Biased and Accurate

Abstract

Are depressive symptoms associated with more biased or more accurate interpersonal perceptions? Both members of committed heterosexual couples (N = 78) reported on their perceptions of their partner’s commitment and behavior daily across a 3-week period. Using the partner’s reports as the benchmark, participants who reported more depressive symptoms not only underestimated their partner’s commitment and overestimated their partner’s negative behavior (greater directional bias) but were also more accurate in tracking changes in their partner’s commitment and behavior across days (greater tracking accuracy). More negative perceptions of the partner’s commitment and behavior was also associated with increases in relationship insecurity and depressed mood, particularly when the partner also reported lower commitment and more negative behavior. These results indicate that depressive symptoms are associated with both more accurate and more biased interpersonal perceptions and suggest that more accurate detection and more biased magnification of interpersonal threat has important implications for the maintenance of depressed mood.

Keywords

depression depressive symptoms bias accuracy relationship perceptions

A longstanding debate involves whether depression produces negatively biased thinking (Beck, 1967) or more realistic and accurate judgments (Alloy & Abramson, 1979). Research supports both positions. Focusing on interpersonal judgments, for example, depressed people are less accurate in recognizing emotion in facial expressions and overperceive sadness and anger (Gilboa-Schechtman, Foa, Vaknin, Marom, & Hermesh, 2008; Leppänen, 2006). However, depressive symptoms are associated with greater accuracy in analytically complex tasks that require integrating social information, such as detecting motivated deception (Lane & DePaulo, 1999) and avoiding the fundamental attribution error (Yost & Weary, 1996).

Prior attempts to reconcile these inconsistencies have focused on the different tasks or contexts that might explain when depression is associated with bias versus accuracy (e.g., Andrews & Thomson, 2009). In contrast, we draw upon recent distinctions between two forms of accuracy—directional bias and tracking accuracy (Fletcher & Kerr, 2010; Gagné & Lydon, 2004; West & Kenny, 2011)—to test whether depressive symptoms are associated with both bias and accuracy. Directional bias reflects the degree to which judgments over- or undershoot some benchmark, such as perceiving expressions as more negative than objective ratings of expression tone. Tracking accuracy, in contrast, indexes the degree to which perceivers accurately track changes in the corresponding benchmark, such as detecting when expressions become more or less negative during an interaction.

The present research examined whether depressive symptoms were associated with directional bias and tracking accuracy by assessing daily perceptions of intimate partners’ commitment and behavior across a 3-week period. Using the partners’ reported commitment and behavior as benchmarks, we predicted that greater depressive symptoms would be associated not only with (a) greater directional bias, including underestimating the partner’s commitment and overestimating the partner’s negative behavior, but also (b) greater tracking accuracy, including more accurately detecting changes in the partner’s commitment and behavior across days.¹ We also explored the consequences of directional bias and tracking accuracy for relationship security and depressed mood.

Bias and Accuracy Within Intimate Relationships

Within intimate contexts, personal well-being and outcomes depend on the actions and continued investment of the partner (Kelley & Thibaut, 1978). Thus, individuals are motivated to assess their partner’s commitment, and behavior diagnostic of their partner’s evaluations, because of the crucial outcomes associated with these judgments, including acceptance and belonging versus rejection and dissatisfaction (Overall, Fletcher, & Kenny, 2012). The resulting perceptions have important consequences. When people perceive their partner is committed and responsive, they feel satisfied and secure in their relationship and constructively cope with relationship difficulties. When people perceive their partner as less committed and intentionally hurtful, they experience chronic insecurities, lower satisfaction, and react defensively to expected rejection (Murray & Holmes, 2009; Reis, Clark, & Holmes, 2004).

There are individual differences in the extent to which people are biased and accurate when assessing their partner’s sentiments and behavior. Dispositions that encompass chronic relationship insecurity, like lower self-esteem and higher attachment anxiety, are associated with perceiving less regard and support than the actual regard and support reported or shown by the partner (e.g., Collins & Feeney, 2004; Murray, Holmes, & Griffin, 2000). Intriguingly, as with the inconsistent findings regarding depression, relationship insecurity is also associated with more accurate perceptions of the partner’s thoughts and feelings (when compared with the partner’s reported thoughts and feelings) during relationship-threatening conflict discussions (Simpson et al., 2011).

Overall et al. (2012) reconciled this contradictory pattern by distinguishing between directional bias and tracking accuracy. We use their study to illustrate how directional bias and tracking accuracy is conceptualized and measured. Overall et al. asked perceivers to judge how positively their intimate partners regarded them at 14 time points during a conflict discussion. To assess directional bias and tracking accuracy, ratings of the actual regard reported by the partner were also gathered at each time point. Using recent statistical procedures by West and Kenny (2011), directional bias was indexed by the mean difference between perceptions of the partner’s regard and the partner’s actual regard across the discussion. Perceptions of the partner’s regard were, on average, lower than the partner’s actual regard, indicating that perceivers generally underestimated their partner’s regard. Tracking accuracy was indexed by the association (i.e., correlation) between perceptions of the partner’s regard and the partner’s actual regard across the discussion (independent of mean-level differences). This association was strong and positive indicating that perceivers accurately tracked the ups and downs of their partner’s changing regard across the discussion.

Overall et al. (2012) also examined whether levels of directional bias and tracking accuracy varied according to how secure individuals were in their relationship. As predicted, greater insecurity was associated with greater underestimation of the partner’s regard (greater directional bias). More insecure women also more accurately tracked changes in regard, specifically detecting when their partner’s regard became more negative during the discussion (tracking accuracy). This pattern demonstrates that individuals can be both biased and accurate; in this case, underestimating their partner’s regard but accurately tracking changes in their partner’s regard. The results also indicate that relationship insecurity is associated with both greater directional bias and greater tracking accuracy.

This mix of directional bias and tracking accuracy makes sense when considering the costs associated with overestimating the partner’s love and acceptance, including the partner’s dissatisfaction, potential rejection, and failure to take remedial action (Haselton & Buss, 2000). Intimates who are relatively cautious in their judgments and vigilantly monitor signs that their partner’s commitment is waning should avoid the costs of overestimating acceptance and be more able to detect possible rejection (Tooby & Cosmides, 1996). Moreover, those whose relationship histories amplify expectations of rejection should be most alert to how they are valued (Murray & Holmes, 2009). Accordingly, insecurity about the partner’s regard should motivate more cautious assessment (greater directional bias) and vigilant monitoring (greater tracking accuracy) of the partner’s current acceptance.

Depressive Symptoms and Perceptions Within Intimate Relationships

The current research investigated whether depressive symptoms are similarly associated with greater directional bias and enhanced tracking accuracy by comparing perceptions of the partner’s sentiments and behavior to the sentiments and behavior reported by the partner. Specifically, we tested the degree to which individuals with elevated depressive symptoms were biased in their perceptions of their partner’s commitment and behavior, and the degree to which they accurately tracked changes in their partner’s commitment and behavior across time. We focused on these judgments because perceptions of the partner’s commitment and behavior are relevant to the interpersonal difficulties associated with depressive symptoms.

Depressive symptoms are associated with a range of behaviors that damage relationship well-being, such as excessive and hostile bids for support and more caustic reactions to conflict (e.g., Davila, Bradbury, Cohan, & Tochluk, 1997; Potthoff, Holahan, & Joiner, 1995; see Rehman, Gollan, & Mortimer, 2008). Relationship difficulties and dissatisfaction, in turn, increase depressive symptoms over time (e.g., Beach, Katz, Kim, & Brody, 2003). Perceptions of the partner undoubtedly play a central role in these reciprocal links. Negative relationship evaluations are intricately tied to perceptions of the partner’s commitment, including behavior that signals the partner’s regard and future responsiveness (Murray & Holmes, 2009; Overall & Fletcher, 2010; Reis et al., 2004). More negative perceptions of the partner’s regard and conduct, in turn, triggers defensive and destructive relationship behavior (Murray & Holmes, 2009; Reis et al., 2004).

Interpersonal theories of depression (and associated evidence) also suggest that depressive symptoms should create biased partner perceptions (Coyne, 1976). Depressive symptoms are associated with seeking not only reassurance (Starr & Davila, 2008) but also self-verifying negative information (Swann, Wenzlaff, Krull, & Pelham, 1992). This unfortunate combination produces doubts about the authenticity of others’ care and support, which sparks a destructive cycle whereby consequent hostilely toned reassurance seeking elicits rejection, which consolidates negative interpersonal expectations and increases the need for further reassurance (see Joiner & Timmons, 2009). Accordingly, individuals with elevated depressive symptoms are likely to actively gather feedback regarding their partner’s commitment but underestimate that commitment. They should also be attuned to potential rejection, and thus perceive the tone and meaning of their partner’s behavior more negatively than is justified.

For these reasons, we predicted that depressive symptoms would be associated with a negative pattern of directional bias, including underestimating the partner’s commitment and overestimating the partner’s negative behavior (when comparing perceptions to the actual commitment and behavior reported by the partner). This prediction is consistent with biases demonstrated outside the relationship domain. Depressive symptoms are associated with elaborating or ruminating on the causes and consequences of negative events, and being unable to disengage processing of negative personally relevant information (see Gotlib & Joormann, 2010). The predicted pattern is also consistent with theoretical models that propose depressed mood arises in response to social loss and adaptively triggers sensitivity toward social threat, avoidance of social risks, and sustained problem analysis (i.e., rumination) to facilitate the resolution of social dilemmas and minimize the possibility of social exclusion (Allen & Badcock, 2003; Andrews & Thomson, 2009). In support, depression is associated with greater attention to socially threatening information (e.g., Mathews, Ridgeway, & Williamson, 1996) and linked with more cautious deliberation of social risks and less willingness to take those risks (e.g., Badcock & Allen, 2003; Pietromonaco & Rook, 1987).

However, greater vigilance regarding social threat should not only produce more directional bias in relationship perceptions but also more accurate detection and careful monitoring of interpersonal threat—that is, tracking accuracy. Accordingly, research outside the relationship domain has found that depressive symptoms are associated with greater motivation to understand others’ attitudes and behavior, integrating more information when generating impressions, and greater attention and sensitivity to diagnostic information (e.g., Edwards, Weary, von Hippel, & Jacobson, 2000; Gleicher & Weary, 1991). Depressive symptoms are also associated with more accurately distinguishing between genuine versus deceptive communications from others (Lane & DePaulo, 1999). Thus, in addition to producing directional bias, depressive symptoms should activate more accurate tracking of changes in the partner’s commitment and behavior, and in particular more sensitive detection of reductions in the partner’s commitment and increases in the partner’s negative behavior.

Prior research investigating the links between depressive symptoms and perceptions of intimate partners do not speak to our predictions. During couples’ discussions of a relationship or personal difficulty, depressive symptoms have been associated with lower (Gadassi, Mor, & Rafaeli, 2011; Papp, Kouros, & Cummings, 2010) and greater (Papp et al., 2010) accuracy of the partner’s thoughts and feelings, or have not been associated with accuracy at all (Thomas, Fletcher, & Lange, 1997). The meaning of these results is hard to unravel because the methods used have not assessed the type or direction of potential inaccuracy. Gadassi et al. (2011), for example, asked independent coders to rate the similarity between participants’ descriptions of their partners’ thoughts and feelings during a conflict and the descriptions provided by the partner (Study 1) and calculated the absolute difference between individuals’ daily perceptions and their partners’ reported feelings across 21 days (Study 2). For women, depressive symptoms were associated with lower coded similarity in the discussion (Study 1) and greater discrepancy in perceptions across days (Study 2). However, ratings of similarity and absolute difference scores do not provide any information about the direction of any bias, such as whether perceptions are more or less negative (directional bias) than the partner’s actual thoughts and feelings. In addition, focusing on judgments made at one time point (Papp et al., 2010) or collapsing similarity assessments across multiple time points (Gadassi et al., 2011; Thomas et al., 1997) does not allow an examination of whether individuals are able to detect changes in their partner’s thoughts, feelings, and behavior (tracking accuracy).

In the current research, perceptions of the partner’s commitment and behavior, and corresponding benchmark ratings by the partner, were gathered daily for 3 weeks. We simultaneously assessed directional bias and tracking accuracy, and predicted that greater depressive symptoms would be associated not only with (a) greater directional bias, including underestimating the partner’s commitment and overestimating the partner’s negative behavior, but also (b) greater tracking accuracy, including more accurately detecting changes in the partner’s commitment and behavior across days.

The Consequences of Directional Bias and Tracking Accuracy

The predicted pattern of directional bias and tracking accuracy should have important consequences for how secure people feel in their relationships. Research by Murray and colleagues suggests that optimistic perceptions are necessary to maintain a sense of conviction and security within relationships (e.g., Murray et al., 2011; Murray, Holmes, & Griffin, 1996). People who perceive their partner in a positively biased fashion maintain greater relationship satisfaction, experience less conflict, and report fewer doubts about their relationship. Conversely, when people possess more negatively biased views of their partner, they experience less trust and security and poorer relationship quality (also Miller & Rempel, 2004). Thus, the directional biases we expected to be associated with depressive symptoms, including underestimating the partners’ commitment and overestimating negative partner behavior, should lead to greater doubts and insecurities about the relationship.

This prediction captures central elements of the interpersonal cycle that sustains the relational difficulties associated with depression (Coyne, 1976; Joiner & Timmons, 2009); doubts about the partner’s care and support trigger reassurance seeking, biased perceptions of the partner’s responsiveness means any reassurance actually delivered by the partner fails to reassure, and resulting insecurity escalates reassurance seeking, which pushes the partner away. Joiner (2000) also outlined that this vicious cycle propagates interpersonal difficulties and maintains depression. As already noted, lower relationship quality predicts increases in depressive symptoms (e.g., Beach et al., 2003). Perceiving lower partner commitment and more negative partner behavior should also lead to greater depressed mood. Thus, the biases associated with depressive symptoms are likely to promote and maintain depressed mood.

Moreover, these damaging effects might be more pronounced when intimates accurately detect lower commitment and more negative partner behavior. Although accurate knowledge of each other can enhance partners’ felt understanding, accuracy can be costly in relationship-threatening situations. Simpson, Oriña, and Ickes (2003) found that greater accuracy when the partner’s thoughts and feelings were threatening for the relationship reduced relationship closeness. Similarly, in daily life, accurate detection of declining commitment and growing negativity is likely to generate relationship insecurity and depressed mood. Accordingly, underestimating partner commitment and overestimating negative partner behavior should foster depressed mood and relationship insecurity, and this should be particularly the case when these biases accurately track reality—that is, in the context of actual drops in partner commitment and increases in negative partner behavior.

To test these ideas, participants were asked to report their daily feelings of relationship insecurity and depressed mood. We predicted that more negative perceptions of the partner’s commitment and behavior would be associated with increases in relationship insecurity and depressed mood, and these deleterious effects would be even more pronounced when negative perceptions were more accurate—that is, on days that the partner’s commitment was actually low and the partner was actually behaving more negatively.

Current Research

To test whether depressive symptoms were associated with directional bias and tracking accuracy, both partners of committed heterosexual couples completed a 3-week daily diary reporting on their own and perceptions of their partner’s commitment and behavior. Comparing individuals’ perceptions of their partner’s commitment and behavior to the commitment and behavior reported by the partner, we hypothesized that depressive symptoms would be associated with both (a) greater directional bias, including underestimating the partner’s commitment and overestimating the partner’s negative behavior, and (b) greater tracking accuracy, including more sensitive detection of reductions in the partner’s commitment and increases in the partner’s negative behavior across days.

To test whether the expected pattern of bias and accuracy cultivated insecurity and depressed mood, daily feelings of relationship insecurity and depressed mood were gathered. We predicted that more negative perceptions of the partner’s commitment and behavior would be associated with increases in relationship security and depressed mood, and these damaging effects would be more pronounced when negative perceptions were accurate—that is, when the partner also reported lower commitment and more negative behavior.

Finally, we assessed other factors prior research has tied to biased relationship perceptions to ensure these variables were not responsible for any links between depressive symptoms and bias and accuracy. We focused on the three most well-established factors that influence bias within intimate relationships. First, people tend to assume close others evaluate and behave in similar ways to the self (assumed similarity or projection; Kenny & Acitelli, 2001). Second, people who are happier in their relationships possess more positively biased perceptions, and vice versa (Fletcher & Kerr, 2010). Third, insecurity in the partner’s acceptance, often indexed by low self-esteem, produces more negatively biased perceptions and greater tracking accuracy (e.g., Murray & Holmes, 2009; Overall et al., 2012). We predicted that depressive symptoms would remain significantly associated with directional bias and tracking accuracy when controlling for these alternative explanations.

Method

Participants

Seventy-eight heterosexual couples who replied to campus wide advertisements were reimbursed $45NZD for participating. Participants were on average 22.44 years old (SD = 4.81) involved in serious romantic relationships (43.6% married or cohabitating) that were on average 2.57 years in length (SD = 1.96).

Materials and Procedure

During an initial session, couples completed the scales described below and were given detailed instructions for completing a 3-week daily diary.

Depressive Symptoms were assessed with the Centre for Epidemiological Studies Depression Scale (CES-D; Radloff, 1977) designed for use with nonclinical samples. The 20-item scale assesses the frequency of symptoms experienced during the past week (e.g., “I felt depressed,” “I felt that everything I did was an effort”). Responses ranged from 0 = rarely or none of the time (less than 1 day) to 3 = most or all of the time (5-7 days) and were scored and summed so that higher scores (out of 60) indicate the presence of more symptoms. Although the CES-D is not a diagnostic tool, scores ≥ 16 are typically considered evidence of meaningful depressive symptoms. Table 1 displays descriptive statistics for this sample; 41% of men and 42% of women scored ≥ 16. As is typical, women reported higher levels of depressive symptoms, but there were no significant gender differences in the effects of depressive symptoms tested below.

Table 1.

Descriptive Statistics and Correlations Across Questionnaire Measures.

	Women		Men		Gender differences		Correlations
	M (SD)	α	M (SD)	α	t	d	1	2	3
1. Depressive symptoms	18.10 (10.68)	.92	15.23 (7.05)	.84	2.25***	.32	.25***	−.61***	−.39***
2. Self-esteem	5.10 (1.18)	.91	5.38 (0.95)	.86	−2.02***	−.23	−.65***	.24***	.25***
3. Relationship quality	6.12 (0.63)	.76	6.12 (0.61)	.83	0.02	.00	−.20*	.16	.34***

Note: Possible scores range from 0 to 60 for depressive symptoms and 1 to 7 for self-esteem and relationship quality. Gender differences are tested using paired-samples t tests and effect sizes are indexed by Cohen’s d corrected for the dependence between means. Correlations above the diagonal are for women; correlations below the diagonal are for men. Bold correlations on the diagonal represent correlations across partners.

p < .10. ***p < .05.

Relationship and Self-Evaluations

Participants also completed the Rosenberg’s (1965) 10-item Self-Esteem Scale (e.g., “On the whole, I am satisfied with myself”; 1 = not at all, 7 = extremely), and the 7-item Perceived Relationship Quality Components Scale (Fletcher, Simpson, & Thomas, 2000) to assess relationship quality (e.g., “How satisfied are you with your relationship?” 1 = not at all, 7 = extremely). These scales were reliable and correlated with depressive symptoms in expected ways (see Table 1).

Daily Diary

At the end of the day for the following 21 days, participants completed a web-based diary record reporting on their relationship-related feelings and behavior. On average, articipants completed 19.3 diary entries. To assess bias and accuracy, each diary record asked participants to report on (a) their own commitment and behavior and (b) perceptions of their partner’s commitment and behavior that day (order counterbalanced across couples). Participants also reported their feelings of relationship insecurity and depressed mood.

Commitment

Single items assessed participants’ own commitment (“I was committed to our relationship”) and their perceptions of their partner’s commitment (“My partner was committed to our relationship,” 1 = not at all, 7 = extremely).

Negative Behavior

Four items assessed participants’ own hurtful and neglectful behavior (e.g., “I acted in a way that could be hurtful to my partner,” “I was critical or unpleasant toward my partner,” “I wanted to be left alone and/or spend less time with my partner,” “I withdrew from my partner and did my own thing,” 1 = not at all, 7 = extremely), and four items assessed positive and responsive relationship behavior (“I was affectionate and loving toward my partner,” “I was supportive to my partner,” “I shared and discussed my feelings and opinions with my partner,” “I tried to maintain or improve the quality of our relationship”). These items capture the range of destructive and constructive responses shown to be important within daily relationship interactions (Rusbult, Verette, Whitney, Slovik, & Lipkus, 1991). The positive items were reverse-scored and the eight items were averaged to provide an overall index of behavior (α = .83), with greater scores reflecting more negative behavior. The same eight items were reworded and scored to assess perceptions of the partner’s negative behavior (e.g., “My partner was critical or unpleasant toward me,” α = .87).

Relationship Insecurity

Participants were asked to rate three items according to how they felt and thought that day, including “I worried about our relationship,” “I felt confident that my partner loves me” (reverse-coded), and “I felt insecure about our relationship” (1 = not at all, 7 = extremely). These items were averaged to assess participants’ feelings of relationship insecurity (α = .81).

Depressed Mood

Depressed mood was assessed with items shown to reliably detect daily changes in depressed mood (Cranford et al., 2006). Individuals rated how much they had felt “sad,” “hopeless,” and “discouraged” that day (1 = not at all, 7 = extremely; α = .87).

Results

The results are organized into three sections. The first section presents baseline models assessing directional bias and tracking accuracy. The second section examines the links between depressive symptoms and bias and accuracy, and considers alternative explanations for the effects. The final section explores the consequences of bias and accuracy for feelings of relationship insecurity and depressed mood.

Are Perceptions of the Partner’s Commitment and Behavior Biased or Accurate?

We used West and Kenny’s (2011) Truth and Bias Model to test the degree to which perceptions of the partner’s daily commitment and behavior were biased and accurate. The person making the judgment of their partner’s commitment and behavior is referred to as the perceiver and their judgments are compared with their partner’s own ratings of commitment and behavior. Our data have a nested structure, with perceivers and partners multiple ratings of commitment and behavior across the 21 days (Level 1) nested within dyad (Level 2). Accordingly, we used multilevel modeling methods for analyzing repeated-measures data within dyads (Kenny, Kashy, & Cook, 2006). Using perceptions of commitment to demonstrate, we modeled the associations across the perceivers’ judgments of their partner’s commitment and the partners’ actual reported commitment (the Level 1 repeated-measures variables) to test the degree to which perceptions of the partner’s commitment were biased and accurate (see Equation 1).

J_{i j} = b_{0 j} + b_{1 j} (partner j ’s actual commitment on day i) + e_{i j} .

In this equation, the judgment of perceiver j of his or her partner’s commitment (J) on a particular day (i) is a function of perceiver j’s intercept (b₀), the effect of the partner’s actual commitment for that day (b₁), and an error term (e_ij) representing random error and all other unmeasured biases that influence the perceiver’s judgments. As specified by West and Kenny (2011), the perceivers’ judgments of the partners’ commitment (the outcome variable) were centered on the partners’ actual commitment by subtracting the grand mean of partners’ commitment from the perceivers’ judgment each day. This centering strategy means that the intercept represents the difference between the partners’ commitment and the perceivers’ perceptions of that commitment, and specifies the direction of that bias (directional bias). A negative intercept would indicate that perceivers were underestimating their partner’s commitment, whereas a positive intercept would indicate perceivers were overestimating.

The predictor variable (partner’s actual commitment) was grand-mean centered across dyads and time points and its coefficient assesses tracking accuracy—the degree to which perceiver’s judgments were influenced by the partner’s actual commitment. A positive coefficient would indicate that perceivers were accurately tracking the degree to which ttheir partner’s commitment varied across the 21-day diary period.

All analyses were conducted using the MIXED procedure in SPSS 19. We first used a no-intercept model to simultaneously estimate the parameters from Equation 1 for men and women separately. To test whether the effects significantly differed across men and women, in a second model, we estimated the fixed effects pooled across men and women, and tested for gender differences by modeling the main and interaction effects of gender (–1 women, 1 men; see Kenny et al., 2006). Both models allowed the error variances to differ for men and women and allowed errors for a given time to be correlated. Both models also allowed directional bias (b_0j) and tracking accuracy (b_1j) to vary by male and female perceivers for each dyad (i.e., be random variables) and these effects to covary within and across dyad members. See Overall et al. (2012) for associated SPSS syntax.

The results from both models are shown in Table 2. Because there were no statistically significant gender differences in directional bias or tracking accuracy (see column marked Gender Diff. t), we attend to the fixed effects pooled across men and women shown in the final columns of Table 2. The average intercept assessing directional bias did not significantly differ from zero, indicating that on average, men and women were not biased in their perceptions of their partner’s commitment across days. The significant coefficient testing accuracy indicated that perceivers also accurately tracked their partner’s commitment across days. Analogous models examining perceptions of the partner’s negative behavior (see bottom half of Table 2) also revealed that, on average, perceivers were not significantly biased and were very accurate in tracking changes in their partner’s behavior.

Table 2.

Directional Bias and Tracking Accuracy of Perceptions of the Partner’s Daily Commitment and Negative Behavior.

Directional bias and tracking accuracy	Women perceivers			Men perceivers			Pooled across men and women
Directional bias and tracking accuracy	B	t	r	B	t	r	Gender Diff. t	B	t	r
Partner’s commitment
Directional bias	.04	0.42	.05	−.14	−1.55	−.17	−1.58	−.05	−0.73	−.09
Tracking accuracy	.22	3.94****	.56	.27	6.20****	.75	0.71	.25	6.99****	.70
Partner’s negative behavior
Directional bias	−.01	−0.08	−.01	−.05	−0.68	−.08	−0.34	−.03	−0.90	.11
Tracking accuracy	.79	19.62****	.93	.70	18.83****	.92	−1.62	.75	28.18****	.94

Note: Fixed effects pooled across men and women are from models estimating the fixed effects constrained to be equal across men and women and testing whether the effects significantly differed by estimating the main and interaction effects of gender (shown in column marked Gender Diff. t). Directional bias and tracking accuracy did not significantly differ across men and women. Approximate effect sizes were computed using Rosenthal and Rosnow’s (2007) formula: r = √(t² / t² + df). Degrees of freedom for each effect varied from 30.20 to 80.95.

****

p < .01.

Table 3 displays tests of whether bias and accuracy varied significantly across perceivers. Although on average directional bias did not differ from zero, there was significant variation in the degree to which women and men perceivers were biased in perceptions of their partner’s commitment and behavior. Similarly, some perceivers were more accurate than others. The covariances presented in Table 3 also suggested that women perceivers who were more biased (underestimating commitment or overestimating negative behavior) also demonstrated greater tracking accuracy. Directional bias and tracking accuracy did not significantly covary across partners with one exception: the more one dyad member under- or overestimated the negativity of their partner’s behavior, the less the partner demonstrated the same directional bias.

Table 3.

Variances and Covariances of Directional Bias and Tracking Accuracy of Perceptions of the Partner’s Commitment and Negative Behavior.

Variances and covariances	Partner’s commitment				Partner’s negative behavior
	Women perceivers		Men perceivers		Women perceivers		Men perceivers
	Estimate	Wald Z	Estimate	Wald Z	Estimate	Wald Z	Estimate	Wald Z
Variances in bias and accuracy
Directional bias	.53	5.36****	.56	5.60****	.37	5.11****	.43	5.58****
Tracking accuracy	.09	2.61****	.04	1.84***	.06	2.81****	.06	3.66****
Within-person covariance
Directional bias and tracking accuracy	−.09	−2.34****	−.02	−0.64	.05	1.91**	.01	0.57
Across-partner covariance
Directional bias and partner’s directional bias	.14	1.73	.14	1.73	−.27	−4.22****	−.27	−4.22****
Tracking accuracy and partner’s tracking accuracy	−.00	−0.18	−.00	−0.18	.01	0.08	.01	0.08
Directional bias and partners’ tracking accuracy	.02	0.48	−.01	−0.15	−.02	−0.69	.01	0.40

Note: Tests of variances are one-tailed (see Kenny, Kashy, & Cook, 2006).

p < .06. ***p < .05.****p < .01.

In sum, on average, perceivers demonstrated substantial accuracy in tracking changes in their partner’s commitment and behavior across days, and did not tend to under- or overestimate their partner’s commitment or negative behavior. However, directional bias and tracking accuracy significantly varied across perceivers. The significant variation in bias and accuracy is important because it indicates that differences across perceivers, such as levels of depressive symptoms, might predict the degree to which perceivers are biased and accurate.

Are Depressive Symptoms Associated With Bias and Accuracy?

To test whether perceivers’ depressive symptoms predicted bias and accuracy, CES-D scores (standardized to aid interpretation of parameter estimates) were entered as a predictor of the between-person variability in directional bias and tracking accuracy (each parameter estimated by Equation 1). In these analyses, the Level 1 intercept (modeling directional bias) and slope (modeling tracking accuracy) were treated as dependent variables predicted by individual differences in depressive symptoms modeled at Level 2 (see Equations 2 and 3).

b_{0 j} = B_{00} + B_{01} (depressive symptoms) + u_{0 j} .

b_{1 j} = B_{10} + B_{11} (depressive symptoms) + u_{1 j} .

Equation 2 examines the effect of depressive symptoms on directional bias (b_0j), where B₀₀ represents the Level 2 intercept reflecting average levels of directional bias across perceivers, B₀₁ is a coefficient testing whether perceivers’ depressive symptoms predict levels of directional bias, and u_0j represents individual variation in bias. We predicted that greater depressive symptoms would be associated with more negative directional bias when perceiving commitment (i.e., underestimation of the partner’s commitment) and positive directional bias when perceiving negative behavior (i.e., overestimation of the partner’s negative behavior).

Equation 3 also gives the cross-level interaction between accuracy and perceivers’ depressive symptoms, which assesses whether tracking accuracy varied according to levels of depressive symptoms. In this equation, the person’s slope coefficient b_1j or tracking accuracy is modeled as a function of the main effect of accuracy (B₁₀), the moderating effect of depressive symptoms on accuracy (B₁₁), and an error term allowing for variation in slopes across perceivers (u_1j). We predicted that greater depressive symptoms would be associated with greater tracking accuracy of both the partner’s commitment and behavior.

Table 4 displays the results testing whether depressive symptoms were associated with directional bias and tracking accuracy when perceiving the partner’s commitment (shown in left half of Table 4) and negative behavior (right half of Table 4). As before, the fixed effects did not significantly differ across men and women (see column marked Gender Diff. t, ps = .34-.95) and so we attend to and present the effects pooled across men and women. As predicted, participants who reported greater depressive symptoms demonstrated both greater bias and greater tracking accuracy in their perceptions of their partner’s commitment and behavior.

Table 4.

The Effects of Depressive Symptoms on Directional Bias and Tracking Accuracy of Perceptions of the Partner’s Commitment and Negative Behavior.

Effect of depressive symptoms	Partner’s commitment				Partner’s negative behavior
Effect of depressive symptoms	B	t	r	Gender Diff. t	B	t	r	Gender Diff. t
Directional bias	−0.18	−2.67****	.23	.06	0.10	2.33***	.28	.86
Tracking accuracy	0.07	2.13***	.28	.86	0.09	3.01****	.28	.96

Note: Results are from models estimating the fixed effects pooled across men and women and testing whether the effects significantly differed across men and women by estimating the main and interaction effects of gender (shown in column marked Gender Diff. t). The effects of depressive symptoms did not differ across men and women. Approximate effect sizes were computed using Rosenthal and Rosnow’s (2007) formula: r = √(t² / t² + df). Degrees of freedom for each effect varied from 54.23 to 123.13.

***

p < .05. ****p < .01.

The impact of depressive symptoms on perceptions of commitment are shown in Figure 1 where the predicted values of perceptions of the partner’s commitment are plotted at low (−1 SD) and high (+1 SD) levels of the partner’s actual commitment (the predictor testing accuracy) for perceivers who reported low (−1 SD) versus high (+1 SD) levels of depressive symptoms. Recall that perceivers’ judgments of the partner’s commitment (on the y axis) are centered on the partners’ actual commitment, so zero represents no directional bias, positive values indicate overestimating commitment, and negative values indicate underestimating commitment (see key on right side of Figure 1). The slopes across low versus high levels of partner’s actual commitment (on the x axis) demonstrate tracking accuracy for perceivers high (top slope in Figure 1) versus low (bottom slope in Figure 1) in depressive symptoms.

Figure 1.

The effect of depressive symptoms on perceptions of partner’s daily levels of commitment across a 21-day period.

As predicted, perceivers with high levels of depressive symptoms tracked changes in their partner’s commitment across the 21 days more accurately (b = .30, t = 6.40, p < .001) than perceivers low in depressive symptoms (b = .15, t = 2.98, p < .01). However, Figure 1 also highlights that this pattern occurs because perceivers high in depressive symptoms were more sensitive to drops in their partner’s commitment. On days when the partner’s actual commitment was high (see right side of Figure 1), there was no significant difference in the perceptions of participants low versus high in depressive symptoms (b = −.09, t = −1.25, p = .22). In contrast, on days when the partner’s actual commitment was low (left side of Figure 1), perceivers high in depressive symptoms underestimated their partner’s commitment (i.e., judgments well below zero) whereas perceivers low in depressive symptoms did not (b = −.27, t = −3.14, p < .01). This pattern indicates that perceivers with greater depressive symptoms accurately detected when their partner’s commitment waned across the diary period, but then overestimated these drops in their partner’s commitment.

The impact of depressive symptoms on perceptions of the partner’s behavior demonstrated a corresponding pattern. Shown in Figure 2, perceivers with high levels of depressive symptoms tracked changes in their partner’s behavior across the 21 days more accurately (b = .62, t = 12.57, p < .001) than perceivers low in depressive symptoms (b = .43, t = 9.07, p < .001). However, the key differences between perceivers high versus low in depressive symptoms occurred when the partner’s negative behavior was high. On days when the partner reported low levels of negative behavior (see left side of Figure 2), there was no difference in the judgments of perceivers low versus high in depressive symptoms (b = .03, t = 0.49, p = .62). In contrast, on days when the partner reported high levels of negative behavior (left side of Figure 2), perceivers high in depressive symptoms reported much greater levels of negative behavior (i.e., biased judgments well above zero) whereas perceivers low in depressive symptoms did not (b = .25, t = 3.80, p < .01). Thus, perceivers with greater depressive symptoms accurately detected when their partner’s negative behavior increased but then exaggerated those increases.

Figure 2.

The effect of depressive symptoms on perceptions of the partner’s negative behavior across a 21-day period.

Alternative Explanations

Own Commitment and Behavior

Consistent with assumed similarity or projection processes (Kenny & Acitelli, 2001), perceptions of the partner’s commitment and behavior is likely to be influenced by individuals’ own commitment and behavior. We reran all analyses controlling for perceivers’ own daily commitment and behavior. Although own commitment (B = .54, t = 15.74, p < .001) and behavior (B = .73, t = 31.46, p < .001) were strongly related to the corresponding judgments of the partner, depressive symptoms were not associated with levels of projection when judging the partner’s commitment (B = −.04, t = −1.13, p = .26) or behavior (B = −.03, t = −1.21, p = .23), and controlling for projection did not eliminate the interactions shown in Figures 1 (B = .06, t = 2.19, p < .05) and 2 (B = .04, t = 1.96, p = .05).

Relationship Quality

Depressive symptoms were also associated with lower relationship quality (see Table 1), and lower relationship quality was associated with more positive judgments of partner commitment (B = .51, t = 5.38, p < .001) and more negative perceptions of partner behavior (B = −.36, t = −4.99, p < .001). Relationship quality did not predict accuracy, however, and controlling for relationship quality did not alter the pattern or associated effects shown in Figures 1 and 2.

Relationship Insecurity

Additional analyses also demonstrated that the results were not a product of chronic insecurity regarding acceptance (indexed by low self-esteem; Murray & Holmes, 2009) or more specific daily feelings of relationship insecurity. Despite being strongly correlated with depressive symptoms (see Table 1), self-esteem did not predict bias and accuracy or alter any of the effects shown in Table 4, and Figures 1 and 2. Daily feelings of relationship insecurity were associated with directional bias (B = −.45, t = −10.46, p < .001) and tracking accuracy (B = .04, t = 2.12, p = .03) in perceiving commitment (in the same manner depicted in Figure 1) as well as greater bias in perceiving the partner’s negative behavior (B = .48, t = 27.53, p < .001). Nonetheless, depressive symptoms continued to significantly predict greater bias (B = −.21 and .15, t > 2.95, p < .01) and tracking accuracy (B = .08 and .07, t > 2.35, p < .05) for both commitment and negative behavior (respectively).

Do Negative Relationship Perceptions Foster Relationship Insecurity and Depressed Mood?

We predicted that the mix of directional bias and tracking accuracy demonstrated by individuals with elevated depressive symptoms would foster relationship insecurity and cultivate depressed mood. To test this, we adopted similar multilevel modeling methods for analyzing repeated-measures data within dyads (Kenny et al., 2006). Using perceptions of the partner’s commitment and relationship insecurity to illustrate, we regressed perceiver’s relationship insecurity on (a) perceiver’s relationship insecurity the prior day, so that any significant effects represented residual change (i.e., increases or decreases relative to prior levels) in insecurity, (b) perceiver’s perceptions of the partner’s commitment, (c) the partner’s actual commitment, and (d) the interaction between perceptions of the partner’s commitment and partner’s actual commitment. All predictors were person-centered and were allowed to vary by male and female perceivers for each dyad (i.e., be random variables).²

In these analyses, the main effect of perceptions of the partner’s commitment tests whether more negative perceptions of the partner’s commitment was associated with within-person increases in relationship insecurity, over and above the impact of the partner’s actual commitment. The interaction term tests whether any within-person increases in relationship insecurity associated with underestimating the partner’s commitment was greater when negative perceptions were more accurate—that is, on days when the partner’s commitment was actually low. Four sets of analyses tested the links between perceptions of the partner’s commitment or negative behavior and relationship insecurity or depressed mood. Across analyses, the fixed effects did not significantly differ across gender (1 exception out of 12; see Gender Diff. t, Table 5) and so we present the effects pooled across men and women.

Table 5.

The Effects of Perceptions of the Partner’s Commitment and Negative Behavior on Daily Feelings of Relationship Insecurity and Depressed Mood.

	Relationship insecurity				Depressed mood
	B	t	r	Gender Diff. t	B	t	r	Gender Diff. t
Partner’s commitment
Perceptions of partner’s commitment	−.50	−12.36****	.87	1.35	−.41	−10.02****	.80	2.45***
Partner’s actual commitment	−.12	−3.23****	.41	0.88	−.17	−4.05****	.47	−0.37
Perceptions × Actual commitment	.08	3.71****	.13	−1.57	.02	0.65	.02	0.82
Partner’s negative behavior
Perceptions of partner’s negative behavior	.45	12.72****	.82	−1.49	.50	12.31****	.81	1.13
Partner’s actual negative behavior	.06	2.43***	.31	0.79	.03	1.13	.14	1.23
Perceptions × Actual negative behavior	.07	3.70****	.08	1.06	.05	2.30***	.06	0.37

Note: All analyses controlled for relationship insecurity/depressed mood on the prior day; thus, significant effects represent predicted decreases or increases relative to prior levels of insecurity/depressed mood. Only the focal fixed effects are shown to simplify the presentation. Results are from models estimating the fixed effects pooled across men and women and testing whether the effects significantly differed by estimating the main and interaction effects of gender (shown in column marked Gender Diff. t). The majority of effects (11 out of 12) did not significantly differ across men and women. The one exception is discussed in the text. Approximate effect sizes were computed using Rosenthal and Rosnow’s (2007) formula: r = √(t² / t² + df).

***

p < .05. ****p < .01.

The results are shown in Table 5. As predicted, perceiving the partner as less committed and behaving more negatively was associated with increased relationship insecurity and depressed mood. Moreover, the significant interactions between perceptions and partner’s actual commitment and negative behavior for three of the four models indicated that the associations between more negative perceptions and relationship insecurity/depressed mood were more pronounced when more negative perceptions aligned accurately with lower commitment and more negative behavior reported by the partner.

To illustrate, Figure 3 plots the significant interaction between perceptions of the partner’s commitment and the partner’s actual commitment on changes in daily relationship security. Perceiving low partner commitment was associated with greater relationship insecurity regardless of whether the partner’s actual commitment was high (b = −.57, t = −13.04, p < .001) or low (b = −.44, t = −9.88, p < .001), but the effect of perceiving low commitment was most negative when perceptions accurately reflected lower commitment reported by the partner (b = −.18, t = −4.54, p < .001).

Figure 3.

The impact of perceptions of partner’s commitment and partner’s actual commitment on perceiver’s daily feelings of relationship insecurity.

Analogous patterns were found when modeling perceptions of the partner’s negative behavior on relationship insecurity and depressed mood (i.e., perceiving more negative behavior by the partner was associated with greater depressed mood, particularly when the partner agreed he or she was behaving more negatively). The only exception was that perceived- and partner-reported commitment independently predicted greater depressed mood but did not significantly interact (see right top of Table 5). The single gender difference (out of 12 effects) also emerged in this model, with the links between perceptions of the partner’s commitment and depressed mood stronger for women (B = −.50, t = − 7.70, p < .01) than men (B = −.31, t = − 7.00, p < .01). Nonetheless, taken together, the results in Table 5 indicate that more negative relationship perceptions were associated with greater relationship insecurity and depressed mood, and there was good evidence (3 of 4 tests) that this was particularly the case on days when negative perceptions more accurately corresponded with lower commitment and more negative behavior reported by the partner. This pattern supports that the mix of directional bias and tracking accuracy that was associated with elevated depressive symptoms is linked with increased relationship insecurity and depressed mood.

Discussion

This study examined whether depressive symptoms are associated with more biased or more accurate social perceptions. Both partners of committed heterosexual couples completed a 3-week daily diary reporting on their own and perceptions of their partner’s commitment and behavior. Comparing individuals’ perceptions of their partner to the commitment and behavior reported by the partner, analyses modeled the degree to which individuals (a) over- or underestimated their partner’s commitment and negative behavior (directional bias) and (b) accurately detected changes in their partner’s commitment and negative behavior across days (tracking accuracy). As predicted, participants who reported greater depressive symptoms exhibited greater directional bias, including underestimating their partner’s commitment and overestimating their partner’s negative behavior (when compared with the commitment and behavior reported by the partner). However, participants with greater depressive symptoms also more accurately tracked changes in their partner’s commitment and behavior across days. Moreover, this pattern held controlling for a series of alternative explanations, including perceivers’ own commitment and negative behavior, relationship quality, self-esteem, and daily levels of relationship insecurity.

The pattern of directional bias and tracking accuracy associated with greater depressive symptoms helps to reconcile prior, seemingly inconsistent, findings. The greater directional bias is consistent with links between depressive symptoms and biased processing of negative interpersonal information (Gotlib & Joormann, 2010), overestimation of negative feelings in others (e.g., Gilboa-Schechtman et al., 2008), and doubts about the sincerity of others’ care and reassurance (Joiner & Timmons, 2009). The greater tracking accuracy is consistent with other research linking depressive symptoms to more thorough processing of interpersonal information, including seeking and integrating more information (e.g., Edwards et al., 2000; Gleicher & Weary, 1991), and more accurate attributions and identification of important interpersonal behavior (e.g., Lane & DePaulo, 1999; Yost & Weary, 1996).

This mix of greater directional bias and tracking accuracy reflects a cautious, vigilant approach to assessing others’ motives and behavior, and in particular detecting interpersonal threat. Such a perceptual approach is consistent with recent models that suggest depressive symptoms reflect an adaptive response to social loss, triggering sensitivity toward social threat, avoidance of social risks, and associated analytical processing (i.e., rumination) of interpersonal problems (see Allen & Badcock, 2003; Andrews & Thomson, 2009). Indeed, perceptions of the partner’s commitment and behavior are crucial because these signal how likely partners will continue to be committed and invested in the relationship versus dissatisfied and rejecting (Murray & Holmes, 2009; Reis et al., 2004). Vigilant monitoring of changes in these signals should prevent further social loss by enhancing detection of any damage to interpersonal connections and subsequent rejection (Tooby & Cosmides, 1996). For example, when sensing partners are dissatisfied, intimates typically try to alter offending self-attributes, and responsive self-regulation efforts work to improve the partner’s evaluations (Overall & Fletcher, 2010; Overall, Fletcher, & Simpson, 2006).

However, vigilant monitoring should not only create more accurate identification of interpersonal threat but also magnify the import and weight given to that threat. Accordingly, individuals who reported greater depressive symptoms more accurately detected when their partner’s commitment weakened (as reported by the partner), but then exaggerated the detected drops in their partner’s commitment. An analogous pattern emerged when examining behavioral signs of the partner’s acceptance and commitment. Although this perceptual pattern may adaptively initiate remedial efforts to prevent interpersonal loss, pushed into overdrive, this pattern is also likely to incur costs, including overreacting to what will often be temporary relationship difficulties, fostering insecurity and propagating depressed mood, and triggering corrective actions that are disproportionate to the degree of threat.

First, even satisfied and invested partners will experience short-term dips in commitment and behave negatively at times, but return to more positive, loving attitudes and behavior once the difficulty has passed. Vigilant tracking coupled with magnifying the significance of any negativity will likely result in intimates reading “too much” into common relationship difficulties, and may reduce the ability for individuals (and their partners) to weather the normal “ups” and “downs” of relationships (Overall et al., 2012; Simpson et al., 2003). This may be one central reason individuals with depressive symptoms react more destructively when experiencing relationship problems (Rehman et al., 2008).

Second, perceiving waning partner commitment and more hurtful, rejecting behavior, particularly in times of conflict or difficulty, is associated with reductions in the degree to which intimates feel secure in their relationship (Murray & Holmes, 2009). Accordingly, in this study, we found that more negative perceptions of the partner’s commitment and relationship behavior were associated with greater increases in relationship insecurity. Moreover, this association was particularly pronounced when the partner reported lower commitment and more negative behavior—that is, when individuals accurately detected and inflated partner negativity. More negative perceptions were also associated with within-person increases in daily depressed mood. This set of findings suggests that the perceptual biases and sensitivity to interpersonal threat associated with elevated depressive symptoms fosters relationship insecurity and propagates depressed mood and, thus, may play a key role in the persistence of depressive symptoms.

Third, attending to and magnifying the meaning of relationship threat may fuel unnecessary or disproportionate remedial actions, such as excessive reassurance seeking to restore felt-security. The links between depressive symptoms and excessive reassurance seeking have been well documented (Joiner & Timmons, 2009). There is also increasing evidence that others find depressive individuals’ attempts to seek reassurance aversive and, thus, ultimately are more rejecting (Benazon, 2000; Katz & Beach, 1997). The perceptual vigilance and bias shown in this study may help to explain why depressed individuals’ reassurance seeking is excessive and aversive. Overestimating and inflating negativity means the need for reassurance does not match the state and quality of the relationship and this mismatch may become tiresome for partners if their reassurance does little to improve biased relationship perceptions or sensitivity to threat. Ironically, the biased and vigilant perceptions shown here might actually generate declines in partner commitment and more negative, rejecting partner behavior.

Strengths, Caveats, and Future Directions

The current research extended prior research on several fronts. The overall pattern reconciles opposing perspectives and research by demonstrating that depressive symptoms are associated with both greater bias and greater accuracy when comparing relationship perceptions to the reports of the partner. The results suggest that these biases are due to a sensitive attunement to important and meaningful contextual changes, such as when partners are thinking or behaving in ways that signal potential rejection. The findings also indicate that this perceptual sensitivity is likely to play an important role in sustaining depressed mood and generating the relationship difficulties associated with depressive symptoms.

Despite the notable strength of assessing perceptions across the course of couples’ daily lives, the correlational nature of the current data prevents causal conclusions. We have focused on how depressive symptoms influence directional bias and tracking accuracy. However, prior research has shown reciprocal links between depression and relationship difficulties (Davila et al., 1997; Rehman et al., 2008), supporting that depression can generate interpersonal stress, which in turn exacerbates depressive symptoms (Hammen, 1991). Indeed, we found that negative relationship perceptions were associated with increases in depressed mood. This raises the possibility that, rather than depressive symptoms shaping bias and accuracy, processes within the relationship were responsible for the perceptual sensitivity and bias we found, and these difficulties caused elevated depressive symptoms. For example, partners who fail to express commitment or vary widely in their day-to-day commitment might elicit more vigilant monitoring and negative bias as well as greater depressive symptoms. However, such within-relationship processes will strongly influence perceived relationship quality and felt-security, and control analyses demonstrated these variables did not account for the associations between depressive symptoms and bias and accuracy. Nonetheless, the links between depressive symptoms and bias and accuracy (and other relationship difficulties) are likely to be self-perpetuating. Tracking these reciprocal processes across time is an important direction for future research.

A principal challenge when assessing bias and accuracy is establishing an appropriate benchmark of reality. Our use of partner reports is consistent with the majority of prior research examining bias and accuracy within relationships (see Fletcher & Kerr, 2010). However, self-serving biases could mean that partners understated their levels of negative behavior. The results suggest this was not the case. On average, across the sample, there was no directional bias and perceivers demonstrated substantial tracking accuracy (see Table 2). This convergence provides strong evidence that our measures generally captured a shared relationship reality. In addition, even if bias in the partner reports meant our index of bias understated or inflated actual directional bias and tracking accuracy, this would not change the meaning or implication of our results. Regardless of overall levels, bias and accuracy significantly differed across high versus low depressive symptoms, and such differences were associated with meaningful changes in relationship insecurity and depressed mood.

Alternative benchmarks that may appear more objective also face challenges to their validity as markers of reality. Third-party observations, such as by family and friends, can be biased because people protect their own relationships by undervaluing others (Gagné & Lydon, 2004). Trained coders can also overestimate or underestimate negativity due to the context and behaviors coded and lack of personal knowledge regarding the intentions and idiosyncratic meaning of specific actions (Gagné & Lydon, 2004). Third parties and objective coders are also only exposed to a limited amount of couples’ behaviors and interactions, whereas we assessed naturally occurring interactions across 3 weeks of couples’ daily life. Furthermore, experimental tasks that assess cognitive biases are also uninformative regarding the veracity of people’s perceptions as they occur in real life. Thus, the use of partner reports as the benchmark of reality allows assessment of bias and accuracy in the ecological context of people’s everyday relationship experiences. Nonetheless, we expect that the same pattern will be evident using other benchmarks, and this is a good direction for future research.

Finally, our data do not generalize to clinical populations. It is possible that clinically diagnosed depression has qualitatively different effects on interpersonal perceptions, such as producing a blanket negative bias that is not sensitive to context. However, previous research using clinical samples provide evidence that a similar pattern might emerge. For example, clinically depressed individuals generally perceive neutral faces to be more sad (Leppänen, 2006) and are also faster to identify anger or sadness in changing expressions (Joormann & Gotlib, 2006). The current methodological approach and statistical design demonstrates how future research can simultaneously distinguish between directional bias and tracking accuracy to clarify the perceptual processes associated with clinically diagnosed depression.

Conclusion

The current research helps to reconcile the debate regarding whether depressive symptoms produce more biased or more accurate interpersonal perceptions. By gathering multiple assessments of interpersonal perceptions and associated benchmarks across couples’ daily lives, we demonstrated that elevated depressive symptoms were associated with greater bias and greater accuracy. Participants who reported more depressive symptoms were more accurate in detecting interpersonal threat, including reductions in their partner’s commitment and increases in their partner’s negative behavior (as reported by the partner), but then amplified those threats, including underestimating commitment and overestimating the partner’s negativity. This perceptual pattern was also associated with increases in relationship insecurity and depressed mood. These results highlight the importance of examining relationship perceptions and processes in understanding the consequences and outcomes of depressive symptoms.

Footnotes

Acknowledgements

We would like to thank Jeffry A. Simpson and Tessa V. West for helpful comments on an earlier draft of this article.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by a University of Auckland Faculty of Science Research Development Grant (UoA 3626244) awarded to the first author.

Notes

References

Allen

N. B.

Badcock

P. B. T.

(2003). The social risk hypothesis of depressed mood: Evolutionary, psychosocial, and neurobiological perspectives. Psychological Bulletin, 129, 887-913.

Alloy

L. B.

Abramson

L. Y.

(1979). Judgment of contingency in depressed and nondepressed students: Sadder but wiser? Journal of Experimental Psychology-General, 108, 441-485.

Andrews

P. W.

Thomson

J. A.

(2009). The bright side of being blue: Depression as an adaptation for analyzing complex problems. Psychological Review, 116, 620-654.

Badcock

P. B. T.

Allen

N. B.

(2003). Adaptive social reasoning in depressed mood and depressive vulnerability. Cognition & Emotion, 17, 647-670.

Beach

S. R. H.

Katz

Kim

Brody

G. H.

(2003). Prospective effects of marital satisfaction on depressive symptoms in established marriages: A dyadic model. Journal of Social and Personal Relationships, 20, 355-371.

Beck

A. T.

(1967). Depression: Clinical, experimental and theoretical aspects. New York, NY: Hoeber.

Benazon

N. R.

(2000). Predicting negative spousal attitudes toward depressed persons: A test of Coyne’s interpersonal model. Journal of Abnormal Psychology, 109, 550-554.

Collins

N. L.

Feeney

B. C.

(2004). Working models of attachment shape perceptions of social support: Evidence from experimental and observational studies. Journal of Personality and Social Psychology, 87, 363-383.

Coyne

J. C.

(1976). Toward an interactional description of depression. Psychiatry, 39, 28-40.

10.

Cranford

J. A.

Shrout

P. E.

Iida

Rafaeli

Yip

Bolger

(2006). A procedure for evaluating sensitivity to within-person change: Can mood measures in diary studies detect change reliably? Personality and Social Psychology Bulletin, 32, 917-929.

11.

Davila

Bradbury

T. N.

Cohan

C. L.

Tochluk

(1997). Marital functioning and depressive symptoms: Evidence for a stress generation model. Journal of Personality and Social Psychology, 73, 849-861.

12.

Edwards

J. A.

Weary

von Hippel

Jacobson

J. A.

(2000). The effects of depression on impression formation: The role of trait and category diagnosticity. Personality and Social Psychology Bulletin, 26, 462-473.

13.

Fletcher

G. J. O.

Kerr

P. S. G.

(2010). Through the eyes of love: Reality and illusion in intimate relationships. Psychological Bulletin, 136, 627-658.

14.

Fletcher

G. J. O.

Simpson

J. A.

Thomas

(2000). The measurement of perceived relationship quality components: A confirmatory factor analytic approach. Personality and Social Psychology Bulletin, 26, 340-354.

15.

Gadassi

Mor

Rafaeli

(2011). Depression and empathic accuracy in couples: An interpersonal model of gender differences in depression. Psychological Science, 22, 1033-1041.

16.

Gagné

F. M.

Lydon

J. E.

(2004). Bias and accuracy in close relationships: An integrative review. Personality and Social Psychology Review, 8, 322-338.

17.

Gilboa-Schechtman

Foa

Vaknin

Marom

Hermesh

(2008). Interpersonal sensitivity and response bias in social phobia and depression: Labeling emotional expression. Cognitive Therapy Research, 32, 605-618.

18.

Gleicher

Weary

(1991). The effect of depression on the quantity and quality of social inferences. Journal of Personality and Social Psychology, 61, 105-114.

19.

Gotlib

I. H.

Joormann

(2010). Cognition and depression: Current status and future directions. Annual Review of Clinical Psychology, 6, 285-312.

20.

Hammen

C. L.

(1991). The generation of stress in the course of unipolar depression. Journal of Abnormal Psychology, 100, 555-561.

21.

Haselton

M. G.

Buss

D. M.

(2000). Error management theory: A new perspective on biases in cross-sex mind reading. Journal of Personality and Social Psychology, 78, 81-91.

22.

Joiner

T. E.

(2000). Depression’s vicious scree: Self-propagating and erosive factors in depression chronicity. Clinical Psychology: Science and Practice, 7, 203-218.

23.

Joiner

T. E.

Timmons

K. A.

(2009). Depression in its interpersonal context. In Gotlib

I. H.

Hammond

C. L.

(Eds.), Handbook of depression (pp. 322-339). New York, NY: Guilford.

24.

Joormann

Gotlib

I. H.

(2006). Is this happiness I see? Biases in the identification of emotional facial expressions in depression and social phobia. Journal of Abnormal Psychology, 115, 705.

25.

Katz

Beach

S. R. H.

(1997). Self-verification and depressive symptoms in marriage and courtship: A multiple pathway model. Journal of Marriage and Family, 59, 903-914.

26.

Kelley

H. H.

Thibaut

J. W.

(1978). Interpersonal relations: A theory of interdependence. New York, NY: Wiley.

27.

Kenny

D. A.

Acitelli

L. K.

(2001). Accuracy and bias in the perception of the partner in a close relationship. Journal of Personality and Social Psychology, 80, 439-448.

28.

Kenny

D. A.

Kashy

D. A.

Cook

W. L.

(2006). Dyadic data analysis. New York, NY: Guilford.

29.

Lane

J. D.

DePaulo

B. M.

(1999). Completing Coyne’s cycle: Dysphorics’ ability to detect deception. Journal of Research in Personality, 33, 311-329.

30.

Dove

N. L.

Agnew

C. R.

Korn

M. S.

Mutso

A. A.

(2010). Predicting nonmarital romantic relationship dissolution: A meta-analytic synthesis. Personal Relationships, 17, 377-390.

31.

Leppänen

J. M.

(2006). Emotional information processing in mood disorders: A review of behavioural and neuroimaging findings. Current Opinion in Psychiatry, 19, 34-39.

32.

Mathews

Ridgeway

Williamson

D. A.

(1996). Evidence for attention to threatening stimuli in depression. Behaviour Research and Therapy, 34, 695-705.

33.

Miller

P. J. E.

Rempel

J. K

. (2004). Trust and partner-enhancing attributions in close relationships. Personality and Social Psychology Bulletin, 30, 695-705.

34.

Murray

S. L.

Griffin

D. W.

Derrick

J. L.

Harris

Aloni

Leder

(2011). Tempting fate or inviting happiness? Unrealistic idealization prevents the decline of marital satisfaction. Psychological Science, 22, 619-626.

35.

Murray

S. L.

Holmes

J. G.

(2009). The architecture of interdependent minds: A motivation-management theory of mutual responsiveness. Psychological Review, 116, 908-928.

36.

Murray

S. L.

Holmes

J. G.

Griffin

D. W.

(1996). The self-fulfilling nature of positive illusions in romantic relationships: Love is not blind, but prescient. Journal of Personality and Social Psychology, 71, 1155-1180.

37.

Murray

S. L.

Holmes

J. G.

Griffin

D. W.

(2000). Self-esteem and the quest for felt security: How perceived regard regulates attachment processes. Journal of Personality and Social Psychology, 78, 478-498.

38.

Overall

N. C.

Fletcher

G. J. O.

(2010). Receiving regulation from intimate partners: Reflected appraisal processes in close relationships. Personal Relationships, 17, 433-456.

39.

Overall

N. C.

Fletcher

G. J. O.

Kenny

D. A.

(2012). When bias and insecurity promote accuracy: Mean-level bias and tracking accuracy in couples’ conflict discussions. Personality and Social Psychology Bulletin, 38, 642-655.

40.

Overall

N. C.

Fletcher

G. J. O.

Simpson

J. A.

(2006). Regulation processes in intimate relationships: The role of ideal standards. Journal of Personality and Social Psychology, 91, 662-685.

41.

Papp

L. M.

Kouros

C. D.

Cummings

M. E.

(2010). Emotions in marital conflict interactions: Empathic accuracy, assumed similarity, and the moderating context of depressive symptoms. Journal of Social and Personal Relationships, 27, 367-387.

42.

Pietromonaco

P. R.

Rook

K. S.

(1987). Decision style in depression: The contribution of perceived risks versus benefits. Journal of Personality and Social Psychology, 52, 399-408.

43.

Potthoff

J. G.

Holahan

C. J.

Joiner

T. E.

(1995). Reassurance seeking, stress generation, and depressive symptoms: An integrative model. Journal of Personality and Social Psychology, 68, 664-670.

44.

Radloff

L. S.

(1977). The CES-D scale: A self-report depression scale for research in the general population. Applied Psychological Measurement, 1, 385-401.

45.

Rehman

U. S.

Gollan

Mortimer

A. R.

(2008). The marital context of depression: Research, limitations and new directions. Clinical Psychology Review, 28, 179-198.

46.

Reis

H. T.

Clark

M. S.

Holmes

J. G.

(2004). Perceived partner responsiveness as an organizing construct in the study of intimacy and closeness. In Mashek

D. J.

Aron

A. P.

(Eds.), Handbook of closeness and intimacy (pp. 201-225). Mahwah, NJ: Erlbaum.

47.

Rosenberg

(1965). Society and the adolescent self-image. Princeton, NJ: Princeton University Press.

48.

Rosenthal

Rosnow

R. L.

(2007). Essentials of behavioral research: Methods and data analysis (3rd ed.). New York, NY: McGraw-Hill.

49.

Rusbult

C. E.

Verette

Whitney

G. A.

Slovik

L. F.

Lipkus

(1991). Accommodation processes in close relationships: Theory and preliminary empirical evidence. Journal of Personality and Social Psychology, 60, 53-78.

50.

Simpson

J. A.

Kim

J. S.

Fillo

Ickes

Rholes

W. S.

Oriña

M. M.

Winterheld

H. A.

(2011). Attachment and the management of empathic accuracy in relationship-threatening situations. Personality and Social Psychology Bulletin, 37, 242-254.

51.

Simpson

J. A.

Oriña

M. M.

Ickes

(2003). When accuracy hurts, and when it helps: A test of the empathic accuracy model in marital interactions. Journal of Personality and Social Psychology, 85, 881-893.

52.

Starr

L. R.

Davila

(2008). Excessive reassurance seeking, depression, and interpersonal rejection: A meta-analytic review. Journal of Abnormal Psychology, 117, 762-775.

53.

Swann

W. B.

Wenzlaff

R. M.

Krull

D. S.

Pelham

B. W.

(1992). Allure of negative feedback: Self-verification strivings among depressed persons. Journal of Abnormal Psychology, 101, 293-305.

54.

Thomas

Fletcher

G. J. O.

Lange

(1997). On-line empathic accuracy in marital interaction. Journal of Personality and Social Psychology, 72, 839-850.

55.

Tooby

Cosmides

(1996). Friendship and the banker’s paradox: Other pathways to the evolution of adaptations for altruism. Proceedings of the British Academy, 88, 119-143.

56.

West

T. V.

Kenny

D. A.

(2011). The truth and bias model of judgment. Psychological Review, 118, 357-378.

57.

Yost

J. H.

Weary

(1996). Depression and the correspondent inference bias: Evidence for more effortful cognitive processing. Personality and Social Psychology Bulletin, 22, 192-200.