《社会心理学》课程教学资源（文献资料）Close relationship——Attachment Anxiety and Reactions to Relationship Threat - The Benefits and Costs of Inducing Guilt in Romantic Partners.pdf_P6-P8

50% serious, 6% steady) that were on average 2.57 years in length (SD 1.96).2 Materials and procedure. During an initial session, couples completed the scales described below and were given detailed instructions for completing a 3-week daily diary. Relationship satisfaction. Participants rated five items developed by Rusbult, Martz, and Agnew (1998) assessing their relationship satisfaction (e.g., “I feel satisfied with our relationship”; “Our relationship is close to ideal”; 1 strongly disagree, 7 strongly agree). Items were averaged to index relationship satisfaction (M 6.01, SD 0.83, .86). Attachment security. The Adult Attachment Questionnaire (AAQ; Simpson et al., 1996) was competed with reference to romantic relationships in general. Nine items assessed attachment anxiety (e.g., “I often worry that my romantic partners don’t really love me”), and eight items assessed avoidance (e.g., “I’m not very comfortable having to depend on romantic partners”; 1 strongly disagree, 7 strongly agree). Anxiety (M 2.99, SD 1.05, .80) and avoidance (M 2.92, SD 1.04, .77) were positively associated (r .14, p .08). Self-esteem. Participants also completed Rosenberg’s (1965) 10-item self-esteem scale (e.g., “On the whole, I am satisfied with myself”; M 5.24, SD 1.08, .89). Daily diary. At the end of the day for the following 21 days participants completed a web-based questionnaire assessing their relationship-related feelings and behavior that day. On average, participants completed 19.3 diary entries. Daily feelings. Participants were asked to rate the degree to which they experienced various emotions when interacting with their partner that day, including hurt feelings (“I was hurt by my partner”), anger (“I was angry at my partner”), and guilt (“I felt guilty”; 1 not at all, 7 very much). Participants also rated the degree to which they perceived their partner felt hurt that day (“My partner was hurt by me”). Relationship threatening events. We assessed relationship threatening events in two ways. Both couple members rated two items assessing the degree to which they behaved in a potentially hurtful manner toward their partner (e.g., “I acted in a way that could be hurtful to my partner”; “I was critical or unpleasant toward my partner”; 1 not at all, 7 extremely). We used the partner’s reports of hurtful behavior to predict individual’s emotional reactions. Participants also reported the amount of relationship conflict experienced that day (“I experienced conflict or disagreement with my partner”; 1 not at all, 7 extremely), which provided an additional index of rejection-related events that threatened the relationship. Longitudinal follow-up. Nine months after completing the diary, both couple members were contacted separately via e-mail and asked to complete an online questionnaire consisting of the relationship satisfaction measure described above. Fifteen couples dissolved within the 9-month longitudinal period, and a further 13 couples were either not able to be contacted or chose not to complete the follow-up questionnaire, leaving a sample of 50 couples for the longitudinal analyses reported below. Intact versus dissolved couples did not significantly differ across the study variables, with two exceptions: Attachment anxiety and daily hurt feelings were higher for dissolved couples (ts 2, p .05). Retained couples maintained high levels of relationship satisfaction across the follow-up period (M 5.92, SD 1.09, .92). Results Daily analyses: Attachment anxiety, hurt feelings, and partner guilt. Our daily analyses tested whether, compared to individuals low in attachment anxiety, individuals higher in anxiety (1) experienced greater hurt feelings on days they faced threatening relationship events and, when hurt, (2) were perceived by their partners as experiencing higher levels of hurt (suggesting that anxious intimates exaggerated their hurt) and, in turn, (3) had partners who felt greater guilt. We expected these effects to be specific to hurt and not anger. All analyses followed Kenny, Kashy, and Cook’s (2006) recommendations for analyzing repeated measures dyadic data using the MIXED procedure in SPSS 20. We included a number of covariates across analyses. To control for shared variance across different forms of insecurity, we controlled for the main effect of attachment avoidance. To isolate the unique associations of hurt, we controlled for the positive association between hurt and anger (hurt ¡ anger B .72, t 52.51, p .01) by either controlling for anger when predicting hurt (and vice versa) or including hurt and anger as simultaneous predictors (see Lemay et al., 2012). To capture residual change in the outcome variables, we also controlled for the level of the outcome variables the prior day. The intercept was modeled as random, and because we wanted to make direct comparisons across high and low anxiety at the same levels of relationship threat or hurt feelings, all predictors were grand-mean centered.3,4 Anxious reactions to threatening relationship events. To illustrate the analysis strategy, we tested the degree to which anxious individuals reacted with greater hurt feelings when encountering hurtful partner behavior by modeling hurt feelings on day i as a function of (a) hurt feelings on day i 1, (b) anger on day i to ensure the associations were specific to hurt, (c) attachment avoidance, (d) attachment anxiety, (e) hurtful behavior reported by the partner on day i, and (f) the interaction between the partner’s hurtful behavior and attachment anxiety. The results are shown in the top left of Table 1. The predicted interaction between anxiety and partners’ hurtful behavior was significant and is plotted in Panel A of Figure 1. We 2 The results reported did not differ according to age, relationship length (log-transformed), and relationship status (cohabiting vs. not), with the exceptions that the greater hurt and lower anger reported by anxious individuals when facing daily conflict were stronger when participants were older and (for anger) in longer, cohabiting relationships. 3 Comparable results emerged using person-mean centering and when conducting the analyses excluding these covariates. The one exception involved the links between attachment anxiety and anger (see Table 1): When hurt feelings were not controlled, the main and interaction effects of anxiety on anger were not significant. These null effects remain supportive of our overall conclusion that hurt is a primary response of individuals high in attachment anxiety and suggest that prior investigations capturing “negative” reactions of anxious individuals were more likely to be assessing hurt-based responses rather than anger. Moreover, this pattern highlights the importance of controlling for the associations between emotions that inevitably covary but have differential antecedents and consequences, such as hurt and anger (see Lemay et al., 2012). 4 We also tested the main and interaction effects of gender across analyses. Only two effects significantly differed between men and women. The interaction between attachment anxiety and daily conflict on hurt feelings shown in Table 1 was stronger for men (B .08, t 6.39, p .01) than women (B .03, t 2.32, p .05), but both were significant. The relatively lower anger in response to hurtful partner behavior by more anxious individuals was also significant for men (B –.07, t 4.13, p .01) but not for women (B –.01, t 0.54, p .59), despite this same reaction to conflict replicating for both men and women. This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 240 OVERALL, GIRME, LEMAY, AND HAMMOND

consider two sets of simple effects to evaluate the meaning of the interaction. First, comparing the slopes of the lines in Figure 1A, the effect of hurtful partner behavior on hurt feelings was greater for individuals higher (dashed line; b .20, t 10.01, p .01) versus lower (solid line; b .08, t 3.53, p .01) in anxiety. Second, contrasting the differences between low and high in anxiety on days of low (left side of Figure 1A) versus high (right side of Figure 1A) hurtful partner behavior, anxious intimates felt greater hurt when their partner behaved in hurtful ways (b .16, t 4.99, p .01) but felt the same low levels of hurt in the absence of hurtful partner behavior (b .00, t 0.05, p .96). Thus, anxious individuals experienced greater hurt specifically on days they faced relationship threats. Analogous models predicting daily anger revealed the opposite pattern (shown in the right top of Table 1 and Panel B of Figure 1). Comparing the effects of partners’ hurtful behavior on anger at low versus high anxiety, individuals lower in anxiety (solid line; b .16, t 6.83, p .01) responded with greater anger compared to individuals higher in anxiety (dashed line; b .07, t 3.00, p .01). Examining differences between low and high anxiety on days of low (left side of Figure 1B) versus high (right side of Figure 1B) hurtful partner behavior, intimates higher in anxiety reported less anger when their partner behaved in hurtful ways (b –.12, t 3.43, p .01) and the same low levels of anger in the absence of hurtful partner behavior (b –.01, t 0.13, p .90). Thus, intimates higher in anxiety responded to relationship threats with less anger compared to individuals lower in anxiety. The pattern shown in Figure 1 was replicated when assessing reactions to conflict and disagreement with the partner (see bottom half of Table 1). Individuals higher in attachment anxiety reacted with greater hurt and less anger compared to intimates lower in Table 1 The Effects of Attachment Anxiety and Threatening Relationship Events on Daily Feelings of Hurt and Anger (Study 1) Attachment anxiety and threatening relationship events Predicting hurt Predicting anger B SE t B SE t Response to partner’s hurtful behavior Prior day criterion .05 .01 3.93 .09 .01 6.16 Other emotion .63 .01 46.18 .70 .02 45.60 Avoidance .03 .03 0.98 .07 .03 2.39 Anxiety .08 .03 2.94 .06 .03 2.09 Partner’s hurtful behavior .14 .02 8.98 .11 .02 6.68 Anxiety Partner’s Hurtful Behavior .06 .01 4.45 .04 .01 3.04 Response to conflict Prior day criterion .05 .01 3.98 .08 .01 6.74 Other emotion .55 .02 34.75 .55 .02 34.42 Avoidance .01 .03 0.49 .06 .03 2.37 Anxiety .07 .02 2.86 .06 .03 2.31 Conflict .15 .01 12.35 .25 .01 21.26 Anxiety Conflict .05 .01 6.15 .03 .01 3.96 p .05. p .01. Figure 1. The moderating effect of individuals’ attachment anxiety on the links between daily levels of partner’s hurtful behavior and individuals’ daily feelings of hurt (Panel A) and anger (Panel B) reported every day for a 3-week period (Study 1). This figure presents two separate two-way interactions (see the top half of Table 1). Panel A graphs predicted values of individuals’ hurt feelings, and Panel B graphs predicted values of individuals’ anger, as a function of their partner’s hurtful behavior and individuals’ level of attachment anxiety. High and low values are indexed at 1 SD above and below the mean. This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. ATTACHMENT ANXIETY AND PARTNER GUILT 241

anxiety. These results support that hurt feelings constitute a primary response to threatening events by anxious intimates, consistent with their dependence and relationship maintenance concerns. Attachment anxiety and guilt induction. Next, we tested whether (a) anxious intimates hurt was perceived to be more intense than the same level of hurt experienced by individuals low in anxiety and, thus, (b) generated greater guilt in the partner. Given equivalent levels of hurt, if partners perceive anxious individuals to feel more hurt than low anxious individuals, this may indicate that anxious intimates express and exaggerate their hurt more. However, anxious individuals should exaggerate hurt only when they are actually feeling hurt and not on days of no or low hurt when the threat that triggers hyperactivating strategies is absent. Thus, the expected difference should emerge on days of greater hurt feelings and not on days of very low levels (or no) hurt. This predicted pattern would be supported by a significant interaction between individuals’ anxiety and self-reported hurt feelings. Adopting the same dyadic approach as above, we predicted the partner’s perceptions of the individual’s hurt feelings on day i by (a) the partner’s perceptions on day i 1, the individual’s (b) attachment avoidance, (c) attachment anxiety, and (d) hurt feelings on day i, and (e) the interaction between attachment anxiety and hurt on day i, which tests our primary prediction. To show the hypothesized exaggerated expressions were specific to hurt, we also included the individual’s (f) anger on day i, and the (g) Anxiety Anger interaction. The results shown in the first column of Table 2 support our prediction. The significant interaction is displayed in Figure 2, Panel A. Partners perceived greater hurt feelings on days individuals’ experienced greater hurt, but this was stronger when individuals were high (b .40, t 11.03, p .01) versus low (b .29, t 9.19, p .01) in attachment anxiety. Revealing the hypothesized distinction, on days individuals experienced high levels of hurt (see right side of Figure 2A), anxious intimates’ hurt was perceived by their partners as more intense than non-anxious intimates hurt experiencing the same high levels of hurt (b .13, t 2.60, p .01). At very low (or no) hurt, when the lack of threat means guilt-induction strategies are not needed, there were no differences in the partner’s perceptions of hurt feelings across levels of anxiety (left side of Figure 2A; b –.02, t 0.37, p .71).5 We conducted parallel analyses predicting the partner’s guilt to test whether the partners of high (vs. low) anxious individuals felt greater guilt on days that individuals reported experiencing hurt. The results are shown in the second column of Table 2, and the significant interaction shown in Panel B of Figure 2. Partners generally reacted to individuals’ hurt feelings with greater guilt, but this tendency was significantly more pronounced for partners of individuals high (b .22, t 5.46, p .01) versus low (b .11, t 2.99, p .01) in anxiety. Specifically, on days that individuals experienced hurt feelings (right side of Figure 2B), the partners of anxious intimates reported greater guilt than partners of non-anxious intimates (b .13, t 1.96, p .05), but there were no differences in levels of partner guilt when individuals were not experiencing hurt feelings (left side of Figure 2B; b –.03, t 0.46, p .65). This pattern supports that when anxious individuals are hurt they express their hurt in ways that magnify the resulting guilt felt by their partner. We also tested whether the heightened guilt in partners of anxious individuals occurred because partners perceived anxious individuals to be more hurt—the indicator that anxious individuals were strategically expressing hurt to induce guilt. To do this, we reran the analyses predicting partner guilt, including the partner’s perceptions of hurt feelings as an additional predictor, and used procedures recommended by MacKinnon, Fritz, Williams, and Lockwood (2007) to compute asymmetric confidence intervals for the indirect effects. The interaction between anxiety and hurt feelings on partner’s guilt was no longer significant (B .04, t 1.59, p .11) when partner’s perceptions of hurt were controlled (a strong predictor of the partner’s guilt; B .38, t 19.89, p .01), suggesting that the greater hurt perceived by partners of anxious intimates lead to the greater guilt felt by those partners on days anxious individuals experienced high levels of hurt (indirect effect .02, 95% CI [.01, .03]). Taken together, these results suggest that anxious intimates express or exaggerate their hurt, which leads to partners perceiving more intense hurt feelings and subsequently feeling greater guilt. In addition, although greater anger was also associated with greater perceptions of hurt and guilt in the partner, this was not magnified for individuals high in attachment anxiety (see Table 2) indicating that these guilt-inducing dynamics are specific to anxious individuals’ experiences and strategic expression of hurt rather than anger. Longitudinal analyses: Anxiety, partner’s guilt, and relationship satisfaction across time. Our final set of analyses examined whether the partner’s guilt experienced across the diary period was associated with changes in relationship satisfaction across the subsequent 9 months.6 Averaging partner guilt across the 3-week period provided an index of the degree to which couples’ interactions across the course of their day-to-day life typically resulted in the partner feeling guilty, which our diary analyses illustrated was amplified by anxious individuals’ expressions of hurt. Thus, any negative links between partners’ guilt and later satisfaction would indicate that these guilt-relevant processes can undermine relationship satisfaction (or vice versa). We first 5 An alternative test is to examine whether partners perceive anxious individuals to feel more hurt than the hurt actually reported by those individuals. The most up-to-date technique for testing bias in perceptions (West & Kenny, 2011) involves an equivalent analytic strategy with the exception that the partners’ perceptions of individuals’ hurt feelings (the dependent variable) are first centered on individuals’ actual self-reported hurt feelings (the predictor) so that the predicted values (as plotted in Panel A of Figure 2) represent the difference between partners’ perceptions of hurt feelings and individuals’ actual self-reported hurt feelings. Accordingly, this approach produced an identical pattern. On days individuals experienced high levels of hurt, partners of anxious intimates were more likely to overestimate the intensity of that hurt compared to partners of non-anxious intimates, indicating that anxious individuals express more hurt than they are actually feeling. 6 We also tested whether the same effects occurred when predicting diary-rated relationship evaluations. There was no evidence that partner’s guilt protected anxious individuals’ relationship evaluations in the shortterm as it did across time in Studies 1 and 2. However, the appeasing benefits of guilt will be muted in the short-term aftermath of specific relationship threats because any reassurance provided by partner guilt may simply dampen but not reverse the heightened reactivity of anxious individuals. Instead, the reassurance provided by partner guilt will build across time as anxious intimates typically encounter partner guilt in response to their proximity-seeking efforts. This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 242 OVERALL, GIRME, LEMAY, AND HAMMOND