EMS and interpersonal problems

Interpersonal problems are a form of psychological distress associated with our interpersonal relationships with an individual or group of individuals [1]. Interpersonal problems are among the most common reasons people seek psychological support and are often reported in clinical populations as well as being key diagnostic markers of personality disorders [2, 3]. Recurrent interpersonal problems also contribute to poorer treatment outcomes in therapy including increased dropout rates, increased disordered symptoms and lower remission rates [4, 5]. Therefore, understanding how recurrent interpersonal problems are caused and maintained is essential for guiding treatment and improving outcomes for people who experience difficulties with mental health.

Researchers have long argued that recurring interpersonal problems may arise from “schemas” or maladaptive beliefs about ourselves, others and the world [6–8]. Early maladaptive schemas (EMSs) form the foundation of Schema Therapy, which is currently one of the most comprehensive theories that describes the role of schemas in psychopathology [7]. It is theorised that EMSs arise from unmet psychological needs during childhood such as needs for autonomy, safety or connection [7]. Young, Klosko and Weishaar [7] argue that EMSs go on to form the building blocks of our personality, guiding how we think, feel and behave in response to a range of situations in the future.

To date, there have been a large number of studies supporting a link between EMSs and interpersonal problems [e.g. 9–16]. People who report higher levels of EMSs have been found to report higher levels of aggression [14, 15], domestic violence victimisation [16], interpersonal conflict [17], marital and peer relationship dissatisfaction [18, 19], bullying victimisation [20], and unhelpful interpersonal behaviour such as avoidant, vindictive, controlling or submissive patterns of behaviour [1, 21]. A recent meta-analysis found a moderately strong relationship between EMSs and interpersonal problems, and EMSs related to beliefs about disconnection and rejection had stronger associations with interpersonal problems when compared to other EMSs [22]. Together, these studies provide support for an association between EMSs and interpersonal problems, particularly for schema related to rejection.

EMSs and experimental designs

Researchers still do not fully understand how EMSs are associated with interpersonal problems. Most of the research investigating EMSs and interpersonal problems has been limited to cross-sectional studies [22]. It is theorised that people with EMSs are more likely to experience negative cognitions, emotional distress and engage in maladaptive behavioural responses when their schemas are activated by stressful or schema congruent contexts in their environment [7]. For example, people who endorse schemas related to rejection are predicted to anticipate rejection, experience greater emotional distress and engage in maladaptive behavioural responses when presented with situations involving rejection (i.e., schema congruent contexts). However, it is also likely that greater interpersonal problems give rise to more maladaptive beliefs and expectations about others, meaning that interpersonal problems could increase EMSs endorsement [7]. Given, EMSs and interpersonal problems might impact one another, further experimental research is needed to investigate Young, Klosko and Weishaar’s [7] theory that EMSs increase problematic interpersonal responses.

One of the challenges of assessing a causal relationship between EMSs and problematic interpersonal responses is that individual differences such as EMSs are challenging to manipulate in an experiment. However, given EMSs are theorised to increase problematic interpersonal responses in stressful or schema congruent situations [7], we would expect to find people with high levels of EMSs to show increased problematic interpersonal responses in situations where there is an interpersonal stressor or high degree of congruence between the situation and the schema. In contrast, according to the Schema Therapy model [7] we would expect to find people with high EMSs to show no differences in interpersonal responses when compared to people with low EMSs in situations with low levels of interpersonal stress or low schema congruence. Therefore, while we are unable to manipulate EMSs, researchers could manipulate the interpersonal situation and assess if EMSs moderate the relationship between the interpersonal situation and any problematic interpersonal responses.

Manipulating the interpersonal situation in the investigation of schemas was used in early research predating Young, Klosko and Weishaar’s [7] theory of EMSs. For example, Koch used lexical decision tasks to assess if people reporting higher expectations about rejection were faster at identifying words related to rejection when compared to words related to acceptance [23]. In this study, Koch [23] found that people with high levels of rejection schema were faster to identify rejection words after being primed with rejection words while people with low levels of rejection schema showed no differences in speed of identifying rejection words regardless of priming. In another study, Baldwin and Meunier [24] found that people with anxious attachment orientations responded faster to rejection words when primed with rejecting interpersonal scenarios when compared to people with more secure attachment orientations. These findings suggest that manipulating the interpersonal situation allows researchers an opportunity to assess how responses to interpersonal situations can be altered in people with higher or lower levels of EMSs. However, one disadvantage of lexical decision tasks is that responses are measured as reaction times to identifying words which provide little insight into the complex range of cognitive, emotional and behavioural responses a person might have in response to a challenging interpersonal situation.

Interpersonal responses and impact of context

Different kinds of problematic interpersonal responses to interpersonal situations have been studied in a number of cross-sectional studies. Pierce and Lydon [25] found that participants with negative expectations about interpersonal relationships reported increased emotional distress and more avoidant coping when presented with a written interpersonal scenario about a stressful interpersonal situation. Participants with negative interpersonal expectations also reported increased emotional distress, negative attributions and maladaptive coping responses after reading through interpersonal vignettes that depicted negative behaviour from a romantic partner [26]. Together, these studies suggest that people with negative beliefs or expectations tend to show increased negative cognitions, distress and maladaptive behaviour in response to stressful interpersonal situations. However, these studies did not attempt to control for, or manipulate, the interpersonal context [25, 26]. This limitation prevents researchers from drawing any causal conclusions between the interpersonal context and interpersonal responses. To address these limitations, the above mentioned schema activation paradigms such as those utilised in early lexical decision tasks [23, 24] and measures of interpersonal responses such as those assessed by interpersonal vignettes [25, 26] could be combined by manipulating the interpersonal context using a series of written vignettes to assess if EMSs moderate the impact of interpersonal situations on interpersonal responses. Such an experimental design could assess if EMSs are associated with changes in the way we respond to challenging interpersonal situations.

Understanding how people with EMSs respond across different interpersonal contexts is also an area largely neglected by the research investigating EMSs and interpersonal problems [22]. The importance given to interpersonal contexts is argued by interpersonal theorists who have found that interpersonal behaviour is open to change across different contexts such as increased interpersonal pressures or responses from others [27]. For example, Farc et al. [28] found that people who experienced physical abuse during childhood were more likely to perceive aggression in children when the cues in the environment were ambiguous [28]. Similar findings have been found in other studies showing increased endorsement of hostility schemas to be associated with increased perceptions of hostile intentions in others when information about another person’s behaviour was ambiguous [29]. Furthermore, people with higher rejection expectations reported increased withdrawal from a group when rejection in an interpersonal scenario was unambiguous compared to ambiguous [30, 31]. Together these findings suggest that individuals with EMSs are more likely to report negative responses (e.g. negative cognitions and behavioural responses) to interpersonal contexts depicting rejection and that responses are likely to become increasingly negative as the ambiguity about rejection in the scenario decreases. These findings are consistent with theoretical models depicting schemas (e.g. Schema therapy model, the rejection sensitivity model) arguing that people with stronger negative beliefs or schemas of themselves and others are more likely to perceive rejection in the environment and that perceptions of rejection are associated with higher levels of distress and more negative interpersonal behaviour responses [7, 31]. Relationship contexts such as whether a relationship is romantic, friendly or familial might also impact problematic interpersonal responses in people with EMSs [22]. For example, Fuhrman, Flannagan and Matamoros [32] argued that people have significantly higher expectations about their romantic partner’s behaviour when compared to friends and family. However, most studies investigating the relationship between EMSs and interpersonal problems have been restricted to self-report questionnaires that measure general trait interpersonal tendencies devoid of any context [22]. Therefore, controlling for or comparing interpersonal responses across different contexts can provide researchers with more information about the role of EMSs in interpersonal problems.

Study aims and hypotheses

The aim of the present study was to develop a new experimental paradigm to assess if EMSs moderate the impact of interpersonal contexts on interpersonal responses. Given that individuals with rejection schema have been found to have stronger associations with interpersonal problems [22], degree of rejection was manipulated in a series of interpersonal vignettes depicting acceptance, ambiguous rejection and rejection. Measuring degree of rejection based on acceptance, ambiguous and unambiguous rejection conditions is consistent with previous studies investigating cognitive, emotional and behavioural responses in people with negative relationship expectancies such as rejection sensitivity and hostility schema [e.g. 33, 35]. The present study’s design also controlled for responses to scenarios depicting friends, family members and romantic partners. Based on the Schema Therapy model proposed by Young, Klosko and Weishaar [7] and earlier research investigating relational schemas, the following hypotheses were formulated:

1. Hypothesis 1: Participants will report increased negative interpersonal responses including negative cognitions, emotional distress and behavioural responses as degree of rejection increases across interpersonal scenarios.
2. Hypothesis 2: EMSs will moderate the relationship between the degree of rejection and negative interpersonal responses including negative cognitions, emotional distress and behavioural responses. That is, participants reporting higher EMSs will report higher levels of negative responses and participants reporting lower EMSs will report lower levels of negative responses to interpersonal rejection as the degree of rejection increases.

Sample size

This research was conducted with approval from the University of New England Human Research Ethics Committee (Approval Number: HE18-248) with all participants providing informed and signed written consent for participation. We conducted a multi-level power analysis using the computer software MLPowSim [33] for a linear mixed model with 13 predictors including differences across individual vignettes, rejection conditions, total EMSs score and the interaction between EMSs and rejection conditions. With a type I error rate of 5% and power greater than 90% we estimated 150 participants were required for the present study.

Participants

Participants were recruited from the first-year psychology student pool at the University of New England, Australia. The University of New England is a regional university with many mature age students who attend externally across Australia. Any person who was living in Australia, spoke English and was 18-years or older was allowed to participate in the present study. Participants were given a $15 supermarket (i.e., Coles) gift card for participation. In total, 176 participants were recruited for the experiment. Eighteen participants did not complete the entire experiment so they were excluded from the dataset due to missing data. As a result, the study sample consisted of 158 participants (n = 43, 27.2% male; n = 114, 72.2% female; n = 1, 0.6% other). Participant ages ranged from 19 to 72 (M = 39.53, SD = 10.44, eight participants did not report their age). Twenty-nine participants were single (18.4%), 40 participants were in a de-facto relationship (25.3%), 66 participants were married (41.8%), five participants were separated (3.2%), 14 participants were divorced (8.9%), and four participants were widowed (2.5%). Four participants reported that their highest level of education achieved was 11 years of schooling (2.5%), 25 had completed 13 years of schooling (15.8%), 28 participants completed a vocational qualification (17.7%), 78 had completed a bachelor’s degree (49.4%), 21 had completed a master’s degree (13.3%), and two had completed a doctorate (1.3%).

Measures

Experiment stimuli

A series of nine interpersonal vignettes were written for the online experiment that depicted scenarios involving interpersonal rejection. To control for the type of relationship, three scenarios related to an interaction with a friend, three with a close family member, and three with a romantic partner. Each individual vignette had three variations of the scenario intended to provoke acceptance (control condition), ambiguous rejection (ambiguous condition), or clear rejection (rejection condition). Degree of rejection was manipulated in each of the nine interpersonal scenarios by changing the way the person in the scenario responded towards the participant including engaging with the participant’s goals in the scenario (control), not engaging the participant’s goals without providing information to identify a cause for non-engagement (ambiguous) and not engaging the participant with information suggesting that the person did not want to engage the participant (rejection). Each vignette and its variation across conditions is listed in S1 Appendix.

Responses to the interpersonal vignettes were broken down into questions about the cognitive, emotional and behavioural responses to each vignette. Participants were first asked to rate the degree of perceived rejection experienced on a sliding scale between 0 (Not at all) and 100 (Severely rejected) after reading each vignette both as a manipulation check and as a cognitive measure of perceived rejection. Negative cognitive responses were developed with reference to Weiner’s [47] attribution theory which identifies three casual dimensions in attribution making: locus of control (internally caused or externally caused), stability and globality over an event. In the current experiment, participants were asked to rate the likelihood of any negative internal, external, intentional, global and stable attributions about the vignettes on a sliding scale from 0 (Extremely unlikely) to 100 (Extremely likely). In particular, participants were asked to rate the degree to which they perceived the scenario to be due to something negative about themselves (internal), something negative about the other person (external), the degree to which the person’s behaviour was intentional or accidental (intentionality), unlikely to change (stability), and the degree to which they believed the behaviour negatively impacts other parts of their relationship (globality).

Emotional distress associated with each scenario was measured by using a subjective units of distress scale asking participants to rate the degree to which they would be angry or upset following the scenario on a sliding scale from 0 (Not upset at all) to 100 (Extremely upset or angry). However, unlike attributions and subjective measures of distress, behavioural responses in grounded theory are often defined by how disproportionate they are to the individual context (e.g. the schema coping style of overcompensation [7]) which meant questions related to specific behaviours would have needed to vary across scenarios making it difficult to differentiate whether changes in condition or changes in behavioural response questions predicted changes in behavioural responses. Due to this variability of behavioural responses across contexts, behavioural responses were measured qualitatively embedded within a predominantly quantitative design to keep questions across scenarios consistent. Behavioural responses were measured by asking participants to record a written response summarising how they would likely respond to the interpersonal situation and why. The behavioural responses were then coded using inductive content analysis to create a series of binary variables scoring endorsement or non-endorsement of each behavioural response so that behaviours could be assessed by statistical analysis with respect to the study hypotheses. To ensure participants followed experiment instructions, participants were asked if they answered vignettes based on how they think they would react rather than what they thought was the best response (Yes/No) at the conclusion of the experiment. Participants were also asked to rate how well they were able to visualise or put themselves in the scenarios on a sliding scale between 0 (Extremely difficult) and 100 (Extremely easy). An example vignette and response layout is presented in S2 Appendix.

Procedure

Participants were first sent a paper-based survey in the mail asking them to respond to some demographic questions and to complete the YSQ-SF-3 and IIP-32 due to the YSQ-SF-3 electronic distribution restrictions. Immediately after completing the paper-based survey, participants were directed to a Qualtrics [48] link to complete the online experiment and RSQ. Data from the paper-based questionnaires and online experiment results were linked using a unique identifier number.

The online experiment was a repeated measures design such that participants completed nine vignettes in total: three vignettes from each of the control, ambiguous and rejection conditions. Participants completed one friend, family and romantic vignette in every condition (i.e., control, ambiguous and rejection conditions). Participants only completed one variation of each vignette to prevent participants from completing the same scenario more than once and keep participants blind to the manipulation of rejection conditions. Presentation of each condition (control, ambiguous and rejection) and type of relationship (friend, family and romantic partner) were randomised in presentation to control for any variation in responses due to order of condition or presentation of scenarios depicting a specific type of relationship. The variation of vignette used in each condition was also randomised to control for any variability in responding across different scenarios. An example of the presentation of conditions and vignettes and allocation to respective presentation order groups is depicted in Fig 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Example summary of experiment depicting the presentation of conditions and vignettes.

https://doi.org/10.1371/journal.pone.0288543.g001

Coding and scoring

Statistical analysis

Given the hierarchical structure of the data, multi-level mixed effect linear and logistic regression models were used to investigate the moderating effect of EMSs (W) on the relationship between degree of rejection (X) and negative responses to interpersonal situations (Y). Multi-level modelling is recommended when assessing a moderator (EMSs) in a repeated measures design with a three-level categorical independent variable (degree of rejection) on a continuous or binary dependent variable (cognitive, emotional and behavioural responses to interpersonal situations; [52]). To assess the moderation of EMSs and rejection condition on each dependent variable we followed the recommendations of Aguinis, Gottfredson and Culpepper [49] but omitted the addition of a random slope due to the small number of observations within participants (n = 9) and non-convergence issues. While the omission of a random slope can inflate the type I error rate, its omission is considered appropriate where there are convergence issues [53]. Given the potential inflation of the type I error rate due to the omission of a random slope and multiple outcome variables the conservative Bonferroni procedure was used to adjust for the type I error rate as recommended by Frane [54]. Consequently, we adjusted our overall alpha level to p = .003 by dividing .05 by all 15 outcome variables and p = .001 for pairwise contrasts across the control, ambiguous and rejection conditions.

The analyses were conducted in Jamovi [55] using the GAMLj module for General Analyses of Linear Models [56] which is a robust module for estimating general linear, mixed and generalised multi-level models [57, 58]. A two-level random effects model using the restricted maximum likelihood (REML) method was estimated for each continuous dependent variable including perceived rejection, negative cognitions, emotional distress and behavioural responses. Independent variables of vignette group, rejection condition, EMS score and the cross-level interaction of EMSs and rejection were added to the model in a stepwise process as recommended by Aguinis, Gottfredson and Culpepper [49].

In step one, the null model with a random intercept was estimated for each dependent variable to determine if multi-level modelling was appropriate [49]. The ICC values for perceived rejection, cognitions and emotional distress were all greater than .1 suggesting it was appropriate to proceed with multi-level modelling. However, all null model ICC for qualitatively coded actions (excluding fighting) were lower than .1 suggesting limited dependency within participants for the qualitatively coded actions. Given our level 1 variance comprised of repeated measures we continued with multi-level modelling as our design implied dependency of scores within participants and failing to control for the dependency of measures even where there are low ICC’s can inflate the type I error rate [59].

In step two, vignette group was added to control for differences in the dependent variables due to differences in the interpersonal scenarios provided in each interpersonal vignette [49]. Vignette group variance included differences in vignettes due to scenarios depicting friends, family and romantic partners. We were unable to control for vignette group in the mixed binary logistic regressions for the qualitatively coded actions due to non-convergence issues associated with having too many categorical variables predicting a single binary outcome variable.

In step three, rejection condition (i.e., control, ambiguous and rejection) was dummy-coded and added to assess if degree of rejection significantly predicted changes in the dependent variables. Simple contrast t-tests were conducted to assess the differences between all three conditions using Bonferroni adjusted p-values.

In step four, total EMS score on the YSQ-SF-3 was added to the model to assess if EMS scores predicted changes in each of the dependent variables [49]. In the final step of model building, the interaction between rejection condition and EMS score was added to assess the cross-level moderation effect of (high and low) levels of EMSs on the relationship between rejection condition and interpersonal responses [49]. For the logistic regression analyses the estimated probability of endorsing each action in each condition was reported using the P symbol. Log-likelihood ratio tests were performed to assess if the addition of each predictor at each step of the model building process significantly explained a unique proportion of model variance. The log-likelihood ratio tests were one-tailed due to the variance only being able to deviate from zero in a positive direction [60].

We computed Cohen’s f² for both the marginal and conditional changes in R² as a measure of effect size for each fixed effect as it was added to the model including vignette group, rejection condition, EMS score and the interaction between rejection condition and EMS scores. The effect size for the change in marginal R² was denoted by f²_m which describes the effect size for the change in the between participant variance (Level 2) explained by the fixed effect [61]. The effect size for the change in conditional R² was denoted by f²_c which describes the effect size for the change in the combined between and within participant variance (Level 1 and Level 2) explained by the fixed effect [61]. The results for the two-level random effects models and effect sizes for each dependent variable of perceived rejection, negative cognitions, emotional distress and behavioural responses are described below and summarised in Tables 1 and 2.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Regression output for the effect of vignette group, rejection condition and EMSs on cognitive and emotional responses to interpersonal scenarios.

https://doi.org/10.1371/journal.pone.0288543.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Regression output for the effect of rejection condition and EMSs on qualitatively coded behavioural responses to interpersonal scenarios.

https://doi.org/10.1371/journal.pone.0288543.t002

Manipulation and assumption checks

Manipulation check measures were reviewed with 96.8% (N = 153) of participants reporting they responded to vignettes based on how they think they would act rather than what they felt was the best course of action. Participants reported the mean ability to visualize themselves in the vignettes to be M = 80.3 (SD = 20.3) on a scale between 0 (Extremely difficult) and 100 (Extremely easy). Significant differences in perceived rejection were found between the different vignette scenarios (F(8, 1252) = 14.56, p < .001, f²_m < .01, f²_c = .01) suggesting that it was appropriate to control for differences in vignettes. Participants reported increased levels of perceived rejection as the degree of rejection in the interpersonal scenarios increased (F(2, 1252) = 436.75, p < .001, f²_m = .12, f²_c = .31). Participants reported higher levels of perceived rejection in the ambiguous scenarios (M = 31.59, SE = 1.36, t(1252) = -18.77, p < .001) and rejection scenarios (M = 46.36, SE = 1.37, t(1252) = -29.16, p < .001) when compared to the control scenarios (M = 4.92, SE = 1.37). Participants also reported higher levels of perceived rejection in the rejection scenarios when compared to the ambiguous scenarios (t(1252) = -10.40, p < .001).

Perceived rejection, self-blame, other-blame, intentionality, stability, globality and emotional distress for each vignette were added together to produce a total score for each participant across the entire experiment. Table 3 shows that perceived rejection in scenarios and interpersonal response variable scores were correlated with similar measures including rejection sensitivity scores (RSQ), trait interpersonal problems (IIP) measures of depression, anxiety and stress (DASS-21).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Correlations (Pearson r) between total interpersonal response scores and trait interpersonal problems, rejection sensitivity and psychiatric symptom distress.

https://doi.org/10.1371/journal.pone.0288543.t003

The assumptions of normality and linearity were satisfied in addition to model variables not showing any signs of multicollinearity. Visual inspection of the scatterplot of residuals suggested signs of double outward box heteroscedasticity which was confirmed by significant log-likelihood ratio test differences when comparing the random-intercept only model of the absolute residuals to the full model of absolute residuals regressed on all of the predictors [62]. Such heteroscedasticity is common in experiments with control conditions and likely reflected reduced variability of responses in the control scenarios when compared to the ambiguous and rejection scenarios [63]. We decided to proceed with interpretation of the multilevel models despite the observed heteroscedasticity to preserve simplicity and interpretability of data given previous simulation studies have found multi-level models to be robust to violations in homoscedasticity when there are equal observations per condition [62, 63].

Qualitative coding of behavioural responses

Eight third order category actions were identified as responses to the interpersonal scenarios using inductive content analysis including (1) engaging, (2) collaborating, (3) supporting, (4) fighting, (5) avoiding, (6) affect expression, (7) uncertainty, and (8) non-engagement. Each of these categories is described below in further detail. Third-order behaviour categories for each scenario were then coded into binary variables as 1 (the behavioural response reported in the scenario) or 0 (the behaviour response not reported in the scenario). The second and third-order categories with specific examples are listed in Table 4. The percentages of participants who reported engaging in each third-order category action across each condition (i.e., control, ambiguous and rejection) are reported in Table 5.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Qualitative categories for actions in response to interpersonal scenarios with examples, number of participants that endorsed the action (N) and their mean Inventory of Interpersonal Problem t-scores (IIP T-Score) and standard deviations (SD).

https://doi.org/10.1371/journal.pone.0288543.t004

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Observed percentages of participants that endorsed each qualitatively coded action across rejection conditions and their mean Inventory of Interpersonal Problem (IIP) t-score (standard deviation).

https://doi.org/10.1371/journal.pone.0288543.t005

Perceived rejection.

Participants with higher EMS scores also reported higher levels of perceived rejection overall (F(1, 156) = 34.34, p < .001, f²_m = .03, f²_c < .01). However, this effect was not statistically significant after adding the cross-level interaction between degree of rejection and EMS scores (ß = 0.03, t(156) = 1.10, p = .272, f²_m = .01, f²_c = .02). EMSs significantly moderated the relationship between degree of scenario rejection and perceived rejection (F(2, 1252) = 18.43, p < .001) such that participants with higher EMS scores reported higher levels of perceived rejection as the degree of scenario rejection increased. Differences in perceived rejection and cognitive responses for people with high, low and mean levels of EMSs across rejection conditions are displayed in Fig 2.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Mean cognitive responses across conditions of interpersonal scenario rejection scored between 0 (Not at all rejected; Extremely unlikely) to 100 (Extremely rejected; Extremely likely).

https://doi.org/10.1371/journal.pone.0288543.g002

Emotional distress.

Participants with higher EMS scores also reported higher levels of emotional distress (F(1, 156) = 24.85, p < .001, f²_m = .03, f²_c < .01). However, this effect became non-significant after adding the cross-level interaction between degree of rejection and EMS scores (ß = 0.02, t(156) = 0.03, p = .258). EMSs significantly moderated the relationship between degree of scenario rejection and emotional distress (F(2, 1252) = 12.64, p < .001, f²_m = .01, f²_c = .01) such that participants with higher EMS scores reported higher levels of emotional distress as the degree of scenario rejection increased. Differences in emotional distress for people with high, low and mean levels of EMSs across rejection conditions are displayed in Fig 3.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Mean emotional distress across conditions of interpersonal scenario rejection scored between 0 (Not at all upset or angry) to 100 (Extremely upset or angry).

https://doi.org/10.1371/journal.pone.0288543.g003

Behavioural responses.

Participant EMS scores did not predict the odds of any qualitatively coded action and did not moderate the relationship between degree of scenario rejection and the odds of any qualitatively coded action (Table 2). Differences in behavioural responses for people with high, low and mean levels of EMSs across rejection conditions are displayed in Fig 4.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Probabilities of engaging in different types of actions by rejection condition and EMS score.

https://doi.org/10.1371/journal.pone.0288543.g004

Interpersonal responses to rejection

To assess participant responses to the interpersonal scenarios, participants were asked to rate the likelihood of negative cognitive attributions, self-reported emotional distress and provide written responses to how participants might respond to the scenarios across conditions. It was predicted that participants would show increased negative cognitions, emotional distress and differences in behaviour as the degree of rejection increased. In support of the present study’s hypotheses, participants were found to report higher levels of emotional distress as the degree of rejection increased. These findings are consistent with previous research reporting increased negative affect following interpersonal rejection [67, 68].

We found that in general participants also reported increased negative cognitions as the degree of rejection increased. In particular, participants reported higher levels of self-blame, other-blame and predicted higher likelihood that the issues in the scenario would globally impact other parts of their relationship as the degree of rejection increased. These findings are consistent with past research showing that people tend to report lower self-esteem and increased other-blame following rejection [69, 70]. Interestingly, participants reported stronger convictions that the person in the scenario was likely to change their behaviour (low stability) in the ambiguous and rejection conditions when compared to the control condition. Participants also perceived behaviour to be more intentional in the control and rejection conditions when compared to the ambiguous condition. However, participants perceived actions in the control condition to have a higher likelihood of being intentional than in the rejection condition. Together, these findings suggest that participants reported lower convictions in perceived intent and stability of rejection in the current experiment. This finding opposes previous experimental research where participants reported increased perceptions of hostile intent following rejection in children [65]. Unlike the experiment conducted by Reijntjes et al. [65] where rejection occurred from peers not known to the participants, in this study rejection was presented as a single transgression within a close interpersonal relationship (i.e., from a family member, friend or romantic partner). Fletcher and Kerr [71] found that individuals were more likely to make positive attribution biases about a person’s actions and transgressions in relationships where there was greater intimacy. Therefore, if participants held relatively strong relationships with their loved ones, it is likely that they might attribute situations involving acceptance to be intentional and stable and unlikely that they would attribute the rejection to be stable or intentional following a single transgression. In contrast, it remains possible that in relationships with limited intimacy (e.g., strangers or peers unknown to the participant) participants might respond more negatively following rejection than how they responded in the present study [65].

Using inductive content analysis of the written responses to the interpersonal scenarios, the present study identified eight overall categories of behavioural responses. One such response included engaging the other person which comprised of actions such as approaching, greeting, making conversation or planning to spend time with the person in the scenario. Unsurprisingly, participants were less likely to engage the person as the degree of rejection increased. Alternatively, participant responses suggested increasing odds of attempting to collaborate about the issue or avoid the person in the scenario in the ambiguous and rejection conditions and increased odds of fighting in the rejection condition. Responses coded as collaborating generally reflected a desire to communicate about the issues presented in the scenarios. In contrast, responses coded in the avoiding or fighting categories consisted of avoiding parts of the interaction, the situation or even the person or engaging in behaviours such as confronting and arguing, making negative comments or retaliating with an intent to hurt the other person. These findings are consistent with previous research that found that increased communication and avoidance following experiences of rejection [67]. These findings are also consistent with research supporting increased aggression following instances of rejection [65].

The present study’s qualitative analysis also identified supporting behavioural responses. Supporting responses generally prioritised the needs of the person in the scenario or involved sacrificing the participant’s interpersonal goals. Participants reported comparably low levels of supporting actions in the control and ambiguous scenarios but significant increases in supporting in the rejection condition. This finding is surprising considering research generally suggests that people tend to show reduced pro-social behaviour following rejection [72]. However, people with low social anxiety have been found to engage in increased pro-social behaviour following rejection while people high in social anxiety did not suggesting some interpersonal characteristics might promote supportive behaviours following rejection [73]. It is also possible that external attributions such as perceptions that your loved one is acting rejecting due to a personal struggle might motivate supporting behaviour after being rejected by someone you typically share a close relationship with. Alternatively, given acceptance was also classified as a supporting strategy it is also possible that acceptance might be more common in situations involving high rejection than in situations where rejection is ambiguous.

The qualitative data also indicated that some participants engaged in behaviours which reflected emotional responses such as expressions of positive and negative affect. Unsurprisingly most expressions of positive affect occurred in the control condition with significantly lower odds of expressing positive affect in both the ambiguous and rejection conditions. In contrast, expressions of negative affect were found to be higher in the ambiguous condition when compared to the control condition. While this finding is consistent with low levels of emotional distress reported in the control condition, participants expressed no differences in negative affect between the control and rejection conditions despite participants reporting the highest amount of subjective distress in the rejection condition. Participants might have reported fewer expressions of negative affect in the qualitative responses because open ended questions did not specifically ask about emotional responses but rather how a participant would act in response to the scenario. Alternatively, increased avoidance in the rejection condition could lead some participants to suppress or hide any expressions of negative affect at higher levels of rejection. Some researchers have argued that avoidance strategies following interaction may at times function to avoid further rejection or damage to the relationship [67]. A minority of behavioural responses were also categorised as uncertainty about how to act or not engaging an interpersonal response. However, changes in uncertainty were not assessed across conditions due to very low base rates and there were no significant differences in responses that did not engage an interpersonal response across any of the experiment’s conditions. Together, our findings combined quantitative and qualitative data to provide researchers with a novel experimental paradigm for assessing the relationship between rejection and emotional, cognitive and behavioural responses to allow for stronger causal attributions in the study of interpersonal behaviour. The categories of engaging, supporting, collaborating, avoiding and fighting also share similarities with the schema coping styles of avoidance, overcompensation and surrender suggested by Young, Klosko and Weishaar [7] and Polyvagal Theory [74] where feelings of safety and security are argued to promote behaviour associated with connecting, while increased stress promotes activation of fight and flight responses.

The moderating effect of EMSs on interpersonal responses to rejection

To assess the moderating effect of EMSs on the relationship between rejection and interpersonal responses. We predicted that individuals with higher EMSs would report increased negative cognitions, distress and problematic behaviours as the degree of rejection increased. Participants reporting higher EMSs were found to report significantly higher levels of perceived rejection, emotional distress and negative attributions (including self-blame, other-blame, globality, stability and intentionality) across all conditions. However, EMSs only moderated the relationship between rejection condition and perceived rejection, emotional distress and self-blame. While these findings provide support for the Schema Therapy model, which argues that individuals with EMSs show increased levels of distress and negative cognitions when EMSs are activated by stressful interpersonal contexts (e.g., rejection), it is surprising that EMSs only moderated the effect of rejection on self-blame and perceived rejection. Together these findings suggest that individuals with higher EMSs tend to report greater distress and negative cognitions regardless of the interpersonal context or situation. Given the finding that interpersonal rejection only activated select negative cognitions in people with higher EMSs, it is possible that EMSs might only moderate the relationship between interpersonal contexts and cognitions unique to that individual’s schema (e.g., only cognitions related to abandonment might be activated in people with abandonment schema). Further research could help clarify these findings by assessing the moderating effect of EMSs on cognitions and scenarios congruent to a specific EMS rather than EMSs in general.

Contrary to predictions, individuals reporting higher EMSs did not show any changes in the odds of engaging in any of the categories of behavioural responses. Changes in the odds of engaging in different behavioural responses were also not observed for EMSs scores across rejection conditions, providing no support for a moderating effect of EMSs on the relationship between rejection and behavioural responses to interpersonal scenarios. These findings suggested that EMSs alone may be less likely to predict interpersonal behaviour. These findings are in contrast to the evidence supporting a link between EMSs and problematic interpersonal behaviour, such as past research that has found individuals with EMSs to report higher levels of aggression [14, 15], interpersonal conflict [17], and trait problem interpersonal behaviour patterns [1, 21]. One explanation for these findings is that most of the studies to date have measured trait dispositional aggression or self-reported perceptions of conflict and interpersonal problems. Given individuals with EMSs by definition hold more negative views of themselves and report higher levels of self-blame, it is possible that use of self-report measures might over-estimate the degree of problematic behaviour in people with higher EMSs [22]. However, adolescents with EMSs have also been assessed as more aggressive from reports from their peers and mothers in two separate studies [75, 76]. While limited, previous research suggests that it is likely that individuals with EMSs engage in more unhelpful interpersonal behaviour. Consequently, it is possible that factors outside of the scope of the present study might influence whether a person with EMSs engages in unhelpful interpersonal behaviour. For example, researchers have found hostility attributions to mediate the relationship between peer rejection and aggressive behaviour [65].

Theoretical and clinical implications

The finding that EMSs predicted and moderated the impact of interpersonal rejection on cognitive and emotional interpersonal responses suggests that there is utility in addressing EMSs in therapy to reduce convictions in negative cognitions and emotional distress in interpersonal relationships. These findings are supported by previous research finding reductions in emotional distress and negative cognitions in select disorders such as major depressive disorder and borderline personality disorder following Schema Therapy and following a couples based schema therapy intervention [77–79]. However, in contrast to the position of the schema therapy model that argues negative cognitions and emotional distress are likely to be activated in stressful situations or when the context is congruent with a schema [7], we found people with higher EMSs were likely to show higher negative cognitions and emotional distress overall regardless of the interpersonal situation.

The findings from the present study also suggest that interpersonal situation or degree of rejection was more likely to predict interpersonal behaviour than EMSs given EMSs were not found to predict behaviour responses. These findings suggest that interpersonal behaviour might be more challenging to change without considering broader interpersonal contextual factors. Such contextual factors might include dyadic interactions that are best addressed using interventions such as couples therapy [80]. For example, Fals-Stewart and Clinton-Sherrod [80] found couples therapy to be superior to individual therapy in reducing intimate partner violence in a sample of offending males. Interventions might also need to consider increasing awareness of interpersonal contexts such as learning to identify when rejection is ambiguous as participants were more likely to collaborate to resolve perceived issues in the ambiguous rejection scenarios while reporting increased likelihood of fighting and supporting the other person when rejection was perceived as less ambiguous.

Limitations and considerations for future research

Together the findings of the present study provide support for the theory that people with higher EMSs experience higher levels of emotional distress and negative cognitions when presented with higher degrees of interpersonal rejection and that EMSs moderate the relationship between the emotional and cognitive responses to interpersonal rejection. However, there are some limitations to consider with respect to the findings of the present study. The limited differences observed in behaviours across people with EMSs might in part reflect the reduced variability of actions due to having action categories as binary outcome variables restricted to select single incident rejection scenarios. In particular, actions were split into different categories and each individual’s action was coded as if the response reflected the category or not. It is possible that people with EMSs might simply express behavioural responses at higher levels of intensity or at greater frequencies; characteristics unlikely to be detected using binary coding of presence of type of behaviour (or not) or single scenario experiences of rejection.

This study also assessed EMSs as an overall EMS score not allowing for different EMSs to be assessed individually. It is likely that changes in responses across interpersonal situations depicting rejection might be more apparent in EMSs more specific to rejection and not others [22]. For example, the disconnection and rejection EMS domain subscale which was not used in the present study due to poor factor structure might show stronger associations between interpersonal rejection and responses. In addition, it is possible that specific EMSs or other individual differences may lend themselves more specifically to certain styles of coping or behavioural actions. For example, researchers have predicted and found the EMS of insufficient self-control and discipline to be more strongly associated with aggressive behaviour [39].

This study also embedded responses to different types of relationships within its design by including an equal number of scenarios specific to friends, family and romantic partners across conditions. While this was intended to control for differences across close relationships, we were unable to assess for differences across types of relationships due to the variability in responses between individual scenarios. It should also be noted that while we were able to control for scenario variation in our statistical analyses for cognitive and emotional responses, the number of categorical variables when analysing our binary behavioural response variables created non-convergence issues when fitting our logistic regressions, preventing us from controlling for scenario variation when assessing the impact of condition and EMSs on behavioural responses. Given that our study has identified types of behavioural responses to the scenarios used, future research might benefit from using such results or other established models of interpersonal behaviour to guide the assessment of behavioural responses. This could include quantitative measures such as likelihood assessments of different responses, frequency count measures or measures of significant others’ cognitions, emotions and responses.