Responses to validating versus reframing support strategies as a function of borderline personality features and interpersonal problems

Abstract

We examined whether borderline personality features may be differentially associated with reactions to social support strategies involving validation (acknowledging distress as normal) vs. reframing (highlighting opportunities for positive outcomes). After completing self-report measures of their borderline personality features and interpersonal problems, participants rated their reactions to receiving either validation or reframing support (randomly assigned) from a close friend with whom they imagined sharing hypothetical negative experiences. Borderline personality features were associated with less positive responses to reframing than to validation, and a tendency to interpret reframing support as an assertion of dominance. Moderated mediation analyses further showed that associations between borderline features and negative responses to reframing support were specifically explained by hostile-dominant interpersonal problems. Findings highlight the importance of validation to help individuals high in borderline features experience the emotional benefits of social support.

Keywords:

• borderline
• interpersonal circumplex
• validation
• support
• dominance

1. Introduction

Borderline Personality (BP) is characterized by emotional/behavioral dysregulation and interpersonal difficulties that exacerbate one another (Mendez et al., 2022). While social support is of great importance to wellbeing and recovery from mental health problems, including the high-risk problems commonly associated with BP, individuals with BP perceive low levels of social support (Thadani et al., 2022) and have difficulties maintaining constructive relationships with loved ones and treatment providers (Barnicot et al., 2011; Beeney et al., 2018; Gamache et al., 2018). Research to understand how individuals with BP experience support can therefore have practical benefits.

According to the biosocial theory (Linehan, 1993), BP arises from a combination of biological vulnerabilities and being raised in an invalidating environment (Lynch et al., 2006; Nelson-Gray et al., 2007). Invalidating environments, marked by abuse or trivialization/neglect of emotional experiences, are associated with the development of negative self-views, unpleasant self-conscious emotions, and poorer ability to adaptively regulate emotional states (Gratz, 2003; Putnam & Silk, 2005). Studies of individuals with BP also link early relationship adversity with perceiving others as more threatening (Berenson et al., 2018a) and less trustworthy (Nicol et al., 2013). The interpretation biases associated with BP reduce emotional benefits from positive interpersonal experiences, resulting in persistent feelings of rejection, even when receiving social acceptance in experimental settings (Gutz et al., 2015; De Panfilis et al., 2015) and in daily life (Bhatia et al., 2013; Lazarus et al., 2018). Berenson et al. (2018b) found that undergraduates high in BP features made negative causal attributions for experiences of social acceptance, interpreting others’ kindness as motivated by pity or an attempt to manipulate them. The tendency for individuals high in BP to doubt support providers’ sincerity suggests that positive words of encouragement may often fail to effectively improve their mood.

1.1. Importance of validation in support

Validation is a social support strategy that encourages distressed people to view their expression of negative feelings as permissible and understandable in the context of the circumstances they are experiencing (Marigold et al., 2014). Greville-Harris et al. (2016) examined physiological arousal and behavior after randomly assigning participants to receive validating, invalidating, or no feedback during a stressful task. While invalidating feedback increased sympathetic nervous system activation and social withdrawal, validating feedback increased parasympathetic nervous system activation and social engagement. Feedback acknowledging negative feelings and circumstances appears to communicate social safety.

Validating feedback is also self-verifying, meaning that it conveys being seen by others in ways consistent with self-views (Marigold et al., 2014). Feeling self-verified helps satisfy basic needs for self-coherence and meaning, and is therefore an important human goal (Swann, 1983). This is true even among individuals with a negative self-image, for whom the motive for self-verification can result in feeling more comfortable with criticism than praise (K. B. Wright et al., 2014). Ayduk et al. (2013) randomly assigned individuals to receive either positive or negative feedback. While those with high self-esteem appreciated positive feedback, those with low self-esteem responded with more pleasant nonverbal affect and lower blood pressure reactivity when they received negative feedback more consistent with their own negative thoughts and feelings.

Marigold et al. (2014) found that individuals with low self-esteem typically perceived social support as less helpful than their self-assured peers, in part because they were less likely to receive validating support that met their needs for self-verification. When those with low self-esteem expressed negative feelings, support providers tended to try to cheer them up through reframing, a strategy that encourages optimism by drawing attention to the ways in which negative experiences are ultimately inconsequential/transient or opportunities for growth (Marigold et al., 2014). While reframing is a commonly used and well-intentioned support strategy that is effective for some individuals, multiple studies showed that its effectiveness significantly varies as a function of self-esteem. Among people with low self-esteem—the population that receives reframing most often—reframing was significantly less effective than validation (Marigold et al., 2014).

No prior research has directly compared experiences of validation versus reframing support among individuals with BP. However, consistent with the emphasis on validation in psychotherapy for BP (Carson-Wong et al., 2018; Fonagy et al., 2015; Linehan, 1993), there are reasons to think validation would be an especially important support strategy for this population. Indeed, BP is not only characterized by low self-esteem (Lynum et al., 2008), but also an interpersonal style that prioritizes autonomy to protect from future risk of emotional invalidation.

1.2. Interpersonal perspective on support strategies in BP

The interpersonal circumplex model (Wiggins, 1982) proposes that relationships function most successfully between individuals who demonstrate complementarity by being similar on the dimension of affiliation but opposite on the dimension of dominance (Alden et al., 1990; Kiesler, 1996; Pincus & Gurtman, 2006). Consistent with this idea, research has shown that patients’ perceptions of treatment helpfulness are positively correlated with the extent of complementarity in their relationship with their therapist (Caspar et al., 2005). It is likely that highly dominant individuals would respond less favorably to support efforts that they perceive as expressions of dominance, both within and outside of clinical settings.

Research suggests that advice giving may be commonly perceived as a signal of interpersonal dominance. Interpersonally dominant individuals often try to influence others’ behavior by giving advice (Schaerer et al., 2018). At the same time, these individuals tend to discount the advice they receive from others because they experience it as a threat to their own autonomy (Tost et al., 2012). To the extent that the reframing support strategy entails a recommendation to think/feel a particular way, it may be experienced as a form of advice giving. Thus, distressed individuals with high needs for autonomy may find recommendations to immediately switch out of a negative emotional state and think/feel positively to be a poor fit.

While the interpersonal profiles of individuals with BP show considerable heterogeneity and elevations in interpersonal problems across the entire circumplex (A. G. C. Wright et al., 2013; Leihener et al., 2003; Salzer et al., 2013; Wilson et al., 2017), they are most commonly characterized by high hostile-dominance (Barnow et al., 2009; Bellino et al., 2006; Girard et al., 2017; Zimmermann & Wright, 2017). Notably, this is the interpersonal style that resembles Linehan’s (1993) description of “butterfly-like” BP patients whose strong autonomy needs present difficulties forming stable attachments to therapists. Interpersonal theory suggests that due to their hostile-dominant interpersonal problems, many individuals high in BP would experience a friend’s attempt to comfort them with reframing support as an assertion of dominance that conveys criticism/doubt of their own experience and independent judgment.

1.3. The current research

We investigated whether BP is differentially associated with reactions to validating versus reframing support, using an experimental design in which participants imagined sharing negative experiences with a friend in hypothetical scenarios (adapted from Marigold et al., 2014, Study 1). We hypothesized that participants high in self-reported BP features would respond significantly less positively to reframing than to validation, such that BP would predict lower ratings of friendship quality, positive reactions, self-verification, and perceptions of caring/helpful intent in the reframing condition. Additionally, based on the interpersonal circumplex model we predicted that BP would be associated with heightened perceptions of reframing as an expression of dominance, and that associations between BP and responses in the reframing condition would be mediated by participants’ level of hostile-dominant interpersonal problems (HDIP). To examine the specificity of our hypotheses to BP and HDIP we repeated our analyses with self-esteem as an additional covariate and alternative mediator, and we also conducted additional analyses controlling for general levels of interpersonal problems. Finally, we also explored the extent to which our results were generalizable to both men and women.

2. Method

2.1. Participants

We recruited adults from the United States for an on-line study of personality traits and attitudes about friendships through Amazon’s Mechanical Turk. All participants provided informed consent. Our sample consisted of 238 participants, after excluding another 22 individuals who failed more than one of 6 embedded simple attention checks or left crucial items blank. Participants included 144 men, 93 women, and 1 individual with a nonbinary gender identity. They had a mean age of 35.37 (SD = 10.02) and identified their race/ethnicity as White (76.9%), Black (7.1%), Asian (6.7%), Hispanic/Latinx (6.3%), and Other (2.9%).

2.2. Materials

2.2.1. State mood

Participants completed the Positive and Negative Affect Schedule (Watson et al., 1988) by rating the extent to which they were currently experiencing 10 positive and 10 negative moods (e.g., “Right now, I feel enthusiastic”) from Very slightly or not at all (1) to Extremely (5). We calculated the mean score after reverse-coding the 10 negative items, α = .90.

2.2.2. Trait self-esteem

Participants rated 6 items from the Rosenberg (1965) Self-Esteem Scale (e.g., “On the whole, I am satisfied with myself”) from 1 strongly disagree (1) to strong agree (4). We calculated the mean score after reverse-coding the 3 negative items, a = .93.

2.2.3. Borderline personality (BP) features

Participants responded to 32 true-false items assessing symptoms of BP from the Schedule for Nonadaptive and Adaptive Personality II (SNAP-II; Clark, 2003). We computed BP scores as the number of symptoms endorsed, a = .94.

2.2.4. Interpersonal problems including hostile-dominant interpersonal problems (HDIP)

Participants completed the Inventory of Interpersonal Problems Short Circumplex, (IIP-SC; Soldz et al., 1995), rating 32 items (e.g., “I try to control other people too much”) from Strongly disagree (1) to Strongly agree (5). We computed the total IIP-SC score to capture the full range of interpersonal problems across the dimensions of dominance and warmth, a = .94, and also computed the HDIP subscale from the 12-items in the hostile-dominant quadrant, a = .92.

2.2.5. Hypothetical scenarios

After identifying the name of their closest friend, participants envisioned themselves in two hypothetical scenarios (adapted from Marigold et al., 2014) in a random order. For each scenario, the participant reflected on how they would think, feel, and react. In one scenario, participants imagined telling their close friend about a poor performance evaluation for either a job or a class. In the other scenario, participants imagined telling their friend about a day full of disappointments and mishaps (e.g., slipping on the sidewalk).

2.2.6. Support strategy conditions

We randomly assigned participants to imagine receiving either validation or reframing support from their close friend, as in Marigold et al. (2014). Each participant was in the same support strategy condition for both scenarios. In the validation condition participants were told: “Your friend wants to assure that you he/she understands what you’re going through, and your feelings are normal.” In the evaluation scenario the friend said: “That’s an awful feeling isn’t it, that’s happened to me before too.” For the bad day scenario, the friend said: “That really is a crappy day! I’m sorry to hear that you went through that.” In the reframing condition, participants were told: “Your friend wants to help ease your distress and lighten your mood.” In the evaluation scenario the friend said: “Don’t worry about it too much, it’s just one evaluation. I’m sure you’ll do better next time.” For the bad day scenario, the friend said: “Don’t worry, I’m sure tomorrow will be a better day.”

2.2.7. Ratings of friend’s support

After imagining their friend’s response to each scenario, the participants rated a series of statements about it, each beginning with “My friend’s response”. These items were adapted from Marigold et al. (2014) and were rated from Not at all (1) to Extremely (7). We created scales by averaging ratings of the relevant items across both scenarios. Overall positive reactions to the support were assessed with 6 items per scenario (e.g., “makes me feel better,” a = .88). Self-verification was measured with four items (e.g., “shows that he/she understands the way that I’m feeling,” a = .91). We measured perceived caring intent with four items (e.g., “lets me know he/she cares about me,” “was intended to make me feel good” a = .78). We also developed three new items (a = .91) to assess perceived dominance: “makes me think that he/she is bossy or controlling”, “makes me think that he/she sees me as weak or incapable”, “shows that he/she is impatient or annoyed with me for how I’m feeling.”

2.2.8. Ratings of friendship quality

Finally, participants rated their friendship with the individual they had imagined supporting them in the hypothetical scenarios, e.g., “I have a very strong relationship with my friend” (Marigold et al., 2014). This scale included 12 items (a = .90) rated from Not at all (1) to Extremely (7).

3. Results

3.1. Differential effects of support strategies as a function of BP

Our hypotheses focused on the interaction between support condition and BP predicting friendship quality, reactions to support, self-verification, perceived caring intent, and perceived dominance. As shown in Supplemental Table 1, we regressed all dependent variables on condition, BP, and their interaction, controlling for mood and gender, after standardizing continuous predictors and centering categorical predictors (validation = −.5, reframing = +.5; woman/nonbinary = −.61, man = +.39). Although BP features showed strong zero-order correlations with total interpersonal problems and low self-esteem (see Supplemental Table 2), the results described below were not substantially changed when we controlled for these variables (Supplemental Table 3).

3.1.1. Friendship quality

BP predicted significantly lower ratings of friendship quality, B = −.14, SE = .05, t = −2.64, p = .009. There was no significant main effect of condition, B = −.11, SE = .10, t = −1.13, p = .258, but the BP x condition interaction was significant as hypothesized, B = −.21, SE = .10, t = −2.24, p = .026, ΔR² = .02 (see, Figure 1). As BP increased, ratings of friendship quality significantly decreased in the reframing condition, B = −.25, SE = .07, t = −3.62, p < .001, but not in the validation condition, B = −.04, SE = .07, t = −.48, p = .634. Participants high in BP features (+1 SD) rated their friendship as significantly lower in quality after they imagined receiving reframing M = 5.75, 95% CI [5.56, 5.93] rather than validation, M = 6.07, [5.87, 6.27].

Figure 1. Friendship quality predicted by BP and Condition.

Note: Predicted values control for gender and initial mood. Error bars represent standard errors.

3.1.2. Reaction to support

Reactions to support did not significantly vary with BP, B = −.09, SE = .07, t = −1.32, p = .190, but a main effect of condition showed that on average participants reacted less positively to reframing than to validation, B = −.40, SE = .12, t = −3.27, p = .001. The hypothesized BP x condition interaction was marginally significant, B = −.24, SE = .12, t = −1.94, p = .05, ΔR² = .01 (see, Figure 2). BP was inversely associated with positive reactions to reframing, B = −.21, SE = .09, t = −2.38, p = .018, but not validation, B = .03, SE = .10, t = .29, p = .775.

Figure 2. Reaction to support predicted by BP and condition.

Note: Predicted values control for gender and initial mood. Error bars represent standard errors.

3.1.3. Self-verification

Significant main effects showed that BP predicted feeling less self-verified by the friend’s support, B = −.29, SE = .08, t = −3.60, p < .001, and that reframing predicted less self-verification than validation, B = −.66, SE = .15, t = −4.58, p < .001. The BP x condition interaction was also significant, B = −.37, SE = .15, t = −2.56, p = .011, ΔR² = .02. As shown in Figure 3, BP was significantly associated with feeling less self-verified by reframing, B = −.48, SE = .11, t = −4.59, p < .001, but not by validation, B = −.11, SE = .11, t = −.96, p = .337.

Figure 3. Self-verification predicted by BP and condition.

Note: Predicted values control for gender and initial mood. Error bars represent standard errors.

3.1.4. Perceived caring intent

A significant main effect showed that BP was associated with perceiving less caring intent in their friend’s response, B = −.15, SE = .06, t(232) = −2.46, p = .015. While there was no significant main effect of condition, B = −.16, SE = .11, t(232) = −1.45, p = .149, the BP x condition interaction was significant as hypothesized, B = −.22, SE = .11, t(232) = −2.04, p = .043, ΔR² = .02. The inverse association between BP and perceived caring intent was significant in the context of reframing, B = −.26, SE = .08, t(232) = −3.34, p = .001, but not validation, B = −.04, SE = .08, t(232) = −.48, p = .634, as shown in Figure 4. Participants high in BP (+1 SD) perceived significantly less caring intent when they imagined receiving reframing support, M = 5.37, 95% CI [5.16, 5.58], rather than validation support, M = 5.75, [5.52, 5.97].

Figure 4. Perceived caring intent predicted by BP and condition.

Note: Predicted values control for gender and initial mood. Error bars represent standard errors.

3.1.5. Perceived dominance

Perceiving the friend’s response as expressing dominance was significantly associated with BP, B = .43, SE = .07, t(232) = 6.30, p = .009. There was no significant main effect of condition, B = .17, SE = .12, t(232) = 1.40, p = .162, but the expected BP x condition interaction was significant, B = .33, SE = .12, t(232) = 2.74, p = .007, ΔR² = .02 (Figure 5). BP predicted perceived dominance in the validation condition B = .26, SE = .09, t = 2.80, p = .006, but this association was significantly stronger in the reframing condition B = .59, SE = .09, t = 6.82, p < .001. Participants high in BP (+1 SD) perceived reframing as significantly more dominant M = 2.32, 95% CI [2.08, 2.55] than validation M = 1.82, [1.57, 2.07].

Figure 5. Perceived dominance predicted by BP and condition.

Note: Predicted values control for gender and initial mood. Error bars represent standard errors.

3.2. Moderated mediation by HDIP

To examine whether the differential associations of BP with negative reactions to reframing (versus validation) support was specifically mediated by HDIP, we conducted moderated mediation analyses for each outcome, based on 10,000 bootstrap samples. As shown in Figure 6, these models identified BP as the predictor, HDIP as the mediator, and support condition as the moderator, with initial mood, gender, and total interpersonal problems (IIP-SC) as covariates (PROCESS Model 14; Hayes, 2018). The relationship between BP and HDIP was significant as expected, a = .30, p < .001. All indexes of moderated mediation were significant (as explained below). Notably, none of the indexes of moderated mediation were significant when we identified trait self-esteem scores as the mediator (instead of HDIP) in these models.

Figure 6. Moderated mediation models.

Note: Each model controlled for total interpersonal problems, gender and initial mood.

3.2.1. Friendship quality

HDIP predicted marginally lower ratings of friendship quality, b₁ = −.14, p = .088, and the HDIP x condition interaction was significant, B = −.20, p = .036. Specifically, HDIP predicted significantly lower friendship quality in the context of reframing, b₃ = −.24, p = .009, but not validation, b₂ = −.04, p = .678. The direct effect of BP on friendship quality was not significant, c’ = −.06, p = .401. As hypothesized, there was evidence of a significant moderated mediation, B = −.06, SE = .03, 95% CI [−.13, −.00], with an indirect effect of BP on friendship quality via HDIP when receiving reframing, B = −.07, SE = .04, [−.15, −.01] but not validation support, B = —.01, SE = .03, [.-09, .05].

3.2.2. Reaction to support

Less positive reactions to support were marginally associated with HDIP, b₁ = −.17, p = .098, and the HDIP x condition interaction was significant, B =  -.27, p = .027. HDIP predicted less positive reactions to reframing, b₃ = −.31, p = .009, but not validation, b₂ = −.04, p = .749. The direct effect of BP on reactions to support was not significant, c’ = .06, p = .507. Moderated mediation was significant, B = −.08, SE = .04, 95% CI [−.17, −.01], with a significant indirect effect via HDIP for reframing, B = −.09, SE = .04, [−.18, −.02], but not validation, B = −.01, SE = .04, [−.08, .06].

3.2.3. Self-verification

Self-verification was significantly predicted by HDIP, b₁ = −.24, p = .046, and by the HDIP x condition interaction, B = −.49, p < .001. The inverse association between HDIP and self-verification was significant for reframing, b₃ = −.48, p < .001, but not validation, b₂ = −.01, p = .954. The direct effect of BP on self-verification was not significant, c’ = −.03, p = .783. Moderated mediation was significant, B = −.15, SE = .05, 95% CI [−.25, −.05], with the indirect effect of BP on self-verification via HDIP being significant for reframing, B = −.14, SE = .05, [−.26, −.05]; but not validation, B = −.00, SE = .04, [−.07, .08].

3.2.4. Perceived caring intent

Perceiving the friend as having less caring intent was marginally predicted by HDIP, b₁ = −.16, p = .076; and the HDIP x condition interaction was significant, B = −.22, p = .036. HDIP predicted perceiving less caring intent for reframing, b₃ = −.28, p = .008, but not validation, b₂ = −.05, p = .635. The direct effect of BP on perceptions of caring intent was not significant, c’ = −.06, p = .455. Moderated mediation was significant, B = −.07, SE = .04, 95% CI [−.14, −.00], with an indirect effect of BP on perceived caring intent via HDIP in the context of reframing, B = −.08, SE = .04, [−.16, −.02], but not validation, B = −.02, SE = .03, [−.09, .05].

3.2.5. Perceived dominance

Finally, perceiving the friend’s support as an expression of dominance was significantly predicted by HDIP, b₁ = .28, p = .004, as well as the HDIP x condition interaction, B = .41, p < .001. The association between HDIP and perceived dominance was significant for reframing b₃ = .49, p < .001, but not validation, b₂ = .08, p = .494. Although the direct effect of BP on perceived dominance remained marginally significant, c’ = .15, p = .061, moderated mediation was also significant, B = .12, SE = .05, 95% CI [.03, .22], and showed a significant indirect effect in the reframing, B = .15, SE = .05, [.03, .22], but not validation condition, B = .02, SE = .04, [−.04, .10].

3.3. Gender differences

Although our hypotheses did not explicitly involve gender, we thought it was important to explore whether the patterns of responses to support strategies that we observed were generalizable across gender. Regression analyses (shown in Supplemental Table 4) revealed no significant BP x Gender interactions, and no significant 3-way [BP x Condition x Gender] interactions, indicating that gender did not moderate the results that we previously reported. Analyses of friendship quality and perceived dominance showed no significant effects of gender at all. The other outcomes (positive reaction, self-verification, perceived helpful intent) showed significant Condition x Gender interactions. We further examined these in 2 × 2 analyses of covariance to compare the adjusted marginal means of each outcome variable for men vs. women in each condition, controlling for initial mood.

Reactions to the friend’s support were predicted by a significant Gender x Condition interaction, F (1, 232) =  10.42, p =  .001, ηp² = .04. Men reacted significantly more positively to reframing, M = 5.36, SE = .11, 95% CI [5.14, 5.58] than women did, M = 4.99, SE = .14, [4.71, 5.27]. By contrast, men reacted significantly less positively to validation, M = 5.45, SE = .11, [5.24, 5.67] than women did, M = 5.89, SE = .14, [5.62, 6.16]. Self-verification ratings showed a similar Gender x Condition interaction, F (1, 232) =  6.38, p = .012, ηp² = .03. In the reframing condition, men reported feeling significantly more self-verified, M = 5.66, SE = .14, [5.39, 5.92] than women, M = 5.26, SE = .17, [4.93, 5.60], but in the validation condition they reported feeling significantly less self-verified, M = 6.02, SE = .13, [5.75, 6.28] than women, M = 6.39, SE = .17, [6.06, 6.72]. Finally, the Gender x Condition interaction was also significant when predicting perceived caring intent, F (1, 232) =  13.01, p < .001, ηp² = .05. While men perceived reframing to have more caring intent M = 5.75, SE = .10, [5.56, 5.94], than women did, M = 5.45, SE = .12, [5.21, 5.69], they also perceived validation to have significantly less caring intent, M = 5.59, SE = .10, [5.40, 5.78], than women did, M = 6.08, SE = .12, [5.84, 6.31].

4. Discussion

We investigated responses to supportive feedback as a function of BP traits and HDIP. We hypothesized that relative to individuals low in BP traits, individuals high in BP traits would have less positive reactions to reframing support, and they would perceive it as less helpful, less self-verifying, and more dominant than support that validates their distress. Moreover, in accord with interpersonal theory, we hypothesized that individuals high in BP traits would perceive reframing support as less helpful, specifically because they have a higher level of HDIP. Our results were consistent with these hypotheses, even when controlling for other types of interpersonal problems and self-esteem. Moreover, moderated mediation analyses showed that negative responses to reframing support were associated with BP through its link with HDIP and not with low self-esteem. Overall, our results suggest that reframing may evoke less favorable reactions (compared to validation) due to violations of complementarity among individuals high in BP, who have strong autonomy needs and high levels of HDIP.

While individuals high in BP perceived reframing support as significantly more dominant than validation (consistent with our hypotheses), it is important to note that BP was positively associated with perceptions of dominance in both support conditions. Moreover, while tests of moderated mediation showed that HDIP significantly mediated the association between BP and perceived dominance in the reframing (vs. validation) condition, a nearly significant direct effect of BP on perceived dominance remained, indicating that other elements of the relationship between BP and perceived dominance are unexplained. Given that the items we created to assess perceived dominance focused on hostile-dominance (antagonism), results for perceived dominance are consistent with prior evidence linking BP with more negatively biased perceptions of others’ interpersonal cues, even when these cues are neutral or positive (Berenson et al., 2018b; Gutz et al., 2015; Berenson et al., 2018a; De Panfilis et al., 2015). To some extent, BP predicts heightened suspicion that any supportive response from the friend might reflect an attempt to control or otherwise harm them.

Because Marigold et al. (2014) conducted their research in samples of mostly women, our study extended the literature on support strategies by also considering men. Notably, the differential responses to support strategies that we found as a function of BP did not vary with gender, however, we did find gender differences independent of BP. Specifically, while women responded more favorably to validation than reframing, men showed the opposite preference. This result is consistent with prior studies on gender differences in coping, because validation can be considered a more emotion-focused (rather than problem-focused) support strategy, and women are more likely than men to use coping strategies centered on verbal communication and emotional support seeking (Brougham et al., 2009; Tamres et al., 2002).

Several methodological weaknesses of our research should be noted. One of the most important limitations is the lack of a control condition (e.g., a neutral or baseline condition that does not include a description of the friend providing a supportive response). In other words, while our results suggest that validation is more beneficial than reframing for individuals high in BP, they do not indicate whether validation is more beneficial than the friend saying nothing at all. As noted above, another limitation is that the two items we used to assess perceptions of the friend’s supportive response as expressing dominance were both negatively toned. Future research might consider adding items that assess perceptions of warm/friendly dominance, to better capture the full dominance dimension of the interpersonal circumplex.

Our results are based on a non-clinical sample with self-reported BP features and cannot be generalized to diagnosed patients. Moreover, while perceiving a support provider’s response as “supportive” depends on multiple factors, including nonverbal cues as well as the ongoing relational context (Goldsmith, 2004; Tanner et al., 2018), our study only examined immediate responses to hypothetical support narratives. In addition, because we did not measure pre-test levels of our outcome variables (e.g., friendship quality ratings) prior to the manipulation of imagined support, our analyses could not capture within-person changes. Also, the cross-sectional nature of this study does not allow us to establish temporal precedence in our moderated mediation analyses, and future research should implement longitudinal designs to examine effects of validation and reframing on interpersonal relations over time. The effect sizes we obtained when comparing self-reported immediate responses to hypothetical validation versus reframing support scenarios as a function of individual differences in BP were small. The size of the real-life impact of the difference between these two types of support for people high in BP may also be small, but this remains an open empirical question.

Extending research on individual differences in experiences of support (Marigold et al., 2014) to consider BP features, our study sheds some light on the interpersonal mechanisms through which those high in BP may be less receptive to support involving positive reframing. Our results suggest that reframing may be perceived as an assertion of dominance that would be aversive to the many individuals with BP who have strong autonomy needs. Moreover, our findings that validation support was received less unfavorably than reframing support among individuals high in BP are consistent with clinical work on the importance of validation in therapy for BP. For example, Carson-Wong et al. (2018) found that BP patients whose therapists did not merely reflect their inner experiences but frequently highlighted the normative, understandable logic behind them reported greater improvement in their emotional states over six months of Dialectical Behavior Therapy. Validation may be especially important for effectively conveying support to those high in BP, in everyday social relationships as well as clinical settings.

Author note

Sydney F. Goldberg is now at the Institute for Graduate Clinical Psychology, Widener University. Kaley M. Michael is now at the Department of Psychology, Millersville University.

Data availability

Data and other materials supporting this study are available from the authors or at https://osf.io/afmbx/?view_only=675f80570f9a4b4aafc2f8c6d7ab597f

Ethics approval

This study received Institutional Review Board approval from Gettysburg College (#19-Psych-10) on 6/20/19. All participants provided informed consent.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/23311908.2022.2146280

Additional information

Funding

This research was supported by the Cross-Disciplinary Science Institute at Gettysburg College (X-SIG), as well as Gettysburg College funding for senior projects.