Equal Partner 1 2 3 4 5 6 7 8 9  24.  If you and your partner decided to end your relationship, whose life would it disrupt more? You Equal Partner  Development and Evaluation of the PCERQ 185  Appendix D-ll Demographics Form 25.  What is your approximate family income now? _ less than $20,000/year _ $20,000 - $50,000/year _ $50,000 - $100,000/year _ more than $100,000/year  26.  How many times have you been arrested?  Development and Evaluation of the PCERQ 186  Appendix D-l2 UDS "Telling Stories" Instructions  TELLING STORIES You are going to see a series of pictures, and we would like you to tell stories about the people shown in the pictures. Try to imagine what is going on. Then tell what happened before, what the people are thinking about and feeling, and what will happen next. Try to write a story, with a beginning and end, about each picture. When you are told to begin, turn to the first page of the Pictures Booklet and look at the first picture for a few seconds. Then turn to the Storytelling Booklet and write your story. You will find these questions spaced out on the to help you write your story. What is happening? Who are the people? What happened before? How did the story begin? What are the people thinking about, and how do they feel? What will happen? How will the story end? These questions are only guidelines for your stories; you need not answer each one specifically. You will have about 5 minutes for each story. You will be told when it's about time to go on to the next story. There are no right or wrong stories. Write whatever kind of stories you like.  Development and Evaluation of the PCERQ 187 Appendix E-l Undergraduate Sample Consent Form  « U B C Dept of Psychology Letterhead» CONSENT FORM Project Title: POWER, RESPONSIBILITY AND PERSONALITY IN RELATION TO ABUSE IN DATING RELATIONSHIPS I freely and voluntarily agree to be a participant in this research project entitled "Power, Responsibility and Personality in Relation to Abuse in Dating Relationships". Dr. Donald Dutton of the UBC Psychology Department s the principal investigator. He may be contacted a 604-822-2151 should you have any concerns. This study is designed to investigate the relationship between a number of attitudes and experiences people have and how they affect behaviour in a marital-type of relationship. More specifically, we are interested in investigating the attitudes people have and how these attitudes shape their relationships. I understand that I will write very brief stories (aboutfiveminutes per picture) based on five picture that will be presented to me. This phase should take approximately 30 minutes. Then I will complete a questionnaire which will take about 1 hour. For participating in this study I will be awarded 1.5 credits (1 per hour) toward afirstor second year Psychology class. I understand that all information is confidential and that this sheet will be removed from the experimental materials to ensure my anonymity. I understand that withdrawal of my consent of my refusal to participate at any time throughout the study will in no way jeopardize or affect my class standing, and that I may withdraw from the study at any time without penalty or consequences. I have read and understood the foregoing.  participant's signature  date  Development and Evaluation of the PCERQ 188  Appendix E-2 Treatment Group Consent Form « U B C Dept. of Psychology Letterhead» Consent Form Project Title: FACTORS RELATED TO ABUSE IN RELATIONSHIPS I freely and voluntarily agree to be a participant in this research project entitled "Factors Related to Abuse in Relationships". Dr. Donald Dutton of the UBC Psychology Department s the principal investigator. He may be contacted a 604-822-2151 should you have any concerns. This study is designed to investigate the relationship between a number of attitudes and experiences people have and how they affect behaviour in a marital-type of relationship. More specifically, we are interested in investigating the attitudes people have and how these attitudes shape their relationships. I understand that I will complete a questionnaire which will take between 1 and 1.5 hours to complete. After completing the questionnaire booklet I will mail it in the envelope provided. Once the booklet has been checked by a lab assistant (I may receive a phone call to answer incomplete questions) I will be mailed $12 for participating. I understand that all the information is confidential (data will not be seen by anybody other than the UBC researchers; treatment group leaders will not know of my scores). This sheet will be removedfromthe experimental material to ensure my anonymity. I understand that withdrawal of my consent or my refusal to participate at any time throughout the study will in no way jeopardize or affect my treatment group involvement, and that I may withdraw form the study at any time without penalty or consequences. I have read and understood the foregoing.  (participant's name - please print) (participant's signature)  phone: mailing address:  date  Development and Evaluation of the PCERQ  189 Appendix E-3 Community Sample Consent Form. « U B C Dept of Psychology Letterhead» Consent Form Project Title: FACTORS RELATED TO ABUSE IN RELATIONSHIPS I freely and voluntarily agree to be a participant in this research project entitled "Factors Related to Abuse in Relationships". Dr. Donald Dutton of the UBC Psychology Department s the principal investigator. He may be contacted a 604-822-2151 should you have any concerns. This study is designed to investigate the relationship between a number of attitudes and experiences people have and how they affect behaviour in a marital-type of relationship. More specifically, we are interested in investigating the attitudes people have and how these attitudes shape their relationships. I understand that I will complete a questionnaire which will take between 1 and 1.5 hours to complete. After completing the questionnaire booklet I will mail it in the envelope provided. Once the booklet has been checked by a lab assistant (I may receive a phone call to answer incomplete questions) I will be mailed $12 for participating. I understand that all the information is confidential (data will not be seen by anybody other than the UBC researchers). This sheet will be removedfromthe experimental material to ensure my anonymity. I understand that withdrawal of my consent or my refusal to participate at any time throughout the study will in no way jeopardize or affect my treatment group involvement, and that I may withdraw form the study at any time without penalty or consequences. I have read and understood the foregoing.  (participant's name - please print) (participant's signature)  phone: mailing address:  date  Development and Evaluation of the PCERQ 190  Appendix E-4 Debriefing Form. « U B C Dept of Psychology Letterhead» "Factors Relations to Abuse in Relationships" Thank you for your participation in our study titled "Factors Related to Abuse in Relationships" conducted under Dr. Donald Dutton, principal investigator. The study you have participated in is designed to investigate the relationship between a number of attitudes and experiences people might have and how they affect behaviour in an intimate relationship. More specifically, we are interested in studying the attitudes people have and how these attitudes shape their marital-style relationship. We are also investigating attitudes about power, self-confidence and responsibility. We hope that this research will lead to some answers as to why some people have successful and fulfilling close relationships while others do not. We hope to be able to design some programs to help people who have such difficulties in their personal lives. All information gathered is confidential and the anonymity of all participants is ensured, as data collected are identified by an identity number only.. If filling out these questionnaires or writing short stories has caused you any concerns, feelfreeto call us at 604-822-2151 to discuss them or to get a referral to a counselor. Additional written material is available in Dr. Donald Dutton's (1995) The Domestic Assault of Women (2nd Edition). Vancouver: UBC Press. Further information will be available in approximately six monthsfromnow by contacting Andrew or Kelley at the above number. Thank you for participating in our study.  


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



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"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items