Affect and Personality as Predictors of Conflict and Closeness in Young Adults' Friendships

Affect and Personality as Predictors of Conflict and Closeness in Young Adults' Friendships

Journal of Research in Personality 34, 84–107 (2000) doi:10.1006/jrpe.1999.2271, available online at http://www.idealibrary.com on Affect and Persona...

86KB Sizes 0 Downloads 20 Views

Journal of Research in Personality 34, 84–107 (2000) doi:10.1006/jrpe.1999.2271, available online at http://www.idealibrary.com on

Affect and Personality as Predictors of Conflict and Closeness in Young Adults’ Friendships Diane S. Berry, Julie K. Willingham, and Christine A. Thayer Southern Methodist University We examined the relations of affect and personality to qualitative aspects of young adults’ friendships. Members of 131 friendship dyads evaluated the quality of their relationships and kept diaries of the conflicts they experienced during a 4week period. Positive and negative affect (PA and NA) emerged as independent predictors of the extent to which people felt close to their friends, of how much irritation they felt toward one another, and of the amount of conflict reported during the recording period. Extraversion, neuroticism, agreeableness, and openness also predicted measures of friendship quality. The data suggest that variations in affectivity are robust predictors of qualitative aspects of people’s social relationships.  2000 Academic Press

Affective experience is certainly influenced by external events. Consider, for example, the unfortunate individual who—in one wretched day—receives negative feedback about his job performance and learns that he is the target of an IRS audit. Anyone subjected to these events would probably feel tense and worried. On the other hand, imagine a person who—in one extraordinary day—receives a promotion and wins the lottery. Most people subjected to this latter sequence of events would feel happy and excited. Clearly, there are intraindividual differences in the levels of positive affect (PA) and negative affect (NA) that people experience. Despite the fact that significant life events influence self-reported affect, however, Suh, Diener, and Fujita (1996) report that the effects of such events are surprisingly fleeting, with most people returning to their emotional ‘‘baseline’’ within 3 to 6 months. In fact, impressive stability in NA and PA is revealed across time periods ranging from 2 to 7 years (Diener & Diener, 1996; Suh, Diener, & Fujita, 1996; Watson & Walker, 1996). Thus, NA and PA may be conceptualized as enduring interindividual differences, or traits. Moreover, numerous studies document that the dimensions of trait NA and

We thank Matt Ansfield and David Watson for comments on a draft of this manuscript. Address correspondence and reprint requests to Diane S. Berry, Department of Psychology, Southern Methodist University, Dallas, Texas, 75275. E-mail: dberry @mail.smu.edu. 84 0092-6566/00 $35.00 Copyright  2000 by Academic Press All rights of reproduction in any form reserved.

AFFECTIVITY, PERSONALITY, AND FRIENDSHIP

85

PA are relatively independent, with correlations typically in the ⫺.1 to ⫺.3 range (e.g., Diener & Emmons, 1984; Watson, 1988a). Systematic links exist between these dimensions of trait affect and the five-factor or ‘‘Big Five’’ model of personality—five broad dimensions that repeatedly emerge from analyses of people’s self-descriptions or ratings provided by their peers (i.e., extraversion, neuroticism, agreeableness, conscientiousness, and openness to experience; cf. McCrae & Costa, 1987). In particular, measures of trait NA are positively correlated with neuroticism (N), but unrelated to extraversion (E), whereas PA is positively correlated with extraversion, but unrelated to neuroticism (Costa & McCrae, 1982; Emmons & Diener, 1985; Schmutte & Ryff, 1997; Tellegen, 1985). The constructs of N and E, however, encompass a variety of behavioral and cognitive components in addition to a predisposition to experience particular affective states. Consequently, the more circumscribed dimensions of trait affect are not simply synonymous with neuroticism and extraversion, but instead constitute the ‘‘emotional core’’ of the broader factors (Watson & Clark, 1992). Although there is an extensive literature on the intrapersonal correlates of these stable individual differences, relatively few studies of the relations between trait affect and people’s interpersonal experiences exist. What data are available, however, suggest links between affect and social experience. For example, Watson and his colleagues asked college students to estimate the number of hours per day they spent with friends or participated in social activities. Positive affect was positively related to the amount of time people spent socializing (Watson, 1988b; Watson et al., 1992). The data revealed mixed results for NA. In some samples, high-NA participants reported more social engagement than did other undergraduates, whereas in other samples NA was unrelated to interaction frequency. More recently, people were asked to describe their social encounters in interaction diaries (Berry & Hansen, 1996; Study 2). Both NA and PA correlated positively with the amount of social contact reported. However, whereas PA predicted engagement in all types of interactions (e.g., opposite-sex interactions; group interactions), NA predicted only the number of same-sex dyadic interactions in which people took part. Studies reveal that trait affect predicts the quality as well as the frequency of people’s social experiences. For example, high-PA participants in the diary study just described characterized their interactions as more pleasant than did other people. In a related study, high-PA women who engaged in brief ‘‘get acquainted’’ conversations subsequently judged their interactions to be more enjoyable than did other women (Berry & Hansen, 1996, Study 1). Moreover, high-PA women’s partners enjoyed their interactions more than did partners of low-PA women. Independent observers who watched videotapes of the interactions also gave higher quality ratings to those involving high-PA dyads than to those featuring low-PA dyads. Given the social con-

86

BERRY, WILLINGHAM, AND THAYER

straints in operation during such brief, initial interactions, the fact that trait affect accounted for substantial variance in these encounters is impressive. Do these traits influence the quality of interactions that take place between people who know each other well, such as romantic partners or close friends? Again, there are surprisingly few investigations of the role of global individual differences in such relationships. Most work in this area focuses on relations between marital satisfaction and neuroticism, which is positively related to marital distress. For example, in a 40-year longitudinal study, spouses’ neuroticism levels at the time of marriage predicted later satisfaction and divorce (Kelly & Conley, 1987). Similarly, Karney and Bradbury (1997) observed a negative relation between neuroticism and newlyweds’ satisfaction. Although the remaining dimensions of the five-factor model have received less attention, research suggests that agreeableness and openness also predict relationship quality. For example, Shaver and Brennan (1992) observed a positive relation between openness and the longevity of college students’ dating relationships. Botwin, Buss, and Shackelford (1997) reported that measures of agreeableness and openness predict newlyweds’ satisfaction. Low levels of agreeableness and openness further predict spousal complaints about a number of negative behaviors presumably linked to dissatisfaction, such as acting in a condescending or inconsiderate manner (Buss, 1991). Although studies of the relations of the complete five-factor model to the nature of close relationships are rare, even fewer consider the role of trait affect in such relationships. Berry and Willingham (1997) reported that trait PA was positively related to relationship commitment and satisfaction in a sample of dating college students. More recently, Watson, Hubbard, and Weise (in press) found that NA and PA both emerged as predictors of satisfaction in a sample of couples married for an average of 17 years. Similar relations were observed in a sample of romantic partners who had been dating for an average of 18 months. Why might dispositional qualities such as trait affect predict satisfaction in close relationships? What, for example, do people high in PA do that makes their interactions with their partners especially satisfying? One possibility is that people who differ in trait affect also differ in how they respond to conflict. Although episodes of conflict occur less frequently than do neutral or positive exchanges among partners (Jacobsen, Follette, & Waggoner, 1982), negative interactions have a disproportionally potent influence on the quality of close relationships (e.g., Rook, 1984). Furthermore, Rusbult and her associates suggest that behaving in a constructive rather than a destructive manner during conflict is a critical determinant of relationship quality, and numerous studies support this view (e.g., Drigotas et al., 1995; Rusbult, Johnson, & Morrow, 1986). Rusbult and her colleagues suggest that responses to conflict (or ‘‘accom-

AFFECTIVITY, PERSONALITY, AND FRIENDSHIP

87

modation styles’’) can be conceptualized along two independent dimensions—active vs passive and positive vs negative—yielding a typology of four general classes of potential responses to dissatisfaction. These are labeled exit, voice, loyalty, and neglect (EVLN). ‘‘Exit’’ behaviors are destructive and active and include acts such as shouting at a partner or threatening to leave a relationship. Active, constructive behaviors are referred to as ‘‘voice’’ and might include discussing a problem or seeking counseling. Responses that are constructive in nature but passive—such as being patient and hoping for improvement—are referred to as instances of ‘‘loyalty.’’ Finally, responses that are simultaneously destructive and passive are instances of ‘‘neglect,’’ examples of which might include sulking or refusing to speak to a partner. The literature convincingly establishes that partners who engage in voice and shy away from exit and neglect tend to have healthier relationships than people who do not, whereas engagement in loyalty seems to have little effect on relationship quality, perhaps because such responses often go unnoticed (Drigotas et al., 1995). Who is most likely to display these behaviors? The literature on personality and coping suggests that high-PA individuals tend to engage in active, positive responses to stressors, whereas high-NA people tend to exhibit negative, emotion-based responses (Watson & Hubbard, 1996). Consistent with this, Berry and Willingham (1997) found that when undergraduates described their typical responses to conflict with a romantic partner, PA predicted engagement in voice and was negatively related to use of neglect. Negative affect predicted engagement in exit and neglect. To summarize, little is known about the relations of global personality dimensions such as trait affect to the nature of people’s close relationships or the potential mediators of such personality/relationship links. Moreover, as this brief review suggests, research in this area tends to focus on romantic relationships. However, dispositional variables may play an important role in other types of close relationships, such as friendships. Surprisingly, friendship processes only recently began to receive much attention in the close relationship literature in general, and very few studies examine predictors of friendship conflict or satisfaction (cf. Fehr, 1996, for a review). In the present study, we examined the relations of affectivity and personality to the quality of young adults’ friendships. To this end, pairs of friends provided evaluations of their relationship. In addition, participants kept daily diaries in which they recorded the frequency and seriousness of conflicts they experienced with their friends during a 4-week period. Based on data linking trait affect to satisfaction in romantic relationships (i.e., Berry & Willingham, 1997; Watson et al., in press), we expected PA to be positively related to friendship quality and NA to independently predict the extent to which these relationships were characterized by conflict and dissatisfaction. Work on personality and coping style (e.g., Watson & Hub-

88

BERRY, WILLINGHAM, AND THAYER

bard, 1996), as well as a previous study of romantic partners (Berry & Willingham, 1997), led us to expect PA to be associated with active, problemfocused responses to conflict (e.g., voice) and NA to be associated with more negative, emotion-focused coping strategies (e.g., neglect, exit). Based on the relations of accommodation style to satisfaction in romantic relationships (e.g., Drigotas et al., 1995), we further expected friendship quality to be negatively related to exit and neglect, positively related to voice, and unrelated to loyalty. Although our primary interest was in trait affect, we also examined the relations of the full five-factor model to these variables. Finally, we examined whether differences in accommodation style could explain any relations observed between affect, personality, and friendship quality. METHOD Participants Data were collected from 262 people. One hundred thirty-one undergraduates enrolled in psychology courses at Southern Methodist University (SMU) took part in return for one semester hour of pass/fail independent study credit. These individuals completed self-report measures and maintained daily records of the conflicts they experienced with their closest samesex friend who (a) lived in the metropolitan Dallas area and (b) was willing to take part in the study. The 131 friends, who comprised the remainder of the sample, completed self-report measures. (For clarity’s sake, we refer to the 131 students who maintained diaries as ‘‘participants’’ and to the remaining people who took part as ‘‘friends.’’) The average length of acquaintance between participants and their friends was 33.9 months, and their mean age was 21.6 years. Eighty-three percent of the friends were SMU students. Fifty-three of the friendship dyads were male and 78 were female.

Diary Procedure Participants met in small groups to receive instructions and materials. They were given daily logs and conflict diaries and asked to complete them before retiring each evening. The daily logs provided space for participants to indicate the amount of time spent with their friend on that day. If appropriate, participants were asked to complete a conflict diary (or diaries) after filling out their daily log. In the diaries, participants described ‘‘each occasion today during which you or your friend felt upset, unhappy, or dissatisfied about something concerning your relationship—something the other said, or did, something about the friendship itself, etc. Please complete a separate sheet for each such incident that may have occurred today’’ (adapted from Drigotas et al., 1995). Participants also evaluated the seriousness of the conflict on two 9-point scales: ‘‘How distressing/upsetting was the incident?’’ (not at all/very) and ‘‘How important is this incident with respect to your friendship?’’ (not at all/very). After receiving instructions, participants completed self-report items and were provided with a packet of self-report measures to give their friends. Accompanying instructions asked the friends to complete the packet while alone and return it to the Psychology Department. If these were not returned within 2 weeks, the friends were contacted and reminded to return the packet. Participants completed the 4-week recording procedure during either the fall of 1996 or the spring of 1997. Recording periods did not include the first or last week of classes, university

AFFECTIVITY, PERSONALITY, AND FRIENDSHIP

89

holidays, or finals. Participants returned completed logs and conflict diaries daily. Items that accumulated over each weekend were passed in on the following Monday. Careful records were kept of when items were received, and participants who failed to return materials on schedule were immediately contacted. In addition, an experimenter met weekly with participants to check their progress. Most people provided complete and usable data. Specifically, only 12 of 143 initial participants’ data were discarded due to their failure to follow directions, to return records in a timely manner, or the failure of their friends to return materials. This yielded the final sample of 131 dyads. After the diary procedure was complete, participants were contacted and asked about the accuracy of their records. One hundred three of the 131 participants provided these data. Specifically, they evaluated the accuracy of their reports of (a) the total amount of interaction they had with their friends and (b) the conflicts that occurred with their friends. These questions were answered on 9-point scales, with end points of ‘‘not at all accurate’’ and ‘‘very accurate.’’ Participants indicated high accuracy on both measures, M’s ⫽ 7.01 and 7.59, respectively.

Self-Report Measures Measures of affect and personality. The participants and their friends completed the PANAS (Positive and Negative Affect Schedule; Watson, Clark, & Tellegen, 1988), a self-report measure of trait positive and negative affect that has well-established psychometric properties, as well as the NEO-FFI (Costa & McCrae, 1992), a 60-item measure of the five-factor model that assesses extraversion, neuroticism, agreeableness, conscientiousness, and openness to experience. Friendship quality. People evaluated the quality of their friendships by responding to the following questions on 9-point scales (not at all/very much): ‘‘Compared to other friendships you’ve had, to what extent do you consider yourself to be close to your friend?’’; ‘‘Compared to other friends you’ve had, how important is this friendship to you?’’; ‘‘How much does your friend ‘get on your nerves’?’’; ‘‘Compared to other friendships, how much conflict do you have with your friend?;’’ and ‘‘How important to you is it that the two of you remain close friends?’’ Accommodation style. Participants and friends described their typical responses to dissatisfaction with their friends on a modified version of Rusbult’s Accommodation Scale (AS; cf. Rusbult et al., 1986). The original AS is a 28-item measure on which people evaluate their responses to dissatisfaction with a romantic partner. Seven different descriptions correspond to each of the four categories of exit (e.g., ‘‘When I’m really angry, I treat my partner badly [for example, by ignoring him/her and saying cruel things]’’), voice (e.g., When my partner hurts my feelings, I say nothing and simply forgive him or her’’), loyalty (e.g., ‘‘When we have troubles, no matter how bad things get, I am loyal to my partner’’), and neglect (e.g., ‘‘When I’m upset with my partner, I sulk rather than confront the issue’’). Previous work established that these scales are internally consistent (e.g., Rusbult et al., 1986). Moreover, individuals’ self-descriptions on these measures converge with descriptions of their behavior provided by their partners and do not correlate with measures of socially desirable responding (Rusbult et al., 1986). Because of our focus on friendships, we modified Rusbult’s original scale to specifically target people’s styles of dealing with problems with a close friend. To this end, a number of scales were simply reworded by substituting the word ‘‘friend’’ for ‘‘partner’’ and ‘‘friendship’’ for ‘‘relationship.’’ Two questions (one voice, one exit) seemed inappropriate for describing friendships, regardless of how they were reworded (‘‘When I’m not satisfied with our relationship, I consider dating other people’’ and ‘‘When we have problems in our relation-

90

BERRY, WILLINGHAM, AND THAYER

TABLE 1 Zero-Order Correlations between Participant and Friend Affect and Personality Scores Friend scores

Participant scores Positive Affect Negative Affect Extraversion Neuroticism Agreeableness Conscientiousness Openness

PA

NA

E

N

A

C

O

.17 ⫺.12 .30*** ⫺.13 .10 .22** .15

⫺.12 .18* ⫺.16 .21* ⫺.06 ⫺.27** ⫺.02

.07 ⫺.11 .26** .00 .18* .07 .09

⫺.10 .10 ⫺.15 .21* ⫺.02 ⫺.21* .02

.00 ⫺.25** .21* ⫺.07 .22* .12 .16

⫺.07 ⫺.10 .13 .08 .05 .10 ⫺.02

.10 ⫺.08 .09 .00 .26** .11 ⫹.37***

Note. Number of participants ⫽ 131; number of friends ⫽ 131. Similarity correlations appear in bold. * p ⬍ .05, ** p ⬍ .01, *** p ⬍ .001.

ship, I get help from someone else [friends, parents, counselor]). These items were eliminated. To maintain an equivalent number of items that measure each response style, a randomly chosen loyalty item and neglect item featured in the original scale were also eliminated. Thus, the revised AS (AS-R) contained 24 items, 6 of which tapped into each of the four response categories.

RESULTS AND DISCUSSION Predictor Variables: Positive and Negative Affect and the ‘‘Big Five’’ Do people become friends with individuals who are similar to them? Correlations between pairs of friends’ scores on the PANAS and NEO scales were significant for five of the seven dimensions (see Table 1). However, consistent with the literature on personality similarity or ‘‘assortative mating’’ among romantic partners (e.g., Buss, 1984, 1991), these positive correlations were low to moderate in magnitude, mean r(129) ⫽ .22. Thus, the data do not suggest that people befriend individuals who are highly similar to them in terms of personality and affectivity. Outcome Variables: Measures of Friendship Quality Global evaluations of quality: Closeness and irritation. Recall that participants and their friends evaluated the quality of their friendships on five items described previously. For purposes of data reduction, we performed varimax rotations separately on the participants’ and friends’ responses. Both analyses revealed very similar patterns of data. As seen in Table 2, three items (‘‘Compared to other friendships you’ve had, to what extent do you con-

AFFECTIVITY, PERSONALITY, AND FRIENDSHIP

91

TABLE 2 Factor Loadings from Varimax Rotation Analyses of Evaluations of Friendship Quality How close? How important is friendship? How important remain close? Get on nerves? How much conflict?

Participants (n ⫽ 131) ⫹.91 ⫺.01 ⫹.93 ⫺.17 ⫹.74 ⫺.18 ⫺.18 ⫹.92 ⫺.08 ⫹.94

How close? How important is friendship? How important remain close? Get on nerves? How much conflict?

Friends (n ⫽ 131) ⫹.90 ⫺.05 ⫹.88 ⫺.09 ⫹.71 ⫺.22 ⫺.15 ⫹.91 ⫺.12 ⫹.90

Note. Factor loadings greater than .7 appear in bold.

sider yourself to be close to your friend?’’; ‘‘Compared to other friendships you’ve had, how important is this friendship to you?’’; and ‘‘How important to you is it that the two of you remain close friends?’’) loaded on one factor labeled ‘‘closeness.’’ The remaining items (‘‘How much does your friend ‘get on your nerves’?’’; and ‘‘Compared to other friendships, how much conflict do you have with your friend?’’) loaded on a second factor, labeled ‘‘irritation.’’ Two composite variables were created by computing the mean of peoples’ responses to items associated with each factor. Alphas for the closeness and conflict scales were .82 and .84, respectively. Analyses revealed significant but low, negative correlations between these composite variables, r’s (129) ⫽ ⫺.27 and ⫺.31 for participants and friends respectively, p’s ⬍ .05. As participants had been instructed to ask a ‘‘close’’ friend to take part, it is not especially surprising that people, on average, reported feeling close to their friends. Specifically, the closeness composite scores ranged from 3.7 to 9, M ⫽ 6.85. Irritation composite scores ranged from 1 to 9, M ⫽ 2.88. (All items were assessed on 1 through 9 scales.) Because these two variables were not normally distributed, we performed a logarithmic transformation on each before conducting additional analyses. Diary measures. On average, participants interacted with their friends on 19.7 of the 28 days described. The average amount of contact between friends during the 4-week period was 68.5 h. As amount of time spent together was unrelated to affect, personality, or friendship quality, it is not discussed further. Seventy-three of the participants did not experience conflict with their

92

BERRY, WILLINGHAM, AND THAYER

friends. The remaining 58 dyads reported from one to seven episodes of conflict, M ⫽ 2.10. Again, a logarithmic transformation was performed on this variable before conducting additional analyses. Recall that participants who reported conflicts also evaluated the seriousness of each on two 9-point scales described previously. Fifty-six of the participants who reported conflicts completed these scales. Because responses to these questions were correlated, r(54) ⫽ ⫹.43, p ⬍ .001, we created a composite ‘‘seriousness’’ measure by computing the mean of the two original items. These were then averaged across a given participant’s conflicts. These values were normally distributed, M ⫽ 4.43, range ⫽ 1 to 8. Relations among measures of friendship quality. Zero-order correlations among the outcome measures appear in Table 3. Pairs of friends agreed in their ratings of how close they were and how irritated they were by one another, r’s (129) ⫽ .63 and .32, respectively, p’s ⬍ .001. Although significant agreement between pairs’ of friends evaluations along both dimensions was observed, a significantly stronger relation existed between mutual assessments of closeness than of irritation, p ⬍ .05, Fisher z test. Previous work indicates that when friends experience feelings of dissatisfaction, they are more likely to avoid the issue than directly confront it (Davis & Todd, 1982; Helgeson, Shaver, & Dyer, 1987; Stiller, 1980). Thus, consistent with our data, overt episodes of conflict tend to be relatively infrequent (Canary, Cupach, & Messman, 1995; Fehr, 1996). The reluctance to express negative feelings in friendships likely makes it difficult for people to recognize when they annoy or irritate friends, yielding the somewhat lower agreement observed for the irritation measure. The number of conflicts reported during the recording period was also unrelated to either participants’ or friends’ irritation ratings, a result also consistent with a tendency to avoid expressing annoyance or anger toward friends. Consequently, the absence of discrete instances of conflict between friends should not be equated with the absence of subjective feelings of irritation or annoyance. Among dyads that do engage in overt conflicts, however, it seems that the nature and frequency of those episodes should influence people’s evaluations of the relationship. In fact, among dyads who reported episodes of conflict, a composite measure of number and seriousness of conflicts predicted participants’ global irritation ratings, r(54) ⫽ ⫹.27, p ⬍ .05, and was marginally related to friends’ irritation ratings, r(54) ⫽ ⫹.22, p ⬍ .08. Potential Mediators: Measures of Accommodation Style Recall that people completed the AS-R, a revised accommodation scale. This modified version of Rusbult’s original measure focused on interactions with friends rather than romantic partners. For each individual, we summed the six reworded items that reflected exit behaviors to create an exit scale.

Participant Closeness Participant Irritation Friend Closeness Friend Irritation N of Conflict Seriousness of Conflict

1.0 — — — — —

Friend closeness ⫹.63*** ⫺.18* 1.0 — — —

Participant irritation ⫺.27** 1.0 — — — —

⫺.32*** ⫹.32*** ⫺.31*** 1.0 — —

Friend irritation ⫺.04 ⫹.18 ⫹.01 ⫺.03 1.0 —

N of conflict ⫹.09 ⫹.19 ⫺.04 ⫹.14 ⫹.12 1.0

Seriousness of conflict

Note. For analyses of participant and friend self-reports N ⫽ 131; for analyses of number of conflicts N ⫽ 58. Analyses of conflict seriousness are based on an N of 56.

1. 2. 3. 4. 5. 6.

Participant closeness

TABLE 3 Zero-Order Correlations among Friendship Outcome Measures

AFFECTIVITY, PERSONALITY, AND FRIENDSHIP

93

94

BERRY, WILLINGHAM, AND THAYER

Parallel procedures were followed to create voice, loyalty, and neglect scales. Alphas for the four scales ranged from .81 to .89.1 Overview of Analyses Our first question was whether trait affect predicts the quality of people’s friendships. Next, we considered whether people who vary in affectivity respond differently to dissatisfaction with their friends and whether such response styles predict friendship quality. We further examined whether affectrelated variations in accommodation style could account for any relations observed between NA, PA, and measures of quality. Finally, we present parallel analyses of relations between the dimensions of the five-factor model, the nature of people’s friendships, and their accommodation styles. Although our discussion focuses on the results of regression analyses, zeroorder correlations among all affect, personality, accommodation, and outcome measures are displayed in Tables 4 and 5. Friendship and Affect Does affectivity predict friendship quality? First, we considered whether dyads that engaged in conflict differed in affectivity from those who did not. To address this, we compared the mean affect scores observed for both groups of dyads. Significant differences between the no-conflict and conflict dyads were revealed for levels of participant PA, t(129) ⫽ 2.60, p ⬍ .01, and friend NA, t(129) ⫽ ⫺2.90, p ⬍ .004. Specifically, participants’ who did not report engaging in conflict were higher in positive affect than were those who reported conflicts with their friends (M’s ⫽ 38.3 vs 35.2). Participants who reported conflict also had friends who were higher in negative affect than participants who reported no conflicts (M’s ⫽ 23.6 vs 19.7). Levels of participant NA and friend PA did not differ between the two groups, t’s ⬍ 1.64, p’s ⬎ .05. Next, we examined the relations of NA and PA to conflict frequency. Only data from those 58 dyads who engaged in conflict were included in this analysis. A regression equation was calculated in which participant PA (PPA) and NA (PNA) and friend PA (FPA) and NA (FNA) simultaneously predicted the number of conflicts in which these dyads engaged (Standardized betas from these equations appear in Table 6). PPA and FNA each emerged as independent predictors of conflict frequency. (All relations dis1 We did have access to people’s scores on the original scale and compared these with the AS-R scores. Significant correlations were found between the two assessments of each response style. However these values were moderate in size, r’s(260) ⫽ .26, .29, .41, and .48 for exit, loyalty, voice and neglect, respectively, p’s ⬍ .001, mean r ⫽ .36. Thus, although the measures are related, they seem to tap into somewhat different constructs. This suggests that a given individual may respond differently to problems with a friend versus a romantic partner.

Extraversion Neuroticism Agreeableness Conscientiousness Openness

Exit Voice Loyalty Neglect

⫹.14 ⫺.12 ⫺.03 ⫹.23** ⫹.24** ⫺.06 ⫺.03 ⫹.23**

⫺.45*** ⫹.29** ⫺.07 ⫺.32*** ⫺.29*** ⫹.11 ⫺.07 ⫺.32***

⫺.15 ⫹.29*** ⫺.15 ⫺.18* ⫺.03

⫹.16 ⫹.01 ⫹.13 ⫹.15 ⫹.01

⫺.42*** ⫹.22* ⫹.06 ⫺.27**

⫺.32*** ⫹.22* ⫹.06 ⫺.27**

⫹.24** ⫺.04 ⫹.19* ⫹.18* ⫹.08

⫹.12 ⫺.02 ⫹.11 ⫹.03 ⫹.01

⫹.06 ⫺.13

⫹.11 ⫺.22**

Friend closeness

⫹.33*** ⫺.24* ⫺.03 ⫹.25**

⫹.19* ⫺.23** ⫺.03 ⫹.25**

⫺.17* ⫹.23** ⫺.19* ⫺.23** ⫺.09

⫺.09 ⫹.07 ⫺.29*** ⫺.22** ⫺.19*

⫺.23** ⫹.24**

⫺.15 ⫹.13

Friend irritation

⫹.03 ⫺.21 ⫺.11 ⫹.26*

⫺.19 ⫺.22 ⫹.08 ⫹.07

⫺.10 ⫹.35** ⫹.02 ⫺.07 ⫹.01

⫺.29* ⫹.18 ⫹.05 ⫺.27* ⫺.27*

⫺.09 ⫹.38**

⫺.32** ⫹.17

N of conflict

⫺.03 ⫺.12 ⫺.05 ⫹.02

⫺.09 ⫺.09 ⫹.20 ⫹.16

⫺.08 ⫹.09 ⫹.08 ⫹.15 ⫹.16

⫺.01 ⫹.11 ⫹.11 ⫹.15 ⫹.08

⫹.08 ⫹.14

⫹.07 ⫺.09

Seriousness of conflicts

Note. For analyses of participant and friend self-reports N ⫽ 131; for analyses of number of conflicts N ⫽ 58. Analyses of conflict seriousness are based on an N of 56.

Friend Friend Friend Friend

Accommodation measures Participant Exit Participant Voice Participant Loyalty Participant Neglect

Friend Friend Friend Friend Friend

⫺.14 ⫹.11 ⫺.11 ⫺.28*** ⫺.12

⫹.25** ⫺.02 ⫹.07 ⫹.10 ⫹.03

⫺.16 ⫹.30***

⫹.07 ⫺.13

Friend PA Friend NA

Personality measures Participant Extraversion Participant Neuroticism Participant Agreeableness Participant Conscientiousness Participant Openness

⫺.36*** ⫹.13

⫹.18* ⫺.11

Participant irritation

Affect measures Participant PA Participant NA

Participant closeness

TABLE 4 Zero-Order Correlations between Friendship Outcome Measures and Affect, Personality, and Accommodation Measures

AFFECTIVITY, PERSONALITY, AND FRIENDSHIP

95

Friend Friend Friend Friend Friend

Extraversion Neuroticism Agreeableness Conscientiousness Openness

⫹.32*** ⫺.04 ⫹.11 ⫹.07 ⫹.26** ⫹.07 ⫹.05 ⫹.16 ⫹.06 ⫹.25**

⫺.28*** ⫺.01 ⫺.26** ⫺.01 ⫺.06 ⫺.19* ⫺.11 ⫺.32*** ⫺.14 ⫺.17

⫹.10 ⫹.02

⫺.03 ⫹.06

Friend PA Friend NA

Personality measures Participant Extraversion Participant Neuroticism Participant Agreeableness Participant Conscientiousness Participant Openness

⫹.22** ⫺.09

⫺.11 ⫹.19*

Voice

Affect measures Participant PA Participant NA

Exit

⫺.02 ⫹.02 ⫹.05 ⫺.03 ⫹.19*

⫹.07 ⫹.09 ⫹.25** ⫹.01 ⫹.21*

⫹.11 ⫺.02

⫹.07 ⫹.01

Loyal

Participant (N ⫽ 131)

⫺.19* ⫺.08 ⫺.18* ⫺.10 ⫺.12

⫺.26** ⫹.21* ⫺.16 ⫺.07 ⫺.23**

.01 ⫺.04

⫺.21* ⫹.26**

Neglect

⫺.22* ⫹.14 ⫺.37*** ⫺.29*** ⫺.16

⫺.12 ⫺.15 ⫺.10 ⫺.16 ⫺.11

⫺.11 ⫹.16

⫺.13 ⫹.08

Exit

⫹.33*** ⫺.15 ⫹.22* ⫹.07 ⫹.23**

⫹.21* ⫺.15 ⫹.01 ⫹.13 ⫹.07

⫹.09 ⫺.11

⫹.23** ⫺.09

Voice

⫹.13 ⫺.08 ⫹.27** ⫹.23** ⫹.03

⫹.02 ⫹.20* ⫹.25** ⫹.08 ⫹.11

⫹.14 ⫺.07

⫺.02 ⫺.06

Loyal

Friend (N ⫽ 131)

TABLE 5 Zero-Order Correlations between Affect and Personality Measures and Accommodation Scores

⫺.18* ⫹.21* ⫺.25** ⫺.07 ⫺.25**

⫺.25** ⫹.14 ⫺.07 ⫺.20* ⫺.09

⫺.06 ⫹.24**

⫺.25** ⫹.21*

Neglect

96 BERRY, WILLINGHAM, AND THAYER

97

AFFECTIVITY, PERSONALITY, AND FRIENDSHIP

TABLE 6 Standardized Betas from Analyses Predicting Friendship Outcome Measures from Trait Affect Self-report

Predictor variable Participant PA Participant NA Friend PA Friend NA

Participant closeness

Participant irritation

Friend closeness

Friend irritation

⫹.17* ⫺.03 ⫺.01 ⫺.11

⫺.33*** ⫺.05 ⫺.02 ⫹.25**

⫹.02 ⫺.20* ⫹.01 ⫺.09

⫺.02 ⫹.06 ⫺.14 ⫹.18*

Diary measures

Predictor variable Participant PA Participant NA Friend PA Friend NA

N of conflict

Seriousness of conflict

⫺.31* ⫺.07 ⫹.07 ⫹.38**

⫹.04 ⫺.12 ⫹.15 ⫹.24

Note. For analyses of participant and friend self-reports N ⫽ 131; for analyses of number of conflicts N ⫽ 58. Analyses of conflict seriousness are based on an N of 56.

cussed are significant at p ⬍ .05.) Specifically, PPA was negatively related to and FNA positively related to the number of conflicts in which the dyads engaged. A parallel regression equation was calculated to examine whether variations in affect predicted the seriousness of these conflicts. As can be seen in Table 6, neither participant or friend affect scores predicted conflict seriousness. Finally, a series of regression equations was calculated in which participant and friend PA and NA were simultaneously entered as predictors of participants’ and their friends’ closeness and irritation ratings (see Table 6). Data from all dyads were included in these analyses. High-PA participants reported feeling closer to and less irritated by their friends than did other people. Participants’ NA levels were negatively related to how close their friends felt to them. Friends’ NA scores further predicted their own and their partners’ feelings of irritation. In sum, PA bore positive relations to measures of friendship quality, and NA independently predicted conflict and irritation. Although no comparable studies of friendship dyads are available, these data are consistent with relations documented between trait affect and the quality of other kinds of relationships. For example, PA is positively related to the satisfaction of dating

98

BERRY, WILLINGHAM, AND THAYER

and married couples (Berry & Willingham, 1997; Watson, Hubbard, & Weise, in press) and to interaction quality in unacquainted dyads (Berry & Hansen, 1996; in press), whereas NA is negatively related to relationship satisfaction (Watson et al., in press). Note that links between affectivity and friendship quality could result from at least two sources. One possibility is that people who vary in trait affect also vary in how they behave toward their friends, creating actual differences in the nature of their interactions and relationships. A second possibility is that although the social relations of people who vary in NA or PA might not differ in any ‘‘objective’’ sense, individuals who differ in trait affect could perceive or interpret their experiences differently. For example, if high-PA people tend to view their relationships ‘‘through rose-colored glasses,’’ they may describe them as being close and conflict free, even if that is not really the case. These processes are both interesting and important, and they are by no means mutually exclusive; both may contribute to the links we observed between affect and experience. On the basis of our data, we believe that the relations revealed between affect and relationship quality cannot be entirely explained by differences in how people who vary in trait affect evaluate their relationships. Several pieces of evidence support this conclusion. First, there is good reason to believe that our outcome measures reflected more than idiosyncratic, subjective views of these relationships. There was significant agreement between dyad members’ evaluations of the amount of irritation in their relationships and substantial agreement between ratings of closeness. Moreover, as correlations between dyad members’ levels of NA and PA were very low (.18 and .17, respectively), similarity in assessments of relationship quality cannot be due to similarity in dyad members’ levels of affect. In addition, trait affect predicted participants’ daily records of the number of conflicts in which they and their friends engaged, and dyads who did not engage in conflict differed in affectivity from those who did. Although we do not claim that these records are purely objective—they do represent the participants’ views—these measures seem more ‘‘concrete’’ than global self-evaluations and less susceptible to subjective interpretation. Additionally, people’s levels of NA and PA in many cases influenced not only their own evaluations of their friendships, but also their partners’ views. For example, high-NA friends reported greater levels of irritation than did other people. Moreover, friends’ NA scores predicted participants’ self-described irritation as well as the number of conflicts participants reported having with them. Thus, people’s levels of trait affect can influence not only their own perceptions of their social experiences, but also the perceptions of those people with whom they interact. In sum, individual differences in trait affect predict friendship quality. Could differences in accommodation style mediate the links revealed between affect and the outcome measures? To address this possibility, we fol-

AFFECTIVITY, PERSONALITY, AND FRIENDSHIP

99

lowed the recommendations of Baron and Kenny (1986) for testing mediation. Given a relation between a predictor and outcome variable, they propose the following four criterion be met before a variable can be identified as a mediator: (a) first, that the potential mediator variables (i.e., accommodation scores) are related to the predictor variable (i.e., NA or PA scores); (b) second, that the potential mediators (i.e., accommodation scores) are related to the outcome variable (i.e., friendship quality); (c) third, that the potential mediator variables continue to predict the outcome measure when that measure is simultaneously regressed on the mediators and the predictor variables; and (d) fourth, that the relation between the predictor and the outcome measure no longer holds in the equation described in c. According to Baron and Kenny, if these four conditions are met, mediation may be inferred. Does affectivity predict accommodation style? To address criteria a, we computed correlations between NA, PA, and people’s AS-R scores (see Table 5). Recall that Berry and Willingham (1997) examined the relations between trait affect and accommodation style in romantic relationships. They found that self-reported propensities to engage in exit and neglect were negatively related to PA and positively related to NA, voice was positively related to PA and negatively related to NA, and loyalty was unrelated to affect. Although all significant relations revealed between affect and accommodation in the present sample were consistent with that pattern, affectivity was not a particularly impressive predictor of accommodation in friendships. Specifically, although participants’ PA scores predicted their use of voice and neglect, friends’ PA scores were unrelated to their accommodation styles. Negative affect was further related to self-reported exit and neglect for participants, but only to neglect for the sample of friends. In addition to an individual’s own personality influencing his willingness to accommodate, it is conceivable that his friend’s personality could influence his responses. Thus, we considered the relations of people’s levels of trait affect to the self-described behaviors of their partners (see Table 5). Several significant relations were revealed. In particular, participants’ NA scores were positively related to the extent to which their friends reported engaging in neglect. Participants’ PA scores further predicted their friends’ use of voice and neglect. In sum, the present data reveal a somewhat less clear pattern of relations between affect and accommodation style than reported by Berry and Willingham (1997). However, participants in the previous study specifically described their ways of responding to romantic partners, whereas people in the current study specifically described how they behaved with friends. This may suggest that accommodation in friendships is less tied to affect than to other factors. Nevertheless, several relations between affect and accommodation were revealed, and the direction of those relations was consistent Berry and Willingham’s data.

100

BERRY, WILLINGHAM, AND THAYER

Accommodation style and friendship quality. Criteria b, described by Baron and Kenny, necessitates the examination of the relations between accommodation style and the measures of friendship quality. Beyond identifying potential mediator variables, these relations are quite interesting in of themselves. As discussed previously, it has been repeatedly shown that the tendency to respond with voice to dissatisfying situations or events in romantic relationships promotes relationship health and satisfaction, whereas engagement in exit and neglect leads to more negative outcomes. To our knowledge, no studies consider the role of these response styles in nonromantic, close relationships. The data displayed in Table 4 reveal a very similar pattern to that documented in romantic relationships. In general, exit and neglect scores tended to be negatively related to participants’ and friends’ perceptions of closeness and positively related to reported irritation. Voice tended to be positively related to closeness and negatively related to irritation. Friends’ propensities to engage in neglect responses predicted the number of conflicts experienced by the dyads. Finally, consistent with previous work, loyalty was unrelated to friendship quality. Tests of mediation: Affectivity and friendship quality. To summarize, a number of relations were revealed between trait affect, friendship quality, and measures of accommodation style. We computed a series of regression equations to assess whether any of the accommodation measures that satisfied criteria a and b for a particular affect/outcome link could further satisfy the final two criteria described by Baron and Kenny. Potential mediator variables that met criteria a and b were revealed for two of the previously described relations between affect and friendship quality: participant voice and friend neglect scores qualified as potential mediators of the relation between participants’ PA levels and their closeness evaluations, whereas participant exit and neglect and friend neglect scores qualified as potential mediators of the relation between participants’ NA levels and their friends’ closeness ratings (see Tables 4 and 5). First, we considered possible mediators of the relation between PPA and participant closeness ratings. A regression analysis was conducted in which PPA, participant voice, and friend neglect simultaneously predicted participant closeness ratings. An examination of the betas revealed that friend neglect scores emerged as the only significant predictor of the criterion variable. PPA no longer predicted the outcome variable in this analysis. Thus, all criteria for mediation were met: Friends of high-PA participants’ were less likely to engage in neglect than were the friends of other participants, and this propensity accounted for the link between PPA and participants’ feelings of closeness. Recall that participant exit and neglect scores and friend neglect scores predicted both PNA and friend closeness ratings. A second regression analy-

AFFECTIVITY, PERSONALITY, AND FRIENDSHIP

101

sis was conducted in which PNA, participant exit, participant neglect, and friend neglect scores were entered as predictors of friend closeness ratings. Participant exit and friend neglect each emerged as significant predictors of the criterion variable, and PNA no longer predicted friend closeness ratings in this analysis. Thus, high-NA participants’ propensities to engage in exit, and their friends’ tendencies to engage in neglect can explain the link between PNA and friends’ closeness ratings.2 Friendship and the Five-Factor Model Do dimensions of the five-factor model predict friendship quality? Paralleling the analyses described previously, we first conducted t tests to compare scores on the five-factor model obtained by individuals in dyads that reported conflict with those of people who did not experience conflict. Only one significant difference was observed; participants in dyads who did not experience conflict were higher in openness (M ⫽ 40.8) than participants in dyads who reported conflict, M ⫽ 38.3, t(129) ⫽ 2.10, p ⬍ .04. No other dimensions differed between groups, all t’s ⬍ 1.27, all p’s ⬎ .05. A series of regression analyses was conducted in which friends’ and participants’ extraversion, neuroticism, agreeableness, conscientiousness, and openness scores simultaneously predicted each measure of friendship quality. Standardized beta weights from these analyses appear in Table 7, and corresponding zero-order correlations appear in Table 4. As before, analyses predicting conflict frequency and seriousness were based on only those individuals who engaged in conflict, whereas analyses of closeness and irritation were based on members of all dyads. Extraverted participants felt closer to their friends than did other people. Similarly, friends high in extraversion gave higher closeness ratings to their friendships than did other people. Friends of agreeable participants reported less irritation than did friends of less agreeable participants. Finally, participants’ levels of openness bore a negative relation to the number of conflicts reported, and friends’ neuroticism scores were positively related to conflict frequency. Dimensions of the five-factor model and accommodation style. Table 5 reveals a number of reliable relations between people’s personality scores and their self-described responses to dissatisfaction in friendships. Moreover, several of these replicated across the two samples of subjects (i.e., r’s be2

Berry and Willingham (1997) examined whether people’s accommodation styles mediated the relations between their levels of trait affect and the quality of their romantic relationships. Because the outcome variables in that study differed from those in the present data set, it is difficult to make direct comparisons between the two studies. Berry and Willingham did find that people’s self-reported tendencies to engage in exit mediated the relationship between PA and whether they were currently in a relationship. For people who were involved in a romantic relationship, exit and voice were identified as mediators of the relation observed between PA and relationship commitment.

Extraversion Neuroticism Agreeableness Conscientiousness Openness

⫺.01 ⫺.04 ⫺.07 ⫺.22 ⫺.10 ⫺.01 ⫹.22 ⫺.02 ⫺.04 ⫹.04

⫹.22* ⫹.02 ⫹.01 ⫹.07 ⫺.04 ⫹.12 ⫹.19 ⫹.04 ⫹.15 ⫺.01

Participant irritation

⫹.20* ⫹.13 ⫹.05 ⫹.16 ⫹.08

⫹.04 ⫺.02 ⫹.03 ⫹.01 ⫺.07

Friend closeness

⫺.03 ⫹.13 ⫺.02 ⫺.13 ⫹.03

⫹.07 ⫺.03 ⫺.24** ⫺.14 ⫺.16

Friend irritation

⫹.04 ⫹.37* ⫹.10 ⫹.05 ⫹.15

⫺.22 ⫺.04 ⫹.14 ⫺.18 ⫺.37***

N of conflicts

⫺.13 ⫹.20 ⫹.08 ⫹.23 ⫹.17

⫹.05 ⫹.17 ⫹.09 ⫹.24 ⫺.01

Seriousness of conflicts

Note. For analyses of participant and friend self-reports N ⫽ 131; for analyses of number of conflicts N ⫽ 58. Analyses of conflict seriousness are based on an N of 56.

Friend Friend Friend Friend Friend

Predictor variable Participant Extraversion Participant Neuroticism Participant Agreeableness Participant Conscientiousness Participant Openness

Participant closeness

TABLE 7 Standardized Betas from Analyses Predicting Relationship Outcome Measures from the Five-Factor Model

102 BERRY, WILLINGHAM, AND THAYER

AFFECTIVITY, PERSONALITY, AND FRIENDSHIP

103

tween participants’ personality and accommodation scores and r’s between friends’ personality and accommodation scores). In particular, extraversion predicted self-reported use of exit, voice, and neglect in both samples. Neuroticism predicted neglect, agreeableness predicted use of exit and loyalty, and openness was related to engagement in voice and neglect; each of these relations was also observed in both samples. Several relations were revealed between people’s personality scores and their partners’ accommodation styles as well (see Table 5 for the full correlation matrix). Of particular note, a relation between people’s use of neglect and their partners’ extraversion scores was observed for both participants and friends. Tests of mediation: Dimensions of the five-factor model and friendship quality. We again followed Baron and Kenny’s recommendations to identify potential mediators of the five relations previously described between people’s scores on the NEO and qualitative aspects of their friendships. We first examined the correlations between friend and participant accommodation scores, friendship outcome measures, and personality scores to identify any potential mediators that satisfied criteria a and b described previously for each personality/outcome link. We then conducted regression analyses to test whether any of these potential mediators satisfied criteria c and d. Two fulfilled all four criteria. First, the reluctance of extraverted participants to engage in exit appears to account for their feelings of closeness to their friends. Parallel results were revealed for friend’s levels of extraversion and friends’ feelings of closeness. Specifically, extraverted friends’ unwillingness to engage in exit accounted for the relation between their levels of extraversion and closeness ratings. Thus, a number of relations were revealed between dimensions of the five-factor model and accommodation styles in friendships. However, variations in accommodation style accounted for only two of the personality/outcome links observed. SUMMARY AND CONCLUSIONS Considerable evidence reveals stable individual differences in people’s propensities to experience positive and negative emotional states. What are the interpersonal consequences of variations in affect? Consistent with our expectations, NA and PA emerged as independent predictors of the extent to which people felt close to, and were irritated by, their friends. Levels of trait affect further differentiated dyads who engaged in conflict from those who did not. Specifically, people who did not report conflicts with their friends during a 4-week period were higher in positive affect than were those who did engage in conflict. Moreover, people who reported conflict had friends who were higher in negative affect than those who did not report conflict. Among those dyads who experienced conflict, trait affect further predicted the frequency with which conflict occurred during a 4-week period. Finally, note that levels of conflict observed in this sample tended to be fairly

104

BERRY, WILLINGHAM, AND THAYER

low and closeness levels high. Given this restricted variability, the fact that NA and PA accounted for significant proportions of the variance in these measures is especially impressive. We also examined relations between dimensions of the five-factor model and friendship quality. As noted previously, a number of studies reveal links between neuroticism and satisfaction in marital relationships. In the present study, friends’ neuroticism scores predicted the number of conflicts in which dyads engaged. People’s extraversion scores also predicted the extent to which they felt close to their friends. The friends of participants’ high in agreeableness reported less irritation than did other friends, and participant openness was positively correlated with the number of conflicts they reported. We suggest that the relations revealed between affect, personality, and friendship quality stem from objective individual differences in the nature of people’s interactions. However, some plausible alternative interpretations warrant mention. First, the direction of the causal arrows between affect and relationship quality may be the reverse of what we suggest. For example, being in a satisfying friendship may increase PA, and experiencing conflict with a friend may heighten NA. However, given the rather impressive stability in affect that has been observed across time (e.g., Watson & Walker, 1996) and the resilience of people’s baseline levels of affect in the wake of various life events (e.g., Suh et al, 1996), we doubt that this is the primary source of relations between trait affect and our outcome measures. As mentioned, a second possibility is that people’s dispositional qualities color their perceptions of their interactions and relationships in a way that may or may not reflect ‘‘true’’ differences in quality. For example, if highPA people report being closer to their friends than do other individuals, is that because their relationships are objectively ‘‘better’’ than others or simply because they interpret events in a more positive light than other people? We tend to favor the former interpretation for the following reasons. First, the fact that substantial agreement was found between dyad members’ evaluations of their friendship indicates that these judgments reflect something more than people’s subjective interpretations of reality. Second, participants’ diary measures would seem more resistant to interpretative biases than global evaluations of satisfaction. Finally, people’s levels of trait affect often predicted their friend’s as well as their own evaluations. In light of this, we suggest that the links revealed between affect and the outcome measures cannot be entirely explained by differences in the way that people who vary in trait affect interpret their own experiences. If this is correct, exactly what do people who vary in trait affect do that yields differences in their relationships? We attempted to address this by testing whether variations in accommodation style mediate links observed between affect, personality, and friendship quality. Relatively few of the

AFFECTIVITY, PERSONALITY, AND FRIENDSHIP

105

relations revealed between affectivity and friendship quality could be explained by these variables. This reflected the fact that although people’s propensities to engage in exit, voice, and neglect were good predictors of friendship quality, affect was not consistently related to people’s accommodation styles. This is somewhat surprising, as previous work linked NA and PA to tendencies to engage in exit, voice, and neglect with romantic partners (Berry & Willingham, 1997). Additional research will be needed to evaluate whether this pattern of results means that affect is a better predictor of responses to dissatisfaction with romantic partners than friends or whether it reflects some idiosyncratic difference between the samples studied. A number of consistent links were observed between accommodation style and dimensions of the five-factor model. Once again, however, the propensities to engage in exit, voice, loyalty, or neglect mediated relatively few of the links observed between personality and friendship quality. Although the categories of the EVLN model are excellent predictors of general relationship satisfaction, it may be that more fine-grained behavioral taxonomies will be more successful in identifying the source of links between affectivity, personality, and friendship quality. Nevertheless, the data clearly indicate that variations in accommodation style are good predictors of the quality of close friendships, a finding that complements and extends recent work on accommodation in romantic relationships. Researchers have paid limited attention to the role that global personality characteristics play in channeling the course of close relationships (Clark & Reis, 1998). Recent work on this topic suggests that individual differences in personality and affect may be key determinants of the extent to which one’s interpersonal experiences are rewarding and satisfying. In particular, the present work adds to a growing literature that suggests that variations in trait affect predict the nature of people’s relationships, ranging from encounters with strangers (Berry & Hansen, 1996, in press), to long-term relationships with romantic partners (Berry & Willingham, 1997; Watson et al., in press), to—as our present data suggest—relationships with close friends. REFERENCES Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 61, 298–307. Berry, D. S., & Hansen, J. S. (1996). Positive affect, negative affect, and social interaction. Journal of Personality and Social Psychology, 71, 796–809. Berry, D. S., & Hansen, J. S. (in press). Personality, nonverbal behavior, and interaction quality in female dyads. Personality and Social Psychology Bulletin. Berry, D. S., & Willingham, J. K. (1997). Affective traits, responses to conflict, and satisfaction in romantic relationships. Journal of Research in Personality, 31, 564–576. Botwin, M. D., Buss, D. M., & Shackelford, T. K. (1997). Personality and mate preferences:

106

BERRY, WILLINGHAM, AND THAYER

Five factors in mate selection and marital satisfaction. Journal of Personality, 65, 107– 136. Buss, D. M. (1984). Marital assortment for personality dispositions: Assessment with three data sources. Behavior Genetics, 14, 111–123. Buss, D. M. (1991). Conflict in married couples: Personality predictors of anger and upset. Journal of Personality, 59, 663–688. Canary, D. J., Cupach, W. R., & Messman, S. J. (1995). Relationship conflict. Thousand Oaks, CA: Sage. Clark, M. S., & Reis, H. T. (1988). Interpersonal processes in close relationships. Annual Review of Psychology, 39, 609–672. Costa, P. T., Jr., & McCrae, R. R. (1992). Manual for the Revised NEO Personality Inventory and the NEO Five-Factor Inventory. Odessa, FL: Psychological Assessment Resources. Davis, K. E., & Todd, M. J. (1982). Friendship and love relationships. In K. E. Davis & T. Mitchell (Eds.), Advances in descriptive psychology (pp. 79–122). Greenwich, CT: JAI Press. Diener, E., & Diener, C. (1996). Most people are happy. Psychological Science, 7, 181–185. Diener, E., & Emmons, R. A. (1984). The independence of positive and negative affect. Journal of Personality and Social Psychology, 47, 1005–1117. Drigotas, S. M., Whitney, G. A., & Rusbult, C. E. (1995). On the peculiarities of loyalty: A diary study of response to dissatisfaction in everyday life. Personality and Social Psychology Bulletin. Fehr, B. (1996). Friendship processes. Thousand Oaks, CA: Sage. Jacobson, N. S., Follette, W. C., & McDonald, D. W. (1982). Reactivity to positive and negative behavior in distressed and nondistressed couples. Journal of Consulting and Clinical Psychology, 50, 706–714. Karney, B. R., & Bradbury, T. N. (1997). Neuroticism, marital interaction, and the trajectory of marital satisfaction. Journal of Personality and Social Psychology, 72, 1075–1092. Kelly, E. L., & Conley, J. J. (1987). Personality and compatibility: A prospective analysis of marital stability and marital satisfaction. Journal of Personality and Social Psychology, 52, 27–40. Lester, D., Haig, C., & Monello, R. (1989). Spouses’ personalities and marital satisfaction. Personality and Individual Differences, 10, 253–254. McCrae, R. R., & Costa, P. T., Jr. (1982). Self-concept and the stability of personality: Crosssectional comparisons of self-reports and ratings. Journal of Personality and Social Psychology, 43, 1282–1292. McCrae, R. R., & Costa, P. T., Jr. (1987). Validation of a five factor model of personality across instruments and observers. Journal of Personality and Social Psychology, 52, 81– 90. Rook, K. S. (1984). The negative side of social interaction: Impact on psychological wellbeing. Journal of Personality and Social Psychology, 46, 1097–1108. Rusbult, C. E., Drigotas, S. M., & Verette, J. (1994). The investment model: An interdependence analysis of commitment processes and relationship maintenance phenomena. Communication and Relational Maintenance, 115–139. Rusbult, C. E., Johnson, D. J., & Morrow, G. D. (1986). Impact of couple patterns on distress and nondistress in dating relationships. Journal of Personality and Social Psychology, 50, 744–753.

AFFECTIVITY, PERSONALITY, AND FRIENDSHIP

107

Schmutte, P. S., & Ryff, C. D. (1997). Personality and well-bring: Reexamining methods and meanings. Journal of Personality and Social Psychology, 73, 549–559. Sillars, A. L. (1980). Attributions and communication in roommate conflicts. Communication Monographs, 47, 180–200. Shaver, P. R., & Brennan, K. A. (1992). Attachment styles and the ‘big five’ personality traits: Their connections with each other and with romantic relationship outcomes. Personality and Social Psychology Bulletin, 18, 536–545. Suh, E., Diener, E., & Fujita, F. (1996). Events and subjective well-being: Only recent events matter. Journal of Personality and Social Psychology, 70, 1091–1102. Tellegen, A. (1985). Structures of mood and personality and their relevance to assessing anxiety, with an emphasis on self-report. In A. H. Tuma & J. D. Maser (Eds.), Anxiety and the anxiety disorders (pp. 681–706). Hillsdale, NJ: Erlbaum. Watson, D. (1988a). The vicissitudes of mood measurement: Effects of varying descriptors, time frames, and response formats on measures of Positive and Negative Affect. Journal of Personality and Social Psychology, 55, 128–141. Watson, D. (1988b). Intraindividual and interindividual analyses of positive and negative affect: Their relations to health complaints, perceived stress, and daily activities. Journal of Personality and Social Psychology, 54, 1020–1030. Watson, D., & Clark, L. A. (1992). On traits and temperament: General and specific factors of emotional experience and their relation to the five-factor model. Journal of Personality, 60, 441–476. Watson, D., Clark, L. A., McIntyre, C. W., & Hamaker, S. (1992). Affect, personality, and social activity. Journal of Personality and Social Psychology, 63, 1011–1025. Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and validation of brief measures of positive and negative affect: The PANAS scales. Journal of Personality and Social Psychology, 54, 1063–1070. Watson, D., & Hubbard, B. (1996). Adaptational style and dispositional structure: Coping in the context of the five-factor model. Journal of Personality, 64, 737–774. Watson, D., Hubbard, B., & Wiese, D. (in press). General traits of personality and affectivity as predictors of satisfaction in intimate relationships: Evidence from self- and partnerratings. Journal of Personality. Watson, D., & Pennebaker, J. W. (1989). Health complaints, stress, and distress: Exploring the central role of negative affectivity. Psychological Review, 96, 309–328. Watson, D., & Slack, A. K. (1993). General factors of affective temperament and their relation to job satisfaction over time. Organizational Behavior and Human Decision Processes, 54, 181–202. Watson, D., & Tellegen, A. (1985). Toward a consensual structure of mood. Psychological Bulletin, 98, 219–235. Watson, D., & Walker, L. (1996). The long-term stability and predictive validity of trait measures of affect. Journal of Personality and Social Psychology, 70, 567–577. Zaleski, Z., & Galkowska, M. (1978). Neuroticism and marital satisfaction. Behavior Research and Therapy, 16, 287–288.