Emotion regulation is the process by which emotions are identified, monitored, managed, and modified (Thompson 1994). As a construct, emotion regulation is complex, incorporating both subjective experience and observable behaviors (Hourigan et al. 2011). Emotion regulation has been conceptualized as a general individual trait that is stable across situations (Eldesouky et al. 2017) and as a context-specific response tendency that is socially acquired and mutable (John and Gross 2004). Emotion regulation abilities are vital across the life-span for dealing with various emotional states, although normative age changes and gender differences in strategy usage have been noted (Zimmermann and Iwanski 2014). Effectively regulating one’s emotions contributes significantly to emotional experiences, social relationships, sense of well-being (Gross and John 2003), and even physical health outcomes such as cardiovascular risk (Trudel-Fitzgerald et al. 2017). In 2013 alone, over ten thousand studies were published on emotion regulation (Gross 2015), and more will certainly come. In light of the popularity of assessing emotion regulation in recent literature, researchers must increase our understanding of the nature of this construct.

Because emotion regulation is thought to include some observable behaviors that are noticeable to other people, researchers studying this construct could benefit from adopting a multi-informant approach. The typical multi-informant approach involves asking multiple people, or informants, about a target individual; generally, informants are relatives, peers, or teachers who know the target well (Kaurin et al. 2016). When couples are involved, a multi-informant approach could involve individuals rating themselves and their partners on a given variable, resulting in four parallel sets of data when both members of the couple dyad serve as respondents (Busby and Gardner 2008).

Although using parent informants to collect data on children and adolescents is considered the “gold standard” (Renk 2005), using partner informants is far less common when assessing adults, with the exception of adults whose self-reports are deemed questionable due to psychosis, cognitive limitations, substance use/abuse, forensic or legal issues, or serious mental illness (Achenbach et al. 2005). Self and partner ratings are most often gathered in studies of intimate couples to examine concordance or divergence in partners’ behaviors or opinions, such as for alcohol and tobacco use (Machado et al. 2017), physical health status (Wolinsky et al. 2016), or sexual dysfunction and dissatisfaction (Gungor et al. 2015). Personality traits have also been widely evaluated using multiple informant reports (Costa et al. 2018). However, relying exclusively on adults’ self-reports remains the dominant method of assessment in personality research and clinical assessment (van der Ende et al. 2012; Vazire 2006). Additionally, it remains unclear whether constructs that involve internal experiences, such as emotion regulation, can be adequately assessed via observer ratings.

Yet, researchers and clinicians alike recognize the limitations inherent to asking individuals to disclose about themselves. In the case of youth, many assume that children’s self-reports are compromised because they may be developmentally unable or unwilling to provide accurate and meaningful data about themselves (Brener et al. 2003; Measelle et al. 2005). Similarly, adults are subject to self-report biases that may result from selective recall, social desirability, or instrument choice (Möricke et al. 2016), or simply from limited cognitive capacities for attention and short-term memory (Avolio et al. 1991). Using an informant, such as a romantic partner, overcomes this single-source bias. Informants can also capture behaviors that occur in different situations or settings (Dirks et al. 2012), which increases predictive ability (De Los Reyes et al. 2015) and clinical utility (Dirks et al. 2012). Lastly, using multiple informants is advantageous at a practical level, as it is more time-efficient, user-friendly, and cost-effective than other methods such as structured interviews or direct behavioral observations (Möricke et al. 2016). Given the many factors that could bias individuals’ self-reports, and the potential for multi-informant approaches to broaden the available data on an individual, it is worth contemplating why multiple informants are not used more often when assessing adults.

One obstacle to using a multi-informant approach to study emotion regulation among adults is the strong likelihood of disagreement between self and informant reports, which is common (Achenbach 2006) and complicates theoretical and statistical interpretation of their reports (Kraemer et al. 2003). One must decide whose report will be relied upon in the event of disagreements, and in which situations (Rodriguez et al. 2016). Individuals’ reports may differ in their predictive power (Loeber et al. 1990), complicating decision-making in clinical practice settings (De Los Reyes et al. 2013). Although clinical decisions for children rely more on parent-provided information than children’s self-report, “best practices” have yet to be established for adults (De Los Reyes et al. 2015). Nonetheless, differences between adult reports are still meaningful in that they convey differences in informants’ perspectives. Thus, in spite of potential complexities, the wealth of material gained from using multiple reporters makes this methodological approach worth exploring.

To address this complexity of interpretation, researchers need to consider and test why self and partner reports about constructs such as emotion regulation might diverge (Costa et al. 2018). Low informant agreement may result in part from the differing abilities of individuals to perceive a target behavior (Loeber et al. 1990). In the case of emotion regulation, one’s thoughts or internal emotional states are not visible to others; even some mood regulation behaviors may not be readily observed by others (Hourigan et al. 2011). Furthermore, sociocultural factors such as traditional gender norms may also influence how women and men perceive and report about behaviors and attitudes, such as sexual activity and satisfaction (Gungor et al. 2015); this may also apply to the gender-stereotyped domain of emotional expression and regulation (Zimmermann and Iwanski 2014). Additionally, studies that have examined intimate partners find that couples exhibit signs of bias, wherein one member assumes similarity to their partner, particularly for constructs that pertain to their relationship (Kenny and Acitelli 2001). Thus, couples may be motivated to perceive their partners in a certain light due to the intimate nature of their relationship.

Differences in opinion also appear to relate to individuals’ unique perspectives and differences in their perceptions and goals (De Los Reyes and Kazdin 2005; Dirks et al. 2012), as well as informants’ own traits (De Los Reyes et al. 2008). Characteristics such as perceived stress and psychopathology may be particularly impactful, as depression is thought to distort the depressed individual’s perceptions, as proposed by the depression-distortion hypothesis (De Los Reyes and Kazdin 2005; Richters and Pellegrini 1989). Alternatively, depressed individuals are hypothesized to reflect upon themselves more accurately than non-depressed individuals, who are thought to exhibit a self-affirmation bias (Moore et al. 2016)—a phenomenon termed the depressive realism hypothesis. Perceptual differences due to the presence or absence of depressive symptoms may therefore widen the gap between self- and informant-reports about one’s own or one’s partner’s abilities to regulate negative emotions.

In addition to intrapersonal traits, qualities of the relationship between informants and the target individual may affect informants’ reports, which is particularly relevant in intimate partner relationships (cf. Keeny and Acitelli 2001). For instance, relationship qualities such as satisfaction may shape informants’ perceptions of and reports about the individual due to leniency bias or halo effects, in which one’s ratings about an individual’s specific attributes are based on one’s general judgment of that individual (Hoyt 2000). Romantically-involved couples in particular may be susceptible to sentiment override, the global affection or disaffection for one’s partner or relationship that creates a perceptual filter (Weiss 1980). Whereas high relationship satisfaction may contribute to a positive sentiment that biases informants to rate their partner more positively on specific dimensions, a detrimental relationship factor such as experiencing domestic violence victimization might negatively affect informants’ ratings of their partner. Thus, global impressions based on broad relationship factors may impair accurate ratings of one’s partner on specific dimensions. Given that women, but not men, appear to be influenced by their global marital sentiments when evaluating partners’ affective expressions (Hawkins et al. 2002), gender differences in sentiment override should be examined. By examining these qualities in relation to informant reports, researchers can learn about intrapersonal and interpersonal factors that may affect individuals’ ratings of themselves and their partners, which may ultimately help professionals decide whose report to rely on, and when, in research and clinical settings.

The present study aimed to investigate factors that contribute to differences between self-reported and partner-reported emotion regulation among adult heterosexual couples. Each member of the couple reported on their own and their partner’s emotion regulation abilities. Each partner’s personal characteristics, as well as qualities about the couple relationship, were hypothesized to contribute to differences in adult couples’ opinions of their emotion regulation abilities. Specifically, the current study examined the influence of perceived stress, endorsement of symptoms of psychopathology, relationship satisfaction, and intimate partner victimization on individuals’ reports of their own and their partner’s emotion regulation abilities. A number of hypotheses were proposed. First, with regard to intrapersonal qualities, higher perceived personal stress and more symptoms of psychopathology were expected to be associated with greater discordance between self-reported and partner-reported emotion regulation abilities. Second, with regard to couple-related characteristics, higher intimate partner victimization and lower relationship satisfaction were expected to be associated with greater discordance between self-reported and partner-reported emotion regulation abilities. Third, concerning the direction of discordance, self-reports of higher perceived personal stress, greater symptoms of psychopathology, higher intimate partner victimization, and lower relationship satisfaction were expected to be associated with lower self-reports of emotion regulation abilities relative to partners’ reports of those abilities. Fourth, given the strong gendered norms in emotional expression and emotion regulation strategies (Zimmermann and Iwanski 2014), gender differences in these associations were explored.

Method

Participants

The sample included 81 male-female dyads recruited for a parenting study of couples raising preschoolers in the Southeast, U.S. Women’s mean age was 33.85 years (SD = 5.20) and men’s mean age was 35.99 years (SD = 7.35). Participants self-identified as primarily Caucasian (women, 76.5%; men, 80.2%), followed by African-American (women, 19.8%; men 18.5%) or Asian (1.2% of both women and men); in addition, some respondents also identified as Hispanic/Latino (women, 6.2%; men, 1.2%). Couples had been in a relationship for 10.40 years (SD = 4.65) on average. For both parents, median educational level was a 4-year college degree. Parents were raising an average of two children on a median annual family income of $65,000. A sensitivity power analysis using APIM Power (Ackerman and Kenny 2016) conducted after data were collected revealed that the final sample size should have provided adequate power (>.80) to detect small-to-medium-sized effects (r = .21).

Procedure

Couples were recruited for a larger parenting study, the “Couples Parenting Preschoolers Study,” with flyers distributed at relevant locations in the local community, including day care centers, and through newspaper advertisements. To be eligible to participate in the larger study, couples had to be married and/or cohabitating parents of a 3–6-year-old child. Couples interested in participating phoned the lab to schedule a 90-min in-home session. All measures were presented electronically on individual laptop computers and each member of the couple completed their assessments in separate, private rooms. Participants’ individual item responses were automatically entered into a database with a randomly assigned family identification number to assure parents anonymity in their responding. Each member reported on their own emotion regulation abilities first, followed by four unrelated measures, before reporting on their partner’s emotion regulation abilities. Each member of the couple received $30 for study participation. All procedures in the larger study were approved by the university institutional review board.

Measures

Emotion regulation

The Negative Mood Regulation Scale (NMRS; Catanzaro and Mearns 1990) consists of 30 items measuring how participants believe they manage negative emotions. The NMRS presents 30 items in which respondents indicate how well they believe they regulate their distress using mood regulation strategies. Items are scored on a 5-point Likert scale, from (1) strongly disagree to (5) strongly agree. Items are then summed and oriented so that higher total scores suggest better regulation of negative emotions. The NMRS demonstrates internal consistency, stability, and concurrent and predictive validity with negative affect (Catanzaro and Mearns 1990) and evidences convergent validity with other emotion regulation measures (Bardeen et al. 2016). In the current study, good internal consistency was observed for both members of the couple (women’s α = .89; men’s α = .88).

A second informant version of the NMRS (Rodriguez et al. 2016), which presents the NMRS with modified language in the items and instructions, was also administered to each member of the couple. Before each of the 30 items, each participant was instructed to report on how well they believe their partner can regulate their distress using mood regulation strategies; thus, instructions and all items in this version parallel those on the self-report version using the same 5-point scale. Higher scores on this Partner NMRS suggest perception of better emotion regulation ability in their partner. A similar adaptation for parents to report on their emerging adult child’s emotion regulation ability (Rodriguez et al. 2016) also demonstrated internal consistency for the emerging adults (α = .90) and their mothers (α = .88) and fathers (α = .89). The current study also observed strong reliability for this Partner NMRS version for both women (α = .94) and men (α = .92).

Intrapersonal functioning

The Perceived Stress Scale (PSS; Cohen et al. 1983) was administered to assess participants’ sense that their lives have been overwhelming, uncontrollable, or unpredictable in the past month. Ten items are presented using a 5-point Likert scale from (1) never to (5) very often. Total scores are averaged across items, with higher scores indicating greater perceived stress. The current sample demonstrated acceptable internal reliability for both women (α = .84) and men (α = .83).

The Revised Symptom Checklist-90 (SCL-90-R; Derogatis 1994) is a measure of a range of mental health symptoms: somatization, obsessive-compulsive, interpersonal sensitivity, depression, anxiety, hostility, phobic anxiety, paranoid ideation, and psychoticism. Respondents indicated on a 5-point Likert scale how frequently they have been bothered by 90 symptoms in the past week, from (0) not at all to (4) extremely. General symptom distress is the total sum across symptoms, with higher scores indicative of greater distress. The SCL-90-R demonstrated strong internal consistency in this study for both women (α = .93) and men (α = .92).

Relationship functioning

The Couple Satisfaction Index (CSI; Funk and Rogge 2007) presents items measuring overall relationship satisfaction; in the current study, 16 items were rated on a 6-point scale, with higher scores indicative of greater satisfaction. The CSI can discriminate between distressed and non-distressed relationships. Scores on the CSI are related to alternative measures of dyadic adjustment and marital satisfaction (Funk and Rogge 2007). Internal consistency was high in the current sample for both women (α = .98) and men (α = .97).

The Revised Conflict Tactics Scale Short Form (CTS-2S; Straus and Douglas 2004) is an abbreviated version of the Revised Conflict Tactics Scales (Straus et al. 1996), a frequently used measure of intimate partner violence. The CTS-2S provides a weighted frequency count of how the couple addresses a wide range of conflict strategies including negotiation, psychological aggression, physical assault, sexual coercion, and injury. Half of the items involve perpetration with parallel items addressing victimization. For the present study, the eight items involving victimization with either psychological or physical aggression were selected, with higher scores indicative of greater frequency of psychological and/or physical victimization. The test authors provide evidence of concurrent validity as well as concordance with the longer version.

Data Analyses

To address the primary research question regarding the factors that predict differences between self-reported and partner-reported emotion regulation, the following two-level model that nested individuals within couples using HLM 7.01 was estimated following the recommendations of Kenny et al. (2006):

$$\begin{array}{l}Y_{ij}\left( {{\mathrm{Emotion}}\,{\mathrm{Regulation}}\,{\mathrm{Discrepancy}}} \right){\mathrm{ = }}\\ \beta _{0ij} + \beta _{1ij}\left( {{\mathrm{Perceived}}\,{\mathrm{Stress}}} \right) + \beta _{2ij}\left( {{\mathrm{Mental}}\,{\mathrm{Health}}\,{\mathrm{Symptoms}}} \right)\\ \quad +\, \beta _{3ij}\left( {{\mathrm{Victimization}}} \right) +\, \beta _{4ij}\left( {{\mathrm{Relationship}}\,{\mathrm{Satisfaction}}} \right)\\ \quad +\, \beta _{5ij}\left( {{\mathrm{Partners}}\prime \,{\mathrm{Perceived}}\,{\mathrm{Stress}}} \right)\\ \quad +\, \beta _{6ij}\left( {{\mathrm{Partners}}\prime \,{\mathrm{Mental}}\,{\mathrm{Health}}\,{\mathrm{Symptoms}}} \right)\\ \quad \, +\, \beta _{7ij}\left( {{\mathrm{Partners}}\prime \,{\mathrm{Victimization}}} \right)\\ \quad \, +\, \beta _{8ij}\left( {{\mathrm{Partners}}\prime \,{\mathrm{Relationship}}\,{\mathrm{Satisfaction}}} \right)\,+\,r_{ij}\end{array}$$
(1)

In this model, emotion regulation discrepancy scores were simultaneously regressed onto participants’ self-reported perceived stress scores, mental health symptoms scores, victimization scores, and relationship satisfaction scores, as well as their partner’s self-reported perceived stress scores, mental health symptoms scores, victimization scores, and relationship satisfaction scores in the first level of the model. The non-independence of couples’ data was controlled in the second level of the model with a randomly varying intercept.

Given that traditional difference scores—in which in which each person’s score is subtracted from their partner’s score—have long been associated with measurement error and can lead to erroneous conclusions (Cronbach and Furby 1970; Griffin et al. 1999), the current study did not adopt that approach. Instead, emotion regulation discrepancy scores were created from a separate analysis that regressed participants’ self-reported emotion regulation ability scores onto partners’ reports of participants’ emotion regulation ability scores. The unstandardized residuals from this analysis served as the index of the discrepancy between self-reported and partner-reported emotion regulation, and this approach is not subject to the same erroneous conclusions that plague traditional difference score approaches (see Castro-Schilo and Grimm 2018; Laird and De Los Reyes 2013; Laird and Weems 2011). Such discrepancy scores may be interpreted in a manner similar to traditional difference scores, however. Specifically, higher discrepancy values indicate that participants’ reports of their own emotion regulation abilities were greater than their partners’ reports of participants’ abilities; lower values indicate that partners’ reports of participants’ emotion regulation abilities were greater than participants’ own reports of their abilities. Significant positive associations would indicate that a given predictor is associated with greater self-reported emotion regulation abilities, whereas significant negative associations would indicate that the predictor is associated with greater partner-reported emotion regulation abilities.

Results

Preliminary Analyses

Descriptive statistics and correlations among all variables appear in Table 1. Mean discrepancies between participants’ self-reports and partners’ reports of their emotion regulation abilities were 10.99 for women (SD = 19.26) and 6.86 for men (SD = 19.04). No mean gender differences across any measures were observed (all p > .05). Note that both women and men viewed their own negative emotion regulation abilities as stronger than how their emotion regulation abilities were perceived by their partners (women, t(80) = 3.76, p < .001; men, t(80) = 4.73, p < .001). Also note in Table 1, men’s self-reported better emotion regulation was associated with lower perceived stress, fewer symptoms of psychopathology, less intimate partner victimization, and greater relationship satisfaction; women’s self-reported better emotion regulation abilities were also associated with lower perceived stress and psychopathology but were not significantly related to their IPV victimization or couple satisfaction. Also shown in Table 1, women’s self-reported emotion regulation abilities about themselves were only marginally related to their reports of their partner’s emotion regulation abilities (r = .21), but men’s self-reported emotion regulation about themselves were significantly related to their reports of their partner’s emotion regulation abilities (r = .38). Not depicted in the table, women’s reports of their partner’s emotion regulation abilities were significantly related to men’s self-reported emotion regulation (r = .37, p < .001), reflecting cross-informant concordance, but men’s reports of their partner’s emotion regulation abilities were only marginally related to women’s self-reported emotion regulation (r = .21, p = .057).

Table 1 Means, standard deviations, and correlations among measures
Full size table

Predictor Analyses

Results from the primary analyses are presented in Table 2. As indicated, greater discordance between participants’ self-reports and their partners’ reports of their emotion regulation abilities was significantly associated with participants’ own perceived personal stress and marginally related to their own mental health symptoms, in support of the first hypothesis. An additional supplemental analysis indicated that participants’ personal stress remained significant, B = −11.02, SE = 2.17, t(69) = −5.08, p < .001, r = .52, even when controlling for participants’ age and minority status and couples’ relationship duration and combined household income. However, participants’ personal mental health symptoms were no longer marginally significant when those additional variables were included, B = −0.10, SE = 0.06, t(69) = −1.56, p = .123, r = .18. With regard to the second hypothesis, couple-related variables were not significantly associated with discordance between self-reported and partner-reported emotion regulation abilities. With regard to the direction of effects proposed in the third hypothesis, when participants reported higher perceived personal stress, they rated their own emotion regulation abilities as significantly lower in comparison to their partners’ ratings of their abilities. Additionally, when participants reported more mental health symptoms, they rated their own emotion regulation abilities as marginally lower in comparison to their partners’ ratings of their abilities.

Table 2 Effects of perceived stress, mental health symptoms, partner victimization, and relationship satisfaction on the difference between personal and partner-reported emotion regulation
Full size table

Gender Analyses

Finally, for the fourth research question, further supplemental analyses indicated that gender did not moderate the association between either stress, B = −1.82, SE = 4.09, t(66) = −0.45, p = .657, r = .05, or mental health symptoms, B = 0.09, SE = 0.11, t(66) = 0.83, p = .411, r = .10, and the difference between self- and partner-reports of emotion regulation abilities. Similarly, gender did not moderate any of the null effects with one notable exception—relationship satisfaction, B = −0.41, SE = 0.16, t(66) = −2.68, p = .009, r = .31. Specifically, men who reported being more satisfied with their relationship also rated their own emotion regulation abilities as better, in comparison to their partners’ reports of their abilities, B = 0.33, SE = 0.13, t(66) = 2.52, p = .014, r = .30. This association was not observed among women, B = −0.09, SE = 0.10, t(66) = −0.89, p = .377, r = .11.

Discussion

In order to consider the value of utilizing couples’ assessment of their partner, the current investigation evaluated factors that may influence adult couples’ reports of each other’s emotion regulation abilities—namely, whether each partner’s personal qualities, as well as characteristics about their relationship, predict discordance between informants. Results partially supported the hypotheses, demonstrating that disagreement between partners’ reports was largely attributable to personal characteristics, such as self-reported stress, and to a lesser extent, endorsement of mental health symptoms. Although gender differences were observed for couple satisfaction, disagreement between partners’ reports did not appear to be attributable to other qualities pertaining to the informants’ relationship, such as intimate partner victimization.

Participants’ higher self-reported stress was associated with lower self-reported emotion regulation abilities in comparison to their abilities as rated by their partner, for both women and men. A similar pattern was suggested by the trends observed for personal reports of greater psychopathology, which was considered simultaneously with perceived stress, thereby reducing that effect. These findings suggest that experiencing stress or distress may skew one’s self-perceptions negatively, consistent with the depression-distortion hypothesis (De Los Reyes and Kazdin 2005; Richters and Pellegrini 1989). Alternatively, these results may indicate a generalized response bias, reflecting a tendency to self-report more negative experiences, rather than actual negative self-perceptions.

In addition, results indicated that men, but not women, who expressed higher couple satisfaction believed that their own emotion regulation abilities were better in comparison to how their emotion regulation abilities were perceived by their partners. For men only, being satisfied with their couple relationship was associated with more positive self-perceptions of emotion regulation. These results are consistent with prior research suggesting that men may have a somewhat stronger positive self-serving bias than women (Mezulis et al. 2004) and that men’s, but not women’s, stronger emotion regulation abilities can buffer their perceived stress to improve their subjective well-being (Extremera and Rey 2015). Further, high relationship satisfaction has previously been associated with low emotional reactivity among men but not women (Peleg 2008). Thus, men reporting higher satisfaction with their relationship may perceive their own emotion regulation abilities to be better due to their generalized positive internal emotional experience. Perhaps future research studies that include men’s relationship satisfaction should include an evaluation of their emotion regulation abilities.

This study contributes to the literature in considering whether adults can serve as informants regarding their partners’ abilities to regulate their negative emotions. Results indicated that women’s ratings of their partners corresponded more highly with their partners’ self-reports than did men’s ratings of their partners. If higher correspondence between partners reflects greater accuracy on the part of the informant, perhaps women are more experienced in perceiving others’ internal states or emotion regulation strategies due to gendered expectations about emotional attunement or gender differences in emotion socialization (Zimmermann and Iwanski 2014). Alternatively, men’s emotion regulation behaviors might be more observable to their partners than women’s choice of strategies, allowing women to report about their partners more readily. Past research indicates that women reported engaging in more social support seeking and dysfunctional rumination whereas men engaged more in passivity, avoidance, and suppression (Zimmermann and Iwanski 2014). Replication is needed to elucidate these gender differences. Future studies could investigate why women’s ratings of their partners are more similar to their partners’ self-reports than are men’s ratings of their partners, and whether this gender difference also exists when couples report on other intrapersonal qualities. This could help inform future researchers on whether women might be more accurate informants about their partners in comparison to men.

Limitations

Several limitations are worth noting. First, this study considered negative mood regulation broadly and did not consider regulation of positive emotions. This focus on negative affect states may have influenced participants’ reporting tendencies. In addition, one limitation inherent to the use of informants is that this data collection strategy still uses only a single method (Holmbeck et al. 2002). Informants face many of the same reporting issues (i.e., biases related to attention, memory, social desirability, etc.) that are present when one person provides a self-report (Avolio et al. 1991; Möricke et al. 2016). This problem is especially true for the subjective, often retrospective data gathered from questionnaires. Additionally, consensus does not necessarily denote accuracy (McCrae and Costa 2013); two raters might agree about a particular behavior if both endorse a faulty stereotype, such as perceived gender differences (Löckenhoff et al. 2014). Ideally, future studies should consider using multiple informants to evaluate bias (cf. Stern and West 2018) in conjunction with employing a variety of methodologies for assessing emotion regulation. Using more objective measures, such as direct behavioral observations, experience sampling methods, or analog tasks, would be advantageous. Nonetheless, assessing individuals’ thoughts, feelings, and sense of well-being is still important to understand subjective experiences and clarify what contributes to differences of opinion.

Other limitations relate to the current study’s sample characteristics. Participants were largely well-educated with moderate income levels on average, with lower socioeconomic status backgrounds somewhat less well-represented. Future studies could draw samples from different socioeconomic status and cultural backgrounds, which could shed light on the relative effect of intrapersonal versus relational influences among various groups. Additionally, the sample did not include same-sex couples, a population which could be examined in future research to further untangle the role of partner gender. Likewise, the current sample was recruited from the community; current findings may not generalize to couples experiencing challenges with clinical levels of psychopathology or documented domestic violence. Replication with clinical populations could prove interesting to determine the degree to which stress and mental health symptoms impact self-reported emotion regulation abilities among individuals with documented psychopathology, who might display more observable emotion regulation strategies. Finally, other intrapersonal or interpersonal characteristics that were not included in the current study might also influence concordance between adult informants. Future research might consider analyzing factors such as self-esteem or relationship communication. Similarly, conducting longitudinal studies could provide insight into changes in informant concordance over time, perhaps as predicted by changes in relationship qualities or intrapersonal factors.

Overall, the aim of the study was to identify intrapersonal and/or interpersonal qualities that contribute to differences between adults’ self-reported and partner-reported emotion regulation in an effort to consider whether partners can provide an alternative perspective for research or clinical purposes. Perceived stress appears to be most critical in explaining discordance in couples’ reports. To some, this finding might suggest that stress itself distorts self-perceptions regarding emotion regulation abilities, or that adults’ perceptions of stress are conflated with their poor ability to cope with distress, such that adults may be unable to provide accurate data about themselves when distressed. However, others may interpret this to mean that partners cannot be accurate reporters on others’ emotion regulation abilities due to the subjective nature of less observable emotion regulation strategies—although this latter concern does not hamper the assessment of children’s emotion regulation abilities. Needless to say, no one has direct access to another individual’s subjectively experienced emotions or internal sensations. Yet observers’ perceptions and perspectives are still valuable by providing alternative information that individuals perhaps cannot or will not offer about themselves. Using multiple informants to assess adult emotion regulation could prove more effective if measures of emotion regulation purposefully rely more heavily on concrete, observable behaviors and less on internal cognitions or subjective feelings that are difficult or even impossible for others to detect. In this manner, researchers and clinicians will be able to formulate a more comprehensive picture of adults’ adaptive emotion regulation abilities.

In addition, emotion regulation has clear implications for personal coping and interpersonal relationships. Using certain coping strategies to regulate negative emotions may open or close communication with others (Folkman and Lazarus 1988), meaning that emotion regulation abilities may not simply affect one’s own well-being, but also the quality of one’s intimate relationships. Additionally, agreement or disagreement between couples’ perceptions of each other’s emotion regulation abilities could possibly serve as an indicator of the couple’s ability to communicate effectively. Thus, improving adults’ understanding about their own emotion regulation abilities, as well as those of their partners, could be useful in enriching the couple relationship in a therapeutic context.