Agreeableness as a Moderator of Interpersonal Conflict

Transcription

1 Agreeableness as a Moderator of Interpersonal Conflict Lauri A. Jensen-Campbell Florida Atlantic University William G. Graziano Texas A&M University ABSTRACT This multimethod research linked the Big Five personality dimensions to interpersonal conflicts. Agreeableness was the focus because this dimension is associated with motives to maintain positive interpersonal relations. Converging responses to both hypothetical conflicts and to diary records of actual daily interpersonal conflicts across a two-week period were assessed. Agreeableness was expected to moderate affective responses and tactical choices during conflicts. Patterns of daily conflict were related to self-reported reactions to hypothetical conflicts and to teacher-rated adjustment in adolescents. As predicted, Agreeableness was related to responsiveness to conflict. We thank Shaun D. Campbell, Deborah A. Kashy, Harry T. Reis, and Stephen Giorgio West for their help and expertise in data collection and analysis. We also wish to thank W. Steve Rholes, Emily Davidson, Howard Kaplan, and Radmila Prislin for their suggestions and comments on earlier versions of this article. We thank Joel Trevino, Gina Sullivan, Amanda Cumberland, Lisa Maloney, Marcel Satsky, Shelley Bannerot, DaShawn Glover, and Trish Ezell for their help in completing the data collection. We are especially grateful to John Moehlman and Paul Kunz, and to the teachers and adolescents of the Brazos Valley for their cooperation. This research was supported by a grant from the Society for Psychological Studies of Social Issues and by a Dissertation Fellowship award from the College of Liberal Arts, Texas A&M University, to Lauri A. Jensen- Campbell. It was also supported by the National Science Foundation Grant SBR and by a grant from the National Institute of Mental Health to William G. Graziano. Correspondence can be sent to Lauri A. Jensen-Campbell, Department of Psychology, Florida Atlantic University, 2912 College Avenue, Davie, FL (USA). Journal of Personality 69:2, April Copyright 2001 by Blackwell Publishers, 350 Main Street, Malden, MA 02148, USA, and 108 Cowley Road, Oxford, OX4 1JF, UK.

2 324 Jensen-Campbell & Graziano Agreeableness differences and use of destructive tactics in conflict were significantly related to evaluations of the individual s adjustment by knowledgeable raters. Among the Big Five dimensions, Agreeableness was most closely associated with processes and outcomes during interpersonal conflict. The five-factor structural approach offers some advantages to personality theory and research. From a measurement perspective there is convergence in empirical work pointing toward a five-factor approach (e.g., Digman, 1997; Digman & Takemoto-Chock, 1981; Kohnstamm, Halverson, Mervielde, & Havill, 1998; Ozer & Reise, 1994; Panter, Tanaka, & Hoyle, 1994). Some writers claim that the five factors have the status of an empirical fact, like the fact that there are seven continents on earth. Others have argued that personality now has established a language for describing the basic structure of personality (Digman, 1997; Digman & Inouye, 1986; Goldberg & Rosolack, 1994; John & Srivastava, 1999). Despite the apparent consensus, Block (1995) objected to the term model, suggesting the term approach as a more accurate substitute. He pointed to deficiencies in the five-factor approach as a general perspective on personality structure. In particular, Block (1995) noted the absence of an overall theoretical explanation for the five-factor approach. Similar concerns were expressed previously by Briggs (1989), who noted that no one had an a priori theoretical reason why these five dimensions, rather than some other five, should emerge, nor had anyone offered a coherent, falsifiable explanation for the five-factor system. One way to address concerns about the atheoretical nature of the five-factor approach is to develop theoretical accounts. It is true that some elements of the five-factor system are more developed theoretically than are others. Among the five factors, Extraversion and Emotional Stability/ Neuroticism are more commonly explored, not only because of their recognized importance for social relations but also because theory has been developed to explain their operation. No comparable theory is available yet for the other three dimensions, although some preliminary steps have been taken (e.g., Caspi, 1998; Kohnstamm et al., 1998; Rothbart & Bates, 1998). Of the five dimensions, Agreeableness is arguably the least well understood. Theory development may have been inhibited in part because of an undifferentiated theoretical understanding of social behavior. Both Extraversion and Agreeableness deal with social interaction, but Extraversion seems to have been the main focus of interest in the

3 Agreeableness & Conflict 325 dispositional underpinnings for social behavior (e.g., Elphick, Halverson, & Marzal-Wisniewska, 1998; Watson & Clark, 1997). Yet the social behavior associated with Extraversion has a different cast from the social behavior associated with Agreeableness (Finch & Graziano, in press; Finch, Okum, Pool, & Ruehlman, 1999; Graziano, Jensen-Campbell, & Finch, 1997; Havill, Besevegis, & Mouroussaki, 1998; Paulhus & John, 1998; Wiggins, 1991). Conceptually, Extraversion deals with social impact, whereas Agreeableness deals with motives for maintaining positive relations with others. A comprehensive dispositional analysis will require a detailed differentiation of these two seemingly different kinds of social behaviors. Some differentiated theoretical accounts of Agreeableness are beginning to emerge (Digman, 1997; Finch & Graziano, in press; Graziano & Eisenberg, 1997; Graziano, Jensen-Campbell, & Finch, 1997; Hogan, 1983; Paulhus & John, 1998; Wiggins, 1991). Compared to the other four factors, Agreeableness is distinctive in its connection to motives for maintaining positive interpersonal relations. Historically, personality labels have been used to describe a basic dimension of personality as social interest that is manifested in cooperation and empathy, friendly compliance (Digman & Takemoto-Chock, 1981), and likability (Hogan, 1983). Taking an evolutionary perspective, Hogan (1983) argued that individuals develop characteristics that allow them to get along with others in a group. Individuals who are unwilling or unable to cooperate or negotiate resources within the group may have been excluded from the group, limiting personal and reproductive rewards (see also Graziano, Jensen-Campbell, Todd, & Finch, 1997; Graziano, Hair, & Finch, 1997). Agreeableness may allow individuals to minimize the negative impact of conflicts and negotiate outcomes that capitalize on the advantages of group living. In a similar vein, Digman and Takemoto-Chock (1981) linked Agreeableness to tensions among group concerns for altruism and individual motives of selfishness. Moving from interpersonal theory to the biobehavioral level of analysis, Rothbart and her colleagues (e.g., Rothbart, 1989; Rothbart, Derryberry, & Posner, 1994; Rothbart & Bates, 1998) suggest that Agreeableness may emerge developmentally from temperamental selfregulative systems. Ahadi and Rothbart (1994) suggest that Agreeableness is linked to temperamental bases of effortful control, specifically the regulation of anger. They found that negative affect (i.e., anger/

4 326 Jensen-Campbell & Graziano frustration) is separable from fear both in infancy and adulthood and loads on the negative pole of Agreeableness. Moreover, Robins, John, and Caspi (1994) found that Agreeableness is related negatively to adolescent antisocial personality and self-reported delinquency. One way to integrate these conceptual analyses of Agreeableness is through specification of levels of analysis. According to McAdams (1995), individual differences in personality may be described at three different levels. Level I is comprised of relatively nonconditional, decontextualized, comparative dimensions of personality like the five-factor traits. Level II is comprised of contextualized strategies, plans, and concerns that enable persons to solve life tasks and accomplish goals. Level III is the domain of the life narrative, in which people construct stories of their lives retrospectively to provide a sense of overall meaning to the life. Level I speaks to what a person has, Level II speaks to what a person does, whereas Level III speaks to the making of the self. McAdams (1995) notes that the levels are conceptually distinct and should not be seen as necessarily derivative of one another. Nonetheless, we suggest that there will be at least a loose linkage, especially between the five-factor dimensions at Level I and the organized systems of personal concerns associated with the self at Level II. McAdams ultimately conceded that the linkages, if they indeed exist, should be established empirically rather than assumed by theorists to be true (p. 380). Our focus is on potential connections between Levels I and II, so we will not discuss Level III in detail here. McAdams (1995) asserts that five-factor personality dimensions are Level I attributes. One way that they might exert their influence on behavior and cognition is through their channeling of persons use of the contextualized strategies and motivation systems of Level II (Graziano, Jensen-Campbell, & Hair, 1996). The implication is that strategies and motive systems (Level II) are mediating links between dispositions at Level I and substantive narrative cognition about the self and others at Level III. (See Graziano, Jensen-Campbell, & Finch, 1997, figure 1; Graziano, Hair, & Finch, 1997.) If this theoretical analysis is valid, then conceptual and empirical work is needed to relate each of the five-factor dimensions to specific strategies and motives (e.g., Agreeableness with motives to minimize interpersonal conflict). It is possible, of course, that Level I attributes could bypass situated motives and strategies and exert their influence through some other route (e.g., automated responses, reflexive social skills).

5 Agreeableness & Conflict 327 Agreeableness may be a particularly good candidate for analyzing the multilevel integration of personality and the situated use of strategies for maintaining positive relationships with others. Agreeableness differences may be a phenotypic, summary expression of a set of underlying multilevel processes of the sort described previously. The Agreeableness variable commonly measured in self-report can be seen as the Level II phenomenological agency that coordinates the demands of the immediate situation with the constraints imposed on the individual by dispositions and residues of life experiences (Graziano, Jensen-Campbell, & Finch, 1997). For example, the high agreeable person may deploy a situated strategy to cope with the challenge of an aggressive adversary during family conflict, constrained by a Level I disposition toward Agreeableness, and a social learning history of success with negotiation as a resolution strategy for family conflict. From a personality development perspective, nowhere in the life course are conflicts among diverse intrapersonal and interpersonal forces and problems of adjustment more apparent than in adolescence (e.g., Grotevant, 1998; Larson & Richards, 1994). One of the most complex life ecologies surrounds the transition from childhood to adolescence. In this phase of the life cycle, physiological changes associated with puberty occur, relations with parents and peers are changing, sex roles are reevaluated, and school structure and academic requirements are altered (Simmons & Blythe, 1987). There is evidence that, during early adolescence, the self-concept is reorganized, and personality processes may be undergoing rapid transition (Eccles et al., 1989). In this age group and social ecology, five-factor dimensions (Level I), and their links to situated strategies and self-evaluations (Level II), may be especially apparent. If the analysis outlined previously is valid, then the five-factor approach might be useful for linking personality to adjustment generally, and to interpersonal conflict in particular in adolescence (Jensen-Campbell, Graziano, & Hair, 1996). Graziano and Ward (1992) used a converging, multimethod assessment that included teacher ratings and computerbased self-report in adolescents to probe links among the Big Five personality attributes and adjustment within the school context. Teachers ratings of overall adjustment in adolescents were related to every Big Five dimension except Factor V (Intellect/Openness). There are some important limitations to the research by Graziano and Ward (1992). First, they assessed the Big Five dimensions in their adolescent research participants through ratings by participants teachers

6 328 Jensen-Campbell & Graziano and counselors. They did not collect adolescent self-report assessments of the five-factor dimensions, however, so there were no data on convergence of teacher rating (R-data) and adolescent self-report (S-data) on the Big Five dimensions. Second, Graziano and Ward did not examine processes that might link personality to adjustment. Personality may exert its influence on adolescent adjustment intrapersonally (e.g., through processes of internal coherence) or interpersonally (e.g., through processes of accommodations to others). It is possible, for example, that the Agreeableness-adjustment link was at least partially mediated by patterns of social exchanges (Finch & Graziano, in press; Finch et al., 1999). All individuals, but especially adolescents, must learn to make social adjustments when faced with opposition from other individuals. Most modern theorists, at least within the psychology of interpersonal relations, agree that opposition or incompatibility is central to defining conflict (Laursen & Collins, 1994; Shantz & Hartup, 1992). The presence of opposition transforms conflict from an individual, intrapsychic phenomenon to an interpersonal, explicitly social event. Mutual behavioral opposition implies that conflict is a temporal event. That is, conflicts unfold over time, so they can involve a progressive transformation of the structure of the interaction (cf. Kelley, 1979, pp ). Conflicts are among one of the most potent sources of distress in daily life (Bolger, DeLongis, Kessler, & Schilling, 1989; Suls, Martin, & David, 1998). Conflicts, however, are not exclusively negative events. Many researchers see conflict as important to social development. Conflict can enhance an individual s understanding of self and his or her interconnectedness with others (Deutsch, 1973; Shantz & Hobart, 1989). Moreover, conflict is seen as responsible for giving participants important information about social boundaries (Vuchinich, 1990). Conflicts occur over time, so they can involve a progressive transformation of the structure of the interaction. Shantz and Hartup (1992) suggested that conflicts contain distinct components and that studying these components separately may contribute to our understanding of conflict events as they occur over time. Most researchers, however, have focused primarily on the initiation and frequency of conflicts, which may bias our understanding of conflict in relationships (Laursen & Collins, 1994; Perry, Perry, & Kennedy, 1992). The frequency of conflicts and the tactics used during conflicts can have very different effects on conflict outcomes.

7 Agreeableness & Conflict 329 Resolutions have recently emerged as an important component for both the eventual outcome of the conflict and the impact on the relationship (Laursen & Collins, 1994). Vuchinich (1990) noted that the outcomes of conflicts often carry the possibility of winning and losing in conflict. Individuals in interpersonal conflict situations often strive to win conflicts and avoid losing them because social standing and self-esteem are often at stake. Understanding the implications of different strategies or tactics to end conflict episodes may help us understand what resolution techniques aggravate or mitigate conflict. Since conflicts are events that occur within a sequence of interpersonal behaviors, characteristics that moderate interpersonal behavior (e.g., Agreeableness) could help predict patterns of interpersonal conflict. These characteristics may moderate the types of strategies used in a conflict, the tactics deployed on line, and ultimately the destructive/constructive nature of the episode (Graziano et al., 1996; Jensen-Campbell, Graziano, & Hair, 1996). It is possible that the motive systems associated with Agreeableness may predispose the individual to perceive persons in the social environment in ways consistent with the salient motives (Finch & Graziano, in press; Finch et al., 1999; Graziano et al., 1996; Graziano, Jensen-Campbell, & Finch, 1997; Graziano, Hair, & Finch, 1997). This working model provides a framework for empirical analysis. The links between Agreeableness and overt conflict behavior are mediated, at least in part, by corresponding differences in social perceptions. These perceptions then steer the individual toward patterns of attributions to other persons and toward corresponding social outcomes. Agreeable people may be greatly motivated to maintain positive relations with other people, and this motive system induces the agreeable person to generate positive perception and attributions to otherwise provocative behavior. These attributions induce the agreeable person to regard the conflict with less negative affect, to select more constructive conflict tactics, and to generate a more constructive pattern of oppositions during conflict than would a low agreeable person (Graziano et al., 1996; Jensen-Campbell et al., 1996). It is possible that other aspects of Agreeableness besides social perception can affect conflict outcomes. These would appear as nonmediated effects in the model. The present research used two different methods to assess responses to conflict. Each method had strengths and weaknesses, but, in combination, they provide a more comprehensive, converging picture of conflict. The first method involved strategy and behavior choices to verbally presented,

8 330 Jensen-Campbell & Graziano hypothetical conflict vignettes. In terms of strengths, vignettes are direct, and focus participants attention on the variables of interest. In addition, the vignette method allows the researcher to control the conflict situation presented and to avoid some ethical problems associated with deception and manipulation of participants perceptions of conflict situations. In terms of weaknesses, however, vignettes may prematurely focus, package, and define the critical elements in interpersonal conflict situations (e.g., relative power of each participant; legitimacy of each perspective). Moreover, responses to vignettes may not predict how the same individual will react to in vivo conflict (cf., Graziano, 1987; Graziano, Hair & Finch, 1997; Greenberg & Folger, 1988). To overcome these weaknesses, a second method was also used. The Rochester Interaction Record (RIR) is a diary report method that does not rely on packaged, verbally presented hypothetical situations. It was explicitly designed to examine naturally occurring daily life experience across diverse substantive content (Reis & Wheeler, 1991; Suls, Martin, & David, 1998). The RIR provides a flexible means for assessing a wide range of life experiences; it is not tied to any specific substantive content. Previous RIR research has examined such diverse topics as self-disclosure, intimacy, satisfaction with same- and other-sex interactions, physical attractiveness, and loneliness (Laurenceau, Barrett, & Pietromonaco, 1998; Nezlek, 1993; Reis & Wheeler, 1991) but has never been used to study naturally occurring interpersonal conflicts in adolescence. The RIR method allows researchers to supplement the usual paper-and-pencil ratings with basic, quasi-ethological reports about frequency and patterns of molecular events (e.g., interpersonal conflict) during daily social life. Following this logic, we predicted, in both hypothetical situations involving vignettes and in actual daily interactions using the RIR method of assessment, that high agreeable participants would use constructive tactics more frequently, and destructive tactics less frequently, than low agreeable participants (Jensen-Campbell et al., 1996). Using the RIR methodology, we expanded our focus to include other conflict components, specifically negative affect and conflict outcomes. We predicted that agreeable adolescents would attempt to minimize affective responses in conflict, to have greater control of frustration, and to report less emotion than their less agreeable peers (Graziano et al., 1996). We also predicted that high agreeable adolescents would report achieving better interpersonal outcomes in conflicts than would low agreeable adolescents. This last prediction was based on the assumption that high agreeable adolescents

9 Agreeableness & Conflict 331 will seek to minimize relationship disruption during conflict episodes by continuing to talk to their conflict partner. Finally, we predicted that high agreeable participants would be rated by their teachers as better adjusted interpersonally than their peers (Graziano, Jensen-Campbell, & Finch, 1997). METHOD Research Participants A total of 167 students in 6th (n = 44), 7th (n = 47), and 8th grade (n = 76) (69 boys, 98 girls) from the Brazos Valley in central Texas (USA) volunteered for the study and were given parental permission for participation. Proportions based on ethnic background were 60.5% European Americans, 28.1% Mexican Americans, 10.2% African Americans, and 1.2% who were classified as Other. Within the Mexican American group, two subgroups were identified: those whose families had lived in Texas for many years and preferred to speak English (68.2%) and those that were new to Texas and had limited fluency in English (31.8%). A total of 155 of the 167 students completed the daily RIR records over a two-week period. (To provide a complete RIR, students could not be absent from school for more than two consecutive days because an experimenter met with each participant each school day to monitor record keeping by participants.) Those who completed the diary phase received $20 for their efforts. (The 12 students who did not provide complete records for the RIR did not differ in levels of Agreeableness from those who completed the two-week study.) Due to experimenter error, 5 research participants (RPs) were dropped from the earlier conflict vignette phase. Overall, 150 students completed all three phases of the project Degrees of freedom vary due to the nature of diary data. Missing data points excluded participants from being included in the analyses. For example, 6 RPs did not report any dyadic conflicts over the two-week period (although they did have nonconflictual interactions). Moreover, participants needed to have conflicts with both male and female partners to be included in the analyses. Some RPs had conflicts with only male or only female partners. Additional analyses not reported here found no evidence that Agreeableness was related to these differences in reporting patterns.

10 332 Jensen-Campbell & Graziano Materials Personality measures. A computerized version of Goldberg s standard (1992) markers was used to measure the five dimensions of personality (for details on the computer assessment methodology, see Graziano, Jensen-Campbell, & Finch, 1997; Graziano, Jensen-Campbell, Steele, & Hair, 1998). Following Goldberg s (1992) trait marker format, the computer instructed students to describe themselves as they are now as compared to other students the same age and sex. Each marker adjective word appeared on the screen one at a time. Using the arrow keys, the student moved a cursor to rate each word as an accurate description, from 1 (strongly disagree) to 5(strongly agree). If the student did not know a word, he or she could press the [F1] key for a definition. Several steps were followed to create composites for each of the five dimensions of personality. First, we used a relatively simple summing procedure. Instead of presenting the markers in a bipolar format (e.g., warm-cold), we separated the poles and presented them in a unipolar format (Briggs, 1989). This allowed us to produce difference scores for each of the five dimensions. For example, agreeableness was the sum of the differences for the person s selfratings of [warm-cold], [kind-unkind], [cooperative-uncooperative], [unselfishselfish], [polite-rude], [trustful-distrustful], [generous-stingy], [flexiblestubborn], [considerate-inconsiderate], and [agreeable-quarrelsome]. Larger numbers indicated higher numbers on agreeableness with a minimum of 40 and a maximum of +40. A paper-and-pencil version of Goldberg s standard (1992) markers was used to measure teacher ratings of the student s personality. Teachers were instructed to describe each student as he/she is now as compared to other students of the same age and sex. The alphas for self- and teacher-rated agreeableness were.83 and.97, respectively. A five-item measure was also created to measure teachers professional evaluation of adolescents overall adjustment. Teachers were asked to rate the students on their academic performance, same-sex peer relations, opposite-sex peer relations, teacher relations, and classroom behavior using a 1 (not well adjusted) to 5(very well adjusted) Likert-type scale. A teacher who reported knowing the participant well was chosen to complete the assessment. During one session, the teacher was asked to assess the student s adjustment at the beginning of the school year and now (i.e., at the end of the school year) (Graziano, Jensen-Campbell, & Finch, 1997; Graziano & Ward, 1992). The assessment covered two time periods because it was possible that adjustment problems appearing at the beginning of the school year might be an unstable, transitory student responses to novel contexts, teachers, or institutional requirements (Eccles et al., 1989). The correlation between ratings of adjustment at the twotimeswas.62 (p <.01). Given the high correlation, we collapsed across the two times to produce a single, more reliable adjustment score for each participant (α =.92).

11 Agreeableness & Conflict 333 Conflict scenarios. Stimulus stories were the same as those used in our previous research on interpersonal conflict among adolescents (Jensen-Campbell et al., 1996). They consisted of 12 stories presenting conflict situations (e.g., behavioral opposition). Stories were divided equally among relationships: siblings, friends, classmates, boyfriends/girlfriends, parents, and teachers (see appendix for sample vignettes). The adolescents task was to tell us how good or bad each choice was in solving conflicts in the different situations/relationships. Each resolution choice was rated from 1 (bad choice) to 9(great choice). They were told that there are many ways to solve a conflict, and although no one way to end a conflict is correct, some ways may be good for some situations and not good for others. We reminded them that no one would know what they answered so they could be as honest as possible. The suggested modes of conflict resolution for the vignettes were adapted from Sternberg and Soriano (1984) and Vuchinich (1990). These were: physical force; threats; wait and see; drop the topic/accept the situation; diffuse conflict by reducing demands; third-party mediation; undermine the opponent s esteem and position; walk away; and compromise. Use of threats was not included in teacher and parent conflicts because differences in power affected the feasibility of the choice. Drop the topic specifically referred to shifting away from the conflict issue but maintaining continued interaction with one s partner, whereas walk away referred to disengaging from interaction with one s partner as a way to end the conflict. The nine resolution tactics were collapsed across relationship type to create global composite scores (see Jensen-Campbell, et al., 1996, for a detailed description of relationship effects). Internal consistencies (α) ranged from.74 (threats) to.86 (reduce demands) with a mean internal consistency of.81. Research has found that different resolution tactics have different outcomes for conflict (e.g., Laursen & Hartup, 1989). Negotiation, dropping the topic, walking away, wait and see, third-party intervention, and diffusing the conflict strategies often allay conflict and allow for continued social interaction and the possibility of reaching equitable solutions (i.e., constructive tactics). Power assertion strategies (i.e., physical force, undermining esteem, and threats), on the other hand, tend to aggravate conflict and often lead to inequitable solutions and discontinued interactions among individuals (i.e., destructive tactics). Rochester Interaction Record. An age-appropriate version of the RIR, suitable for young adolescents, was adapted from procedures used by Reis and Wheeler (1991) to study interaction in college students. We created a record that measured daily ratings of every interaction lasting longer than 10 minutes or more and any conflict that occurred in briefer interactions. This was done because previous RIR work had participants rate any interaction that was 10 minutes or longer. We thought, however, that this method might be less useful because conflicts

12 334 Jensen-Campbell & Graziano may be relatively rare events, and the likelihood that conflicts would coincide with a 10-minute event would be small. Conflicts were defined as happening when what you are doing or saying is opposed, blocked, or objected to by another person and you object back. In other words, someone does or says something against you, and you do or say something against them back. Conflicts included everything from small verbal disagreements to physical fights. The RIR uses a simple Likert-type format to evaluate the quality of interactions. For every conflict, RPs were instructed to rate each episode from 1 (not angry at all; feelings not hurt at all; no force used; no verbal insults used; no negotiation; problem not solved; stopped talking to them) to7(very angry; feeling very hurt; lots of force used; lots of verbal insults; lots of negotiation; problem solved completely; continued to talk to them). Individual RIR items were then categorized into superordinate rating categories: constructive tactics (negotiation), destructive tactics (physical force and verbal insults), negative affect (anger and hurt feelings), and outcomes (problem solved and continuing to talk to conflict partner). Procedure Personality description. Data collection took place in three distinct phases. In the first two phases, each adolescent was assessed individually in a self-paced computer format tailored to the adolescent s own first language (English or Spanish; see Graziano, Jensen-Campbell, Steele, & Hair, 1998, for a more detailed description). Participants were taken from class and shown how to use the computer to describe themselves as accurately as possible. When the participants finished the scales, they were thanked and escorted back to class. Finally, teacher reports were collected using paper-and-pencil measures of levels of adjustment and Goldberg s (1992) adjectives on each of the adolescents. Conflict scenarios (Phase II). Approximately 10 months after completing the personality measures, the same adolescents were asked to participate in the second phase of the study. Following procedures reported by Sternberg and Soriano (1984), participants read each conflict vignette and evaluated the appropriateness of each resolution tactic. Each participant was then left to complete his/her own randomized version of all 12 vignettes using the self-paced computer format. Once they had completed all vignettes, the adolescents were asked if they had any difficulties or problems with the stories and were thanked for their participation.

13 Agreeableness & Conflict 335 Rochester Interaction Record (Phase III). Approximately one month later, we assessed perceptions of daily conflict using the event-contingent method of the RIR. To gain trust and avoid problems of self-presentation, we guaranteed each participant s confidentiality. Pilot testing was conducted to identify potential problems. In addition, actual RPs were given phone numbers of several contact persons in case they ran into difficulty. Participants initially met with the experimenters to learn the RIR procedures that they were to follow. Small groups of approximately participants were formed. Prior to the meeting, RPs who spoke Spanish were assigned to a Spanish-speaking group. The other RPs were each randomly assigned to a group within the constraint that one of his/her undergraduate research assistants was the same sex as the participant. Each group had two undergraduate research assistants who were responsible for answering further questions regarding the RIR format and procedures. These two undergraduates were the RP s contact persons in case of an emergency. In addition, all RPs had the first author s phone number. Research participants were told that we did not know a lot about adolescent conflict and that we wanted them to help us explore the kinds of conflicts in which people their ages are involved. We told them that conflicts were not necessarily bad things and were often good because they helped us understand others and ourselves. We emphasized that this was not a test and that there were no right or wrong answers. We then passed out a sample RIR to each RP and explained what an interaction and conflict was. We told them that they probably have conflicts every day that they do not even think about much and really are not big deals to them. We then had them give us examples of situations that are and are not conflicts. After they understood what a conflict was, we explained each question on the form and defined possible unknown words on the list (e.g., interaction). Each RP was also furnished with a helpful hint fact sheet that explained the RIR in detail. RPs were then given their individual RIR folder packet. Each packet contained RIR forms, a name/initial reminder sheet, contact phone numbers, the helpful hint sheet, and a reminder tablet. We told them to pick a time(s) to do the RIR forms each day (e.g., before they went to bed). The reminder tablet was to be used to record all interactions/conflicts during the day so that it would be easier to remember them when they went to do the RIR. After the initial meeting, we met with the RPs every day to collect forms and give out new ones. Research assistants used this time to establish rapport, to answer any questions, and to check records for possible problems. RPs were then thanked for their time and paid for their participation.

14 336 Jensen-Campbell & Graziano RESULTS Convergence of S-data and R-data Personality Measures (Phase 1) On average, adolescents and teachers rated the adolescents as relatively high on Agreeableness (see Table 1 for descriptive statistics). Adolescent self-ratings on Agreeableness correlated with the teachers ratings of Agreeableness, r (160) =.20, p <.01. This result is statistically significant, but the absolute magnitude of this correlation is small. Recall, however, that the teacher and student measures were administered in different formats, and correlation involved only single student teacher pairs. Had the rating been aggregated across multiple teacher raters, the correlation would probably have been larger (Graziano, Jensen- Campbell, & Finch, 1997; Kolar, Funder, & Colvin, 1996). There was no evidence of mean differences of self-reported data (S-data) in Agreeableness between the two sexes, t(160) = 1.63, ns. For teacher-rated (R-data) Agreeableness, there was a significant sex difference in mean levels of Agreeableness, t(156) = 3.28, p <.001. Teachers rated female participants as significantly higher in Agreeableness (M = 25.57; SD = 14.79) than male participants (M = 17.34; SD = 16.59). Hypothetical Conflict Situations (Phase II) Overall, adolescents reported that constructive strategies (e.g., compromise and diffusing the conflict) were more appropriate than destructive strategies (e.g., physical force and threats; see Table 1 for descriptive statistics). We predicted, however, that Agreeableness would moderate perceptions of the appropriateness of conflict resolution choices in hypothetical situations. Iterative sets of regression analyses were conducted to examine the unique contribution of each of the five dimensions of personality. As predicted, Agreeableness uniquely moderated the use of both constructive and destructive conflict resolution tactics. Persons higher in self-rated Agreeableness reported that the use of physical action, threats, and undermining partner s self-esteem was less appropriate than did persons lower in Agreeableness. Persons higher in self-rated Agreeableness also reported that the use of compromise and dropping the topic were more appropriate than did persons lower in Agreeableness (see Table 2 for regression weights). Emotional Stability was also related to the use of physical action and undermining self-esteem. Persons higher

15 Agreeableness & Conflict 337 Table 1 Descriptive Statistics for Personality and Conflict Ratings M SD Extraversion Self-Ratings Teacher Ratings Agreeableness Self-Ratings Teacher Ratings Conscientiousness Self-Ratings Teacher Ratings Emotional Stability Self-Ratings Teacher Ratings Intellect Self-Ratings Teacher Ratings Hypothetical Conflict Strategies Drop Topic/Accept Situation Walk Away Wait and See Compromise rd-Party Intervention Diffuse Conflict Physical Force Undermine Esteem/Position Threats RIR Diary Ratings Mean Interactions per Day Mean Conflicts per Day Anger Feelings Hurt Physical Force Verbal Insults Negotiation Problem Solved Continued Talking

16 Table 2 Summary of Regression Analyses for the Five Domains of Personality (S-Data) Predicting Hypothetical Conflict Tactics DOMAINS OF PERSONALITY Extroversion Agreeableness Conscientiousness Emotional Stability Intellect Drop Topic/Accept Situation B-weight t-value * * Walk Away B-weight t-value Wait and See B-weight t-value * Compromise B-weight t-value * rd-Party Intervention B-weight t-value Diffuse Conflict B-weight t-value

17 Table 2 Continued DOMAINS OF PERSONALITY Extroversion Agreeableness Conscientiousness Emotional Stability Intellect Physical Force B-weight t-value * * 0.74 Undermine Esteem/Position B-weight t-value * * 0.28 Threats B-weight t-value * *p <.05.

18 340 Jensen-Campbell & Graziano in emotional stability reported that the use of these destructive tactics was less appropriate than did persons lower in emotional stability. Finally, Intellect was negatively related to dropping the topic and waiting to see what would happen. There was no evidence that personality moderated step-down behavior, third-party intervention, and walking away from the conflict. Next, Agreeableness was treated as a continuous variable and centered to avoid problems with unstandardized solutions (Aiken & West, 1991). To analyze participants evaluations of conflict resolution choices, an Agreeableness 9 (type of resolution choice) 2 (sex of participant) repeated measures factorial was conducted. Following a significant multivariate effect, simple effects analyses of interactions were conducted. There was an overall multivariate self-rated Agreeableness Tactic interaction, F(8, 151) = 5.42, p <.001. Self-rated Agreeableness moderated the endorsement of walking away, compromise, physical force, undermining partner s self-esteem, and use of threats (Fs > 6.06, ps <.01) (see Table 3 for bivariate correlations). There was no evidence of a Sex of Participant Tactic interaction, Sex of Participant Self-Rated Agreeableness interaction, or Sex of Participant Agreeableness (S-data) Tactic interaction α(fs < 3.07, ps >.05). Next, an Agreeableness 9 (type of resolution choice) 2 (sex of participant) repeated measures factorial was conducted to examine the effects of teacher-rated Agreeableness (R-data). There was again an overall multivariate Agreeableness Tactic interaction, F(8, 147) = 4.26, p <.001. Students higher on teacher-rated Agreeableness endorsed walking away, wait and see, and compromise more than did students lower on Agreeableness. Students lower on teacher-rated Agreeableness endorsed physical force and threats more than did students higher on Agreeableness (see Table 3 for bivariate correlations). There was no evidence of a teacher-rated Agreeableness Sex of Participant Tactic interaction, F(8, 147) =.84, ns. Collapsing across type of relationship in conflict scenarios may conceal potentially useful information about the situated strategies agreeable adolescents use. Supplementary analyses were conducted using an Agreeableness (S- and R-data) 2 (sex of participant) 5 (type of relationship) 8 (type of resolution choice) repeated measures factorial to examine this possibility. Because other analyses revealed that the majority of the sample did not have romantic partners, same-sex friends

19 Agreeableness & Conflict 341 Table 3 Intercorrelations Among Agreeableness and Hypothetical Conflict Tactics Self-Rated Teacher-Rated Agreeableness Agreeableness Drop Topic/Accept Situation Walk Away 0.18* 0.25* Wait and See Compromise 0.40* 0.23* 3rd-Party Intervention Diffuse Conflict Physical Force 0.41* 0.23* Undermine Esteem/Position 0.39* 0.10 Threats 0.34* 0.15* *p <.05; +p <.10. and romantic partners were combined to create a close friend category. Threats were also removed from the analyses since using threats was not a choice for all relationship types. There was no evidence, however, of an Agreeableness Tactic Relationship interaction, Fs = 1.47,.73, for S- and R- data respectively, ns. 2 Compared to the other four dimensions of personality, Agreeableness was uniquely related to the endorsement of both constructive and destructive tactics using the vignette methodology. Moreover, the type of relationship the vignette described did not moderate Agreeableness s relation to resolution tactics (e.g., parent versus sibling). Agreeable adolescents consistently endorsed constructive tactics more than did their less agreeable peers. Conversely, low agreeable adolescents endorsed destructive tactics more than did their high agreeable peers. Rochester Interaction Record (Phase III) Overview. Conflicts occur over time and involve a progressive transformation of the structure of the interaction. Thus, conflicts contain distinct 2. There was a significant Resolution Tactic Relationship interaction, F(28, 131), p <.001, that is beyond the scope of this article. Please contact the authors, if you are interested in a more detailed description of this finding.

20 342 Jensen-Campbell & Graziano components within each conflict event. These various components represent distinct features that contribute to our understanding of conflict events. The RIR record specifically focused on negative affect, constructive tactics, destructive tactics, and conflict outcomes. In relation to Agreeableness, we first predicted that agreeable adolescents would attempt to minimize negative affective responses in conflict. Two measures were collected for negative affect: anger and hurt. We then predicted that high agreeable participants would use constructive tactics more frequently, and destructive tactics less frequently, than would low agreeable participants. To examine destructive tactics, we collected two measures: physical force and verbal insults. To examine constructive tactics, a measure of negotiation was collected. Finally, it was predicted that high agreeable adolescents would report having better interpersonal outcomes in conflicts than would low agreeable adolescents. Interpersonal outcomes were measured by how well the conflicts were solved and by how much participants continued to talk to their conflict partners. Statistical analyses. To test our hypotheses about Agreeableness, we conducted iterative sets of hierarchical linear models (HLMs; Bryk & Raudenbush, 1992; Kenny et al., 1998). In our data, conflicts are nested within individuals. HLM procedures treat both lower (interactions) and upper (persons) levels as sampling units so inferences could be made about both interactions as well as persons. This random coefficient estimation approach specifically involves a two-step procedure of hierarchically nested regressions. First, we calculated regression equations that estimated how much each RP reported systematic ratings with his/her partners based on the sex of partner. Categorical variables of partner sex and RP sex were coded to center around zero ( 1 for males; +1 for females). In the second step, the first-step regression coefficients are regressed on Agreeableness, sex of RP, and the cross product of Sex Agreeableness (see Bryk & Raudenbush, 1992; Kenny et al., 1998). We report analyses for standardized Agreeableness scores for our analyses, which produced identical results (i.e., t-values) to our analyses using unstandardized centered scores. The analyses presented here are weighted least squares procedures and not ordinary unweighted least squares procedures. The weighting scheme used for the first-step regressions is computed so that dyads that engaged in more interactions get more weight than dyads that interacted less frequently. The interactions, however, are weighted more

21 Agreeableness & Conflict 343 only to the extent that they yield new independent information. The degree of nonindependence in these analyses is measured by the intraclass correlation. The second-step regressions are also weighted in such a way (e.g., 1/variance of coefficients) that highly stable coefficients are given more weight than less stable ones (Kenny et al., 1998). Thus, means derived from these procedures follow this weighting scheme. We might note here that traditional procedures for analyzing RIR data (e.g., Reis & Wheeler, 1991) implicitly weight by the number of interactions for an RP. HLM analyses were limited to single interaction partner reports, which involved 72% of all conflicts. 3 Descriptive information. Descriptive statistics concerning the average number of conflicts and interactions per day as well as the mean levels of negative affect, negotiation, physical force, verbal insults, continued talking, and problem solved are presented in Table 1. On average, participants reported 1.5 conflicts per day. Although similar to past reports with adult samples (Bolger et al., 1989; Suls et al., 1998), adolescents reported more conflicts per day than did adult participants in prior research. Negative affect. We predicted that agreeable adolescents would attempt to minimize affective responses in conflict and to report less negative emotion. Contrary to our predictions, Agreeableness (S-data) was positively related to self-reports of anger (b =.23), t(123) = 2.16, p <.03. This effect was qualified, however, by interactions involving sex of the interaction partner and sex of the respondent (RP). First, there was evidence of a self-rated Agreeableness Sex of Partner interaction, t(104) = 2.47, p <.02 (see Table 4 for b-weights). Persons rating themselves higher on Agreeableness reported more anger with male partners than did persons who rated themselves lower on Agreeableness (b =.39). In interactions with female partners, however, there was no evidence that Agreeableness was related to reports of anger (b =.07). Second, there was also an Agreeableness Sex of RP interaction, t(123) = 2.04, p <.04. For female RPs, Agreeableness was positively related to self-reports of anger (b =.36), t(69) = 2.57, p <.03. For male 3. More detailed descriptions about the multi-level analyses can be obtained from the authors.

22 Table 4 Regression Coefficients Using Agreeableness to Predict RIR Components SELF-RATED PERSONALITY TEACHER-RATED PERSONALITY Female Male Female Male Variable Overall Participants Participants Overall Participants Participants Affective Intensity Anger 0.23* 0.36* * 0.52* 0.04 Feelings Hurt * 0.70* 0.05 Destructive Tactics Physical Force * 0.90* 0.04 Verbal Insults * 0.98* 0.02 Constructive Tactics Negotiation 0.38* 0.33* 0.47* 0.24* 0.53* 0.05 Outcomes Conflict Solved * Continue Talking * * 0.21 Note. Ratings could range from 1 to 7.

23 Agreeableness & Conflict 345 RPs, there was no evidence of a relationship between Agreeableness (S-data) and anger (b =.11), t(54) =.69, ns. As predicted, teacher-rated Agreeableness (R-data) was negatively relatedtoself-reports of anger (b=.27),t(123)= 2.91,p<.004(seeTable 4). There was also a significant Agreeableness (R-data) RP Sex interaction, t(123) = 2.35, p <.02. Agreeableness (R-data) was negatively related to self-reports of anger for female RPs (b =.52), t(69) = 4.07, p <.001. For male RPs, there was no evidence of a relationship between Agreeableness and anger (b =.04), t(54) =.29, ns. For feelings being hurt, there was a self-rated Agreeableness Sex of RP interaction, t(121) = 2.22, p <.03. For female RPs, Agreeableness (S-data) was positively related to self-reports of feelings being hurt (b =.32), t(69) = 1.78, p <.08. There was an Agreeableness Sex of Partner interaction for female participants, t(59) = 2.34, p <.02. Agreeableness levels of female adolescents were only positively related to feelings being hurt with male partners (b =.47). There was no evidence of a relation between Agreeableness and hurt feelings for female partners (b =.17). For male RPs, there was no evidence that feelings being hurt was related to Agreeableness (S-data) (b =.14), t(52) =.77, ns. As predicted, there was a main effect for teacher-rated Agreeableness, t(121) = 3.54, p <.001. Agreeableness (R-data) was negatively related to feelings being hurt (b =.37). There was also an Agreeableness RP Sex interaction, t(121) = 3.10, p <.002. For male RPs, Agreeableness (R-data) was not significantly related to feelings of being hurt (b =.05), t(52) =.34, ns. For female RPs, Agreeableness (R-data) was negatively related to self-reports of feelings being hurt (b =.70), t(69) = 4.75, p <.001. In sum, negative affect was systematically related to Agreeableness differences. Two findings are of particular interest. First, the relation between Agreeableness (S- and R- data) and negative affect held only for girls. Second, teacher-rated Agreeableness and self-reported Agreeableness produced different results. As predicted, teacher-rated Agreeableness was negatively related to negative affect during conflict episodes. For self-rated Agreeableness, the opposite pattern emerged. The relation between teacher-reported Agreeableness and negative affect was larger in magnitude, however, than the link between self-reported Agreeableness and negative affect (b =.52 versus.36).