Validation of a Self-Report Measure of Unrealistic Relationship Expectations Heather M. Foran and Amy M. Smith Slep Stony Brook University, State University of New York Cognitive models of intimate partner aggression implicate maladaptive relationship beliefs as antecedents to aggression and targets for intervention. However, existing self-report measures of relationship beliefs have failed to differentiate aggressive and nonaggressive individuals, raising questions about their assessment of and role in understanding aggression. To address these concerns, the authors developed and tested a new measure of unrealistic relationship beliefs in a sample of 453 community couples. Structural validity, concurrent validity, discriminant validity, internal consistency, and temporal stability of the new measure were examined. The final scale demonstrated adequate internal consistency ( s .83���.84), test���retest reliability (rs .68���.74), and concurrent validity (small to moderate associations with predicted variables). Unrealistic relationship beliefs significantly differentiated aggressive and nonaggressive men, and this association remained significant after the authors controlled for other related variables. Keywords: intimate partner violence, unrealistic expectations, psychometric properties, assessment Supplemental materials: http://dx.doi.org/10.1037/1040-3590.19.4.382.supp The role of cognitions in marriage has received considerable attention over the past few decades. Baucom, Epstein, Sayers, and Sher (1989) enumerated several types of cognitions thought to influence couple functioning. These include attributions, assump- tions, standards, selective attention, and expectancies. Attributions are the meanings and explanations an individual assigns to a particular behavior or marital event, assumptions are beliefs about how things are in the relationship or other life circumstances, selective attention refers to idiosyncratic perceptions, and expect- ancies are beliefs about the future. Standards are the beliefs about how things or people should be and are the focus of the current article. Standards may reflect general beliefs or relationship-specific beliefs, and they may vary from healthy to dysfunctional.1 General irrational beliefs are dysfunctional beliefs about oneself and one���s world and have been linked with poor individual functioning (Ellis, 1962), but they appear to be less relevant for relationship functioning than relationship-specific beliefs (Epstein & Eidelson, 1981). Relationship-specific beliefs are beliefs or expectations about the relationship, one���s own behavior in the relationship, or about the partner���s behavior (e.g., partners should read each oth- er���s minds, sexual performance should be perfect Eidelson & Epstein, 1982). Whether high relationship-specific beliefs are healthy or dysfunctional may depend on whether it is realistic they will be met in the relationship. Some studies have found high expectations to be linked to increased satisfaction (Baucom, Ep- stein, Rankin, & Burnett, 1996 DeBord, Romans, & Krieshok, 1996), whereas other studies have found high expectations to be linked with less satisfaction, poorer communication, and more negative marital behaviors (e.g., Bradbury & Fincham, 1993 Eidelson & Epstein, 1982 Emmelkamp, Krol, Sanderman, & Ru ��phan, 1987 Epstein & Eidelson, 1981 Holtzworth-Munroe & Stuart, 1994 Woodward, Carless, & Findlay, 2001). For couples that are highly satisfied and display positive marital behaviors, a specific belief may be met, whereas it may not be met in a dissatisfied relationship characterized by negative interactions. Preliminary support for this hypothesis was found by McNulty and Karney (2004) in a longitudinal study of newlywed couples. They found that marital behaviors and attributions moderated the asso- ciation between change in marital satisfaction and beliefs. Couples with positive marital behaviors and attributions benefited from having high expectations over time, whereas couples with more negative behaviors and attributions saw greater deterioration in satisfaction when they endorsed higher relationship beliefs. Hence, the degree to which the expectations are unrealistic for that par- ticular couple may dictate the impact on the couple. Furthermore, certain types of beliefs may be unrealistic for nearly all relation- ships (e.g., my partner doesn���t agree with all my ideas, so it is unlikely that s/he respects me Eidelson & Epstein, 1982). 1 Standards, when dysfunctional, have also been referred to as irrational beliefs and unrealistic expectations these terms will be used interchange- ably in this article (e.g., Azar, Robinson, Hekimian, & Twentyman, 1984 Ellis, 1986). However, it is important to note that expectancies are not included under this umbrella and have been operationalized differently in the literature (see Baucom et al., 1989). Heather M. Foran and Amy M. Smith Slep, Department of Psychology, Stony Brook University, State University of New York. This research was supported in part by National Institute of Mental Health Grant RO1MH57985. Correspondence concerning this article should be addressed to Heather M. Foran, Department of Psychology, Stony Brook University, Stony Brook, NY 11794-2500. E-mail: heather.foran@sunysb.edu Psychological Assessment Copyright 2007 by the American Psychological Association 2007, Vol. 19, No. 4, 382���396 1040-3590/07/$12.00 DOI: 10.1037/1040-3590.19.4.382 382

One way that unrealistic relationship expectations may nega- tively impact relationships is by leading to blaming the partner when these expectations are inevitably not met. Baucom, Epstein, Daiuto, et al. (1996) reported that the number of unmet standards was related to blaming the partner for the problem, negative affect, and a response style aimed at hurting the partner. Vangelisti and Alexander (2002) found that anger, in particular, was the most common reaction to not having expectations met by one���s partner. Furthermore, unrealistic expectations, which lead to anger and blaming of the partner, may also lead to aggression. Cognitive models of intimate partner violence suggest that aggressive indi- viduals become angry when their partners fail to meet their ex- pectations, which in turn leads to the use of verbal and physical aggression as a method to express such anger and hurt the partner (Meichenbaum, 1977 Novaco, 1978). Empirical evidence support- ing the link between aggression and anger is well established (for reviews, see Norlander & Eckhardt, 2005 Stith, Smith, Penn, Ward, & Tritt, 2004) however, evidence for a link between aggression and unrealistic relationship expectations has been mixed. Partner-aggressive and nonaggressive men have not differed on unrealistic expectations in any of the studies using the currently available self-report measures (Eckhardt, Barbour, & Davison, 1998 Eisikovits, Edleson, Guttmann, & Sela-Amit, 1991 Holtzworth-Munroe & Stuart, 1994). For example, Holtzworth- Munroe and Stuart compared aggressive distressed men, nonag- gressive distressed men, and nonaggressive satisfied controls using the Relationship Beliefs Inventory (RBI Eidelson & Epstein, 1982) and the Inventory of Specific Relationship Standards (ISRS Baucom, Epstein, Rankin, et al., 1996) and only found significant differences between the distressed and satisfied groups. In contrast to the lack of differences found with self-report methodology, differences have emerged when using an articulated- thought methodology (Eckhardt et al., 1998). This methodology (Davison, Robins, & Johnson, 1983) was utilized by Eckhardt and colleagues to induce anger after listening to marital conflict sce- narios and compare online thoughts (irrational beliefs) among aggressive distressed, nonaggressive distressed, and satisfied non- aggressive men. Verbalizations during the anger induction were coded for irrational beliefs. Aggressive men were significantly more likely to have rigid, demanding expectations for the hypo- thetical partner���s behavior than were nonaggressive men. Interest- ingly this study assessed irrational beliefs using both articulated- thought and self-report methodology and only found differences among aggressive and nonaggressive men using the articulated- thought methodology. Accordingly, based on the differential re- sults depending on methodology used, it has been proposed that irrational expectations are most ���accessible��� during the provoking event and that individuals may not be conscious of their expecta- tions to the degree that is necessary to differentiate these beliefs using self-report methodology (e.g., Eckhardt & Dye, 2000). It may indeed be the case that certain unrealistic expectations are only accessible under priming conditions however, it may also be the case that current self-report measures of unrealistic expec- tations of the type that might relate to aggression are not sensitive enough to identify these long-hypothesized aggression��� expectation connections (Holtzworth-Munroe & Stuart, 1994). Ex- isting measures may not focus specifically enough on unrealistic expectations that might be particularly related to violence. Evi- dence from two related literatures has suggested that the latter is worthy of further investigation. In the marital literature, attributions have been the most exten- sively studied cognition. Both self-report and articulated-thought methods of assessing negative, hostile attributions appear to be valid and predict partner aggression (Eckhardt et al., 1998 Fin- cham, Bradbury, Arias, Byrne, & Karney, 1997 Holtzworth- Munroe & Hutchinson, 1993). This suggests that at least one type of relationship cognition can be adequately assessed with self- report methods, and it may imply that additional attention to measure development will result in similarly useful self-report approaches to measuring unrealistic relationship expectations. Of particular relevance, self-report measures of unrealistic ex- pectations in another family dyad, namely parents��� expectations for their children���s behavior, do appear to successfully tap these cognitions. Self-reported unrealistic expectations for child behav- ior distinguish physically abusive and nonabusive mothers (e.g., Azar et al., 1984 Azar & Rohrbeck, 1986). Mothers who hold unrealistic expectations are more likely to interpret their children���s behavior as hostile (i.e., make attributions of hostile intent) and respond with more severe punishment (Azar, 1989). Furthermore, social information processing models of child abuse that explicate the role of unrealistic expectations in the propagation of abuse are consistent with those applied to intimate partner violence (see Holtzworth-Munroe, 1991 McFall, 1982 Milner, 1993). Finally, as is the case with child abuse, unrealistic beliefs are a fairly standard target of intervention in partner violence treatment and prevention programs (e.g., Heyman & Neidig, 1997 Holtzworth- Munroe et al., 1995). Thus, only a few studies have demonstrated differences in relationship expectations among partner-aggressive and nonag- gressive individuals, and many studies have failed to document differences. It remains unclear whether the inconsistent findings are best attributed to a lack of accessibility of unrealistic expec- tations via self-report methodology or to the lack of self-report measures specifically designed to tap unrealistic expectations re- lated to aggression. Considering the theoretical and clinical prom- inence of unrealistic expectations in the partner violence literature, further refinement of their assessment may facilitate developments in both domains. Therefore, given that the marital attribution and child physical abuse literatures indicate that self-report measures can indeed differentiate cognitions of aggressive and nonaggres- sive individuals, we sought to develop and validate a self-report measure that might assess unrealistic expectations in intimate relationships that would relate to partner aggression. Description and Development of the Construct We aimed to develop a measure that specifically assessed un- realistic marital expectations that might better capture the expec- tations that are thought to contribute to partner violence. Currently in the relationship literature, two measures of specific relationship expectations/standards have been studied in relation to aggression: ISRS (Baucom, Epstein, Rankin, et al., 1996) and RBI (Eidelson & Epstein, 1982). These measures do not demonstrate associations with partner violence, but they do distinguish satisfied from dis- tressed couples. The ISRS was designed to be a more comprehen- sive measure of relationship standards, assessing 12 domains (e.g., finances, household responsibilities). It measures a broader range 383 A NEW MEASURE OF RELATIONSHIP EXPECTATIONS

of standards, rather than necessarily unrealistic expectations, as indicated by its positive association with dyadic adjustment (Bau- com, Epstein, Rankin, & Burnett, 1996). The RBI, in contrast, was designed to assess dysfunctional expectations and negatively cor- relates with dyadic adjustment. It assesses two types of expecta- tions for relationships (mindreading and sexual perfectionism) in addition to three types of assumptions about relationships (partners can���t change, disagreement is destructive, and sexes are different). To gain further insight into why these measures might not be sensitive to expectations that relate to aggression whereas self- report measures of unrealistic expectations for children do relate to child abuse, we compared the RBI and ISRS to expectation mea- sures in the parent���child arena (e.g., Child Abuse Potential Inven- tory (CAP) Rigidity subscale Milner, 1986). One difference ap- peared to be the focus of the expectation. Measures of parental expectations were narrowly focused on beliefs about what parents thought children should be like. In contrast, measures of relation- ship expectations were more variably focused and included expec- tations for how respondents themselves should be (e.g., expecting personal sexual perfectionism), how couples generally should be, and what partners should do. We expected relationship expectations or standards focused on the partner to more closely associate with partner violence than these more global relationship expectations, which do relate to general marital distress. Partner-violent men attribute more hostile intent and blame to their partner for ambiguous situations (Holtzworth-Munroe & Hutchinson, 1993), perhaps suggesting that they have unrealistically high expectations for their partner���s behavior specifically, but not for themselves or couples in general. Cognitive views of emotion (e.g., Ortony, Clore, & Collins, 1988) would suggest that although violations of partner-focused expec- tations would be likely to lead to anger (which, in turn, contributes to violence), violations of self-focused and couple-level expecta- tions would be relatively unlikely to elicit anger, instead eliciting anxiety, sadness, or shame. Supporting this distinction, van Dijk and Zeelenberg (2002) found that disappointment caused by an- other person���s behavior led specifically to feelings of anger and seeing the other person as having character flaws, whereas disap- pointment with oneself or one���s situation was linked more with feelings of sadness. Therefore, it appears that the specific focus of the beliefs, rather than just whether the belief is held at a high level, may be particularly important. The ���other-directed perfectionism��� content that is apparent in the CAP Rigidity subscale may explain one reason why other measures of relationship expectations have not been linked to aggression even when the beliefs were held at high levels. Thus, we attempted to capture partner-focused perfection- istic expectations about what a partner should do in a fashion similar to measures of parental unrealistic expectations using some of the content domain from the RBI that has previously been identified as unrealistic beliefs relevant to intimate relationships. Specifically, items assessed expectations for partner perfectionism in mindreading, sexual behavior, household responsibilities, and handling of conflict and marital disagreement. Overview of the Present Study The purpose of this study was to evaluate the reliability and validity of a new measure of unrealistic relationship expectations. We first examined the structure of the scale. Although the content of the items varied (sexual behavior vs. household responsibili- ties), we expected a one-factor structure due to the other-directed perfectionism commonality across items. Similarly, the item con- tent of the CAP Rigidity subscale (Milner & Wimberley, 1979) also varies (cleanliness vs. voicing disagreement with rules), but results have indicated that the items make up one factor, presum- ably due to tapping the construct of other-directed perfectionism. Using the supported factor structure, we then examined (a) reli- ability via internal consistency and temporal stability and (b) concurrent and discriminant validities. We had several specific validity hypotheses. Unrealistic expec- tations should positively relate to perpetration of partner aggres- sion and negatively relate to marital satisfaction. Both marital satisfaction and relationship expectations should independently, and perhaps interactively, predict partner aggression. Specifically, we hypothesized that the association between expectations and aggression would be stronger at lower levels of relationship satis- faction than at higher levels. This hypothesis was based on the assumption that for individuals in more poorly functioning rela- tionships, high expectations are less likely to be met by the partner and thus more likely to lead to anger and aggression. This would be consistent with the interaction found by McNulty and Karney (2004) when examining relationship satisfaction change longitu- dinally. We also hypothesized that unrealistic expectations would positively relate to other cognitive, mood, and marital-functioning variables (e.g., anger and negative affect, desired relationship change, and attributions for partner���s behavior). In addition, it was critical to evaluate the degree of overlap between expectations on the new measure and other established cognitive constructs such as unrealistic parental expectations and attributions, as each type of cognition should independently predict variables of interest. Predictors of partner aggression overlap with variables that predict parental aggression (Slep & O���Leary, 2001) therefore, it is important to determine whether relationship expec- tations, rather than parental expectations, specifically predict part- ner aggression. We hypothesized that unrealistic relationship ex- pectations would be related to, but distinct from, unrealistic parental expectations and that relationship expectations would account for unique variance in partner aggression over and above that predicted by parental expectations. A second cognitive vari- able, attributions, seemed particularly important to discriminate from partner-focused unrealistic expectations because of robust relations between negative partner attributions and partner aggres- sion (Holtzworth-Munroe & Hutchinson, 1993) and because of the hypothesized link between expectations and attributions in social cognitive models of family violence (Holtzworth-Munroe & Stu- art, 1994). Therefore, we hypothesized that expectations and attri- butions would be less than perfectly related and that each would significantly and uniquely predict partner aggression in the context of the other. Method Participants Married or cohabitating heterosexual couples were recruited to participate in a larger study of how families cope with conflict. A total of 453 community couples from Suffolk County, New York, 384 FORAN AND SLEP