Adolescent prosocial behavior: the role of self-processes and contextual cues

Peer‐ and teacher‐reported prosocial behavior of 339 6th‐grade (11–12 years) and 8th‐grade (13–14 years) students was examined in relation to prosocial goals, self‐processes (reasons for behavior, empathy, perspective taking, depressive affect, perceived competence), and contextual cues (expectations of peers and teachers). Goal pursuit significantly predicted prosocial behavior, and goal pursuit provided a pathway by which reasons for behavior were related to behavior. Reasons reflected external, other‐focused, self‐focused, and internal justifications for behavior; each reason was related to a unique set of self‐processes and contextual cues. Associations between prosocial outcomes and sex and race (Caucasian and African American) were mediated in part by self‐processes and contextual cues. The implications of studying prosocial behavior from a motivational perspective are discussed.

Sharing, helping, and cooperative forms of behavior are hallmarks of social competence in childhood and adolescence. In addition, these behaviors have been related theoretically and empirically to other forms of social competence such as social acceptance and approval (e.g., [6], [32]), and to intellectual competencies such as academic performance (e.g., [47]). Research on the development of prosocial behavior has identified a range of self‐processes likely to motivate displays of prosocial actions including perspective taking, empathy, levels of moral reasoning, and affective functioning ([18], [20]). Qualities of social interactions with parents and peers have also been linked to displays of prosocial behavior (see [7], [18]). Despite this growing body of work, understanding of the psychological, social, and developmental underpinnings of prosocial behavior is still somewhat limited. For example, studies of mediating and moderating factors that link self‐processes and social influences to prosocial behavior are rare ([18]), and little is known about prosocial behavior beyond early and middle childhood ([20]). In addition, few researchers have examined multiple self‐processes and social influences in relation to prosocial actions in a single study.

In light of these limitations, the current research was designed to extend understanding of prosocial behavior in several ways. First, we developed a conceptual framework that integrates aspects of developmental models of prosocial behavior ([18]), work on social motivation ([48]), and self‐determination theory ([37]). The model allows for the possibility that individuals can have multiple reasons for their prosocial actions ([40]), that each reason might be related to unique sets of self‐processes and social influences, and that social goal pursuit serves as a pathway that links these variables to prosocial behavior. Second, we examined self‐processes and social cues as mediators between sex, race, and prosocial behavior; we also examined the role of sex and race as moderating variables. Finally, we focused on the prosocial behavior of young adolescents in middle school. At this age, prosocial behavior has been related positively to self‐processes such as perceived competence ([9]), emotional well‐being ([9], [50]), and altruistic moral reasoning ([8]). However, processes typically related to prosocial behavior in young children such as empathy and perspective taking ([18]), and socialization processes that might motivate prosocial actions at school, have rarely been studied in adolescent samples.

In the current study, we conceptualized prosocial behavior as the outcome of multiple self‐processes and contextual affordances, utilizing theoretical perspectives on motivation (e.g., [37], [48]) and the development of prosocial behavior (e.g., [18]). As shown in Figure 1, our conceptual model depicts prosocial behavior as being motivated most proximally by an individual's intentions or goals to be prosocial. In turn, we hypothesized that prosocial goal pursuit is influenced in part by a set of self‐processes including reasons for prosocial behavior, empathy, perspective taking, depressive affect, and perceived competence, and social cues in the form of perceived expectations for prosocial behavior. We developed this model to characterize young adolescents' prosocial behavior in middle school classrooms. Therefore, classmates and teachers were considered to be the most proximal and salient source of social cues.

Graph: 1 Model of adolescent prosocial behavior.

Goal pursuit and self‐processes as predictors of prosocial behavior. Personal goals in the social domain have been defined in a number of ways, including what an individual wants to achieve in a specific situation (e.g., to help others; [37], [46]), and as reasons for engaging in certain types of behavior (e.g., to gain approval from others; [14]). We adopted both definitions, focusing on the content of students' classroom goals as well as the reasons they have for achieving them. We use the words goals and reasons to refer to what students want to achieve, and why they want to achieve it, respectively. However, both of these constructs can be viewed as aspects of personal goals. With respect to goal content, our interest was in the extent to which students want to achieve specific prosocial outcomes, that is, to help, cooperate, and follow rules in their classes at school. Evidence suggests that pursuit of these goals is related to helping, sharing, and cooperative classroom behavior (e.g., [44]).

We defined the reasons why a student might want to pursue these prosocial goals to reflect types of behavioral regulation ([36]), with external reasons for behaving reflecting fear of punishment or a desire to comply, introjected reasons reflecting desires to maintain a positive sense of self either through gaining social approval (other‐focused) or avoiding negative feelings of guilt or shame (self‐focused), and internal reasons reflecting personal valuing of prosocial behavior. It is noteworthy that these types of behavioral regulation correspond closely to levels of moral reasoning, with external reasons reflecting the simplest level and an internalized concern for others reflecting the most complex and sophisticated level of reasoning. However, rather than assuming a developmental continuum from extrinsic or hedonistic to internalized regulation ([29]), we simply hypothesized that adolescents can have multiple reasons (e.g., extrinsic as well as intrinsic) guiding their behavior.

As shown in Figure 1, we also hypothesized that an additional set of self‐processes would motivate goal pursuit as well as determine reasons for behaving in prosocial ways. These self‐processes include empathy, perspective taking, depressive affect, and perceived competence. Empathy reflects a tendency to experience the affective states of others and is therefore, believed to be the affective basis of altruistic forms of prosocial behavior ([25]). Empirical evidence has linked empathy to prosocial outcomes in adolescents (e.g., Eisenberg et al., 1991), and significant relations between empathy and introjected and internal reasons for prosocial behavior, but not between empathy and external reasons have been reported ([36]). In light of this work, we expected empathy to be related most proximally to internal reasons for prosocial behavior and unrelated to external reasons. Social perspective taking reflects the ability to understand the emotional state or thinking of another person and is believed to be a requisite cognitive skill for experiencing empathy ([19]). Few researchers have included perspective taking in studies of adolescent prosocial behavior although it has been associated consistently with altruistic behavior and empathy in young children (e.g., [17], [42]). Given these conceptual and empirical links between empathy and perspective taking, we also predicted that perspective taking would be related to internal but not external reasons for behavior.

Depressive affect has also been studied in relation to prosocial behavior, typically assessed as a global disposition reflecting general levels of sadness and unhappiness. During adolescence, depressive affect has been related negatively to students' prosocial behavior ([9], [50]) and pursuit of goals to help, share, and cooperate with others ([49]). Young adolescents who experience relatively high levels of depression also tend to report high levels of compliance and concern about receiving positive evaluations from others (e.g., [27]). Therefore, we expected depressive affect to be related significantly and positively to external and other‐focused reasons for prosocial behavior but unrelated to internal reasons. Finally, perceived competence is believed to motivate efforts to achieve personal goals to the extent that the individual believes she has the ability to achieve them ([2]). Previous studies have documented that adolescents' beliefs about their ability to be helpful are related to their displays of prosocial behavior ([9], [31]). In line with this research, our model depicts perceived social competence as a predictor of goal pursuit; we did not have specific predictions about the relation of perceived competence to various reasons for behaving in a prosocial manner.

Contextual cues. We focused on social norms as contextual cues relevant for understanding adolescents' displays of prosocial behavior at school. Specifically, we predicted that adolescents' perceptions of teachers' and peers' expectations for prosocial behavior would be related to prosocial goal pursuit as well as to reasons for behaving prosocially. Social learning theory ([2]) would predict that within a classroom setting, students learn from teachers and peers as to which types of behavior are valued and rewarded. Therefore, adolescents' perceptions that others expect them to behave in prosocial ways should partly motivate them to do so. In addition, the effects of social learning tend to be particularly strong if the source of information is perceived as being powerful, similar to oneself, or a source of affiliation ([2]). It is reasonable to expect that adolescent students perceive their teachers and peers differently as a function of these dimensions, with teachers being viewed as having social power and peers as being similar and a source of social approval and acceptance. If true, perceived expectations of teachers and peers are likely to be related to different reasons for engaging in prosocial actions. For example, perceiving expectations from teachers to behave prosocially should invoke external reasons for behavior (e.g., fear of punishment). In contrast, perceived peer expectations for prosocial behavior should invoke empathy and thus, internal reasons for behavior; in addition, fear of social disapproval from peers might also result in wanting to maintain or gain social approval as a reason for prosocial behavior.

Sex, race, and grade level. Of final interest were the effects of sex, race, and grade level on adolescents' prosocial behavior. The effects of these variables on prosocial behavior are poorly understood in part because theoretical models that discuss them are rare. However, previous empirical findings provide some insights to guide further examination of ways in which sex, race, and grade level are related to prosocial behavior. With respect to sex differences, a fairly consistent finding in the literature is that adolescent girls tend to behave in a prosocial manner more often than boys (e.g., [20]). To some extent, this difference is likely to reflect how prosocial behavior is operationalized, the source of information, and issues of social desirability (see [18]). At a conceptual level, it also is likely that sex differences in prosocial behavior might be explained by more proximal, psychological processes and perceived social cues of interest in this study. For instance, girls tend to be more empathic ([30]) and report more internalized reasons for prosocial behavior ([8]) than boys. Therefore, we examined the possibility that self‐processes and social cues partly mediate relations between sex and prosocial behavior. In addition, sex also might serve as a moderating variable; depressive affect has been related to prosocial behavior more strongly for adolescent girls than for boys ([3], [27]). In light of this evidence, we also examined the role of sex in moderating relations between depressive affect and prosocial behavior, predicting that the relation would be stronger for girls than for boys.

Few researchers have studied the prosocial behavior of African American adolescents. However, it is important to understand the positive, adaptive behaviors that these students display at school. Therefore, we examined several issues with respect to race. First, there is conflicting evidence concerning group differences in behavior. On the one hand, Caucasian students in a Caucasian‐majority middle school (87% Caucasian) were reported by peers to display classroom prosocial behavior more frequently than were African American students in an African American majority middle school (92% African American; [46]). On the other hand, elementary‐age African American and Caucasian students' reports of their peers' classroom prosocial behavior yielded nonsignificant group differences when sex and classroom composition (majority White and majority African American) but not school were taken into account ([28]). To explore these findings further, we examined classroom prosocial behavior of Caucasian and African American middle school students attending the same school, while taking into account students' sex, as well as classroom composition (majority White vs. majority African American).

As with sex differences, we also considered the possibility that if significant group differences in prosocial behavior are found as a function of race, self‐processes and contextual cues might partly mediate relations between race and prosocial behavior. Finally, we examined whether race serves as a moderating variable, with expectations from peers and prosocial behavior being related differently for Caucasian and African American students. In the domain of academic achievement, several researchers have documented that peer expectations and values for achievement are related significantly to achievement‐related outcomes for Caucasian students but are not significantly related for African American students (e.g., [10], [12], [41]). We extended this work by examining this phenomenon in the social domain, that is, whether race also serves as a moderator of the relation between peer expectations and prosocial behavior.

Finally, we examined grade level as a predictor of prosocial behavior. Little is known about developmental differences in prosocial behavior during early adolescence. However, models that focus on stage‐environment fit (e.g., [16]) highlight early adolescence as a critical period during which developmental needs and contextual affordances often conflict. For example, as children enter into early adolescence they tend to express stronger needs for autonomy and self‐control ([33]). At the same time, school contexts tend to provide fewer opportunities for young adolescents to fulfill these needs; teachers tend to have a heightened focus on maintaining control in the classroom especially as sixth‐grade students make the transition from elementary school into middle school ([16]). Therefore, it is reasonable to expect that sixth‐grade middle school students, who tend to experience fewest opportunities for autonomy and self‐control in the classroom, would report more frequent use of external reasons for their prosocial behavior than their older peers.

Summary. In this study we examined young adolescents' prosocial behavior in relation to goals to behave prosocially, self‐processes (reasons for behavior, depressive affect, empathy, reasons, perspective taking, perceived competence), and contextual cues (expectations for behavior of peers and teachers). We predicted that adolescents' goals would serve as a pathway linking self‐processes and contextual cues to displays of prosocial behavior. We further expected that self‐processes and contextual cues would be related to goal pursuit by way of reasons for prosocial behavior, with empathy, perspective taking, and perceived peer expectations for prosocial behavior being related to internalized reasons for prosocial behavior and depressive affect and perceived teachers' expectations being related to more external reasons. Finally, we examined the effects of sex, race, and grade level on prosocial behavior.

In this study, prosocial behavior was operationalized as helping and sharing, as reported by peers and teachers. Therefore, behavior scores reflect tendencies to display prosocial behavior consistently over time. In support of this assessment strategy is evidence that adolescents tend to behave prosocially with some consistency ([20]). However, we also assumed that consistent displays of prosocial behavior might also be context specific, that is, that prosocial behavior can differ as a function of classroom setting, depending on expectations of teachers and peers in a specific class. Therefore, we assessed adolescents' perceptions of their teachers' and peers' expectations for them to behave prosocially in the same class in which their prosocial behavior was assessed. Finally, we included following classroom rules as a construct in our measures of goal pursuit and reasons for prosocial behavior. We included this motive for behavior because many classroom rules and expectations for behavior are prescriptions for prosocial actions ([21]).

The participants were 339 sixth‐grade and eighth‐grade students from a middle school in a suburban, predominantly middle‐class community in a mid‐Atlantic state. Forty‐seven percent of these students were sixth graders (n=162), 52% were girls (n=178), and 49% were Caucasian (n=165), 44% were African American (n=148), with the remaining students being of other ethnic status (n=26). Although data on individual students were not available, 26.2% of the students attending the school received free‐reduced lunch. The proportion of Caucasian to African American students school wide was the same as that found in our sample. All students of two sixth‐grade Caucasian female social studies teachers and two eighth‐grade Caucasian male social studies teachers (18 classes) were asked to participate; 83% of these students received parental permission to participate.

Data were gathered in the Spring semester during regular 45‐min class sessions. The study was presented as a survey of sixth‐grade and eighth‐grade students' opinions about their classroom experiences in middle school. Students were told that they did not have to answer any of the questions if they did not want to. Teachers remained in their classrooms to complete rating scales. Students and teachers were told that all of their responses would be confidential and that data would be stored in a locked file cabinet that was accessible only to the first author. All students and teachers signed consent forms that stated assurances of confidentiality.

Peer nominations and teacher ratings were used to assess prosocial behavior and self‐reports were used to assess the remaining variables. For all questions, students were instructed to answer with reference to their experiences in the class they were in at the time of data collection.

Depressive affect. The three‐item depression subscale of the Weinberger Adjustment Inventory, Short‐Form ([43]) assessed depressive affect (e.g., "I often feel sad or unhappy"; 1=False, 5=True). Items were averaged to yield a composite score; M(SD)/α=2.45(1.06)/.62, respectively.

Empathy and perspective taking. The [13] Interpersonal Reactivity Index (IRI; 1983) asks students to respond to items representing reactions to someone in distress. Seven empathy items reflect affective responses (e.g., "I often am concerned and feel for kids at school less fortunate than me"), and seven perspective‐taking items reflect the tendency to take another's point of view (e.g., "I try to look at everybody's side of a disagreement before I make a decision; 1=not true about me, 4=very true about me") Ms(SDs)/αs of averaged empathy and perspective‐taking scores were 2.90(.52)/.68 and 2.59(.54)/.73, respectively.

Perceived social competence. [22] perceived social competence scale (six items) from the Self‐Perception Profile for Adolescents asks in the following format: "Some teenagers usually do the right thing BUT other teenagers often don't do the right thing." Students then indicate which part of the sentence is "really true" or "sort of true" for them. M(SD)/α of averaged scores were 2.76(.73)/.79.

Reasons for prosocial behavior. Students' reasons for behaving in a prosocial manner were measured with three scales adapted from [36] Prosocial Self‐Regulation Questionnaire (SRQ–P). For each of three behaviors, cooperating with a classmate (e.g., "Why would you cooperate with a classmate in this class?"), helping (e.g., "Why would you help a classmate in distress or having a lot of troubles?"), and following rules (e.g., "Why would you follow rules in the classroom?"), students were asked to respond to each of 17 reasons for behavior (1=never the reason, 4=always the reason; see Table 1 for items). Reasons represented four types of behavioral regulation: external (four items), M(SD)/α=2.28(.63)/.78); introjected—self‐focused (three items), M(SD)/α=2.15(.81)/.86); introjected—other‐focused (three items), M(SD)/α=2.51(.81)/.79); and internal (seven items), M(SD)/α=2.73(.72)/.87).

1 
 Reasons for Prosocial Behavior: Results of Principal Components Analyses (PCAs)

1 Note. Factor loadings for each item are shown. N for Cooperating, Helping, and Following Rules PCAs=339.

A principal components analysis (Varimax rotation) confirmed the four types of reasons for each of the three behaviors, with the four factors explaining 65%, 67%, and 69% of the variance in reasons for cooperating, helping, and following rules, respectively (see Table 1). Given the consistent factor structure, items were averaged to form four scores (one for each reason) for each aspect of behavior. Because scores for each reason were correlated significantly across behaviors, they were then averaged to create composite scores for each type of reason (external: rs=.60,.27, and.45, p<.001 for cooperating/helping, helping/rules, and cooperating/rules, respectively; self‐focused: rs=.70,.70, and.62, p<.001, for cooperating/helping, helping/rules, and cooperating/rules, respectively; other‐focused: rs=.80,.79, and.73, p<.001, for cooperating/helping, helping/rules, and cooperating/rules, respectively; and internal: rs=.71,.67, and.70, p<.001, for cooperating/helping, helping/rules, and cooperating/rules, respectively).

Expectations for prosocial behavior. Items adapted from the Cooperation subscale of the Classroom Life Measure ([26]) assessed perceptions of teachers' and peers' expectations to behave positively in class (e.g., "In this class, the teacher wants me to share my ideas and materials with other students"; 1=never, 5=always). M(SD)/α of three averaged peer items=3.32(.88)/.76 and of three averaged teacher items=3.63(.80)/.66.

Social goal pursuit. Social goal pursuit was operationalized as prosocial outcomes that students try to accomplish in the classroom (see, [45]). Fourteen items asked students about efforts to share and help peers with problems (e.g., "How often do you try to share what you've learned with your classmates?"), and to follow classroom rules and keep social commitments (e.g., "How often do you try to do what your teacher asks you to do?"; 1=never and 5=always. Scores were averaged to form a social goal pursuit score; M(SD)/α=3.78(.56)/.83.

Prosocial behavior. Peer nominations and teacher ratings were used to assess prosocial behavior. Peer nomination scores were obtained by asking students: "Who shares and cooperates?" and "Who helps other kids when they have a problem?" For each behavior, students were given a list of their same‐sex classmates participating in the study. Students were asked to cross out their own name and then place a check mark in front of the names of those classmates who display the behavior most of the time. They were instructed to check as many or as few names as they liked. Because data collection took place at the end of the Spring semester and students were part of teams in which students took several classes together, it is reasonable to assume that the students knew each other well enough to make valid nominations. For each behavior, the percentage of nominations received was computed by dividing the number of nominations received by the number of names on the roster. The two scores were related (r=.88, p<.001) and therefore, averaged to form a nomination score. Teachers rated each student on the frequency of the same two behaviors in their class (1=never, 5=always). The two rating scores were correlated (r=.51, p<.001) and therefore, averaged to form a prosocial rating score. The nomination and rating scores were related (r=.43, p<.001) and therefore, were standardized within classroom and averaged to form a composite prosocial behavior score.

Background information. Students completed a general information sheet that asked them their sex and race; for race, students were asked to check one of the following: Caucasian, African American, Hispanic, Asian, and Other. These categories were recommended by the school principal as being representative of the student population and familiar terms to students.

Results are presented in two sections, descriptive relations among variables and results of regression analyses. The sample for the following analyses consisted of 263 students: of the total sample, 279 students had complete data (due to unforeseen time constraints, 53 students in three classes were unable to complete the Perceived Social Competence scale), and because most of the students self‐reported as either Caucasian or African American, data obtained from students in the "other" category (n=26) were excluded from analyses. A series of one‐way analyses of variance (ANOVAs) compared scores of the 263 students with those of the excluded students; no significant differences on any of the study's variables were found.

Correlations. In support of our hypothesized pathways were significant, positive correlations between prosocial goal pursuit and prosocial behavior, and between goal pursuit and each predictor except depressive affect (see Table 2). Behavior also was related significantly to each predictor except other‐focused and external reasons for behavior, and reasons were related significantly to other self‐processes and social cues. Finally, empathy and perspective taking were related significantly to internal and introjected reasons but not to external reasons, and distress was related significantly and positively to other‐focused and external reasons. Contrary to predictions, teacher and peer expectations were related positively to each of the four reasons.

2 
 Intercorrelations Among Variables

2 Note.n=263. 
 *p<.05 ; **p<.01 ; ***p<.001.

Group differences. 2(sex) × 2(grade level) × 2(race) ANOVAs assessed main effects of sex, grade level, and race for each variable. Compared with boys, girls reported significantly higher levels of empathy, stronger perceived social competence, more frequent internal and self‐focused reasons for behavior, higher levels of perceived expectations from teachers and peers, more frequent social goal pursuit, and were seen by teachers and peers as displaying more prosocial behavior (see Table 3). For all variables except depressive affect, perceived expectations from teachers, and perspective taking, Caucasian students' scores were significantly higher than those of African American students. Finally, sixth graders reported more frequent external reasons for behavior than did eighth graders. One interaction was significant: Caucasian females (n=77) reported more frequent self‐focused reasons for behavior than did African American females (n=64); similar differences were not found for males, F(1, 262) =4.92, p<.05, η=.14; Ms(SDs)=2.47(.88) and 1.99(.73) for Caucasian and African American females, and 2.06(.79) and 2.02(.79) for Caucasian (n=65) and African American males (n=57), respectively.

3 
 Variables as a Function of Sex, Grade Level, and Race: Results of ANOVAs

3 Note.n=263. 
 *p<.05 ; **p<.01 ; ***p<.001.

We examined differences in prosocial behavior as a function of race further by taking into account classroom composition. Of the 18 classrooms participating in this study, 8 consisted of more African American students than Caucasian students and 8 consisted of more Caucasian students than African American students (in all cases, the majority group included at least three more students than the minority group). Therefore, we compared behavior scores as a function of race for students in each set of classes: significant differences were found in the Caucasian‐majority classrooms such that Caucasian behavior scores were higher than African American scores, t(172)=2.48, p<.01; Ms(SDs)=0.13(.63) and −0.18(.82) for Caucasian and African American students, respectively. In contrast, prosocial behavior did not differ as a function of race in the African American majority classrooms, t(105)=−0.69, ns; Ms(SDs)=−0.02(.75) and 0.11(.72) for Caucasian and African American students, respectively.

We hypothesized that self‐processes and contextual cues would be related to prosocial behavior by way of goal pursuit and that empathy, distress, perspective taking, and expectations from teachers and peers would be related to goal pursuit in part, by way of reasons for prosocial behavior. We also predicted that differences in prosocial behavior as a function of sex and race would be mediated partly by self‐processes and contextual cues. To test these mediation hypotheses, we conducted a series of multiple‐regression analyses. We included Sex × Depressive Affect and Race × Peer Expectation interaction terms in the models to test our moderation hypotheses. Variables were mean centered to control for possible effects of collinearity.

Predictors of prosocial behavior. We hypothesized that students' self‐processes and social perceptions would be related to prosocial behavior by way of prosocial goal pursuit. To examine this, prosocial behavior was regressed on self‐processes, contextual cues, and prosocial goal pursuit. Demographic variables were entered at the first step, self‐processes and contextual cues at the second step, prosocial goal pursuit at the third step, and the Sex × Depressive Affect and Race × Peer Expectation interaction terms at the last step. Depressive affect, perceived expectations from peers, and prosocial goal pursuit were significant predictors of prosocial behavior at the last step (see Table 4). These results established prosocial goal pursuit as a direct predictor of prosocial behavior when taking into account other self‐processes and social cues.

4 
 Predictors of Prosocial Behavior: Multiple‐Regression Analyses

4 Note.n=263. Sex was coded such that 0=male and 1=female; race was coded such that 0=Caucasian and 1=African American. 
 *p<.05 ; **p<.01 ; ***p<.001.

Results also supported our mediation hypotheses for sex and race. Specifically, the decreasing beta weight for sex at Step 2 combined with significant relations between sex and peer expectations (see Table 3), and between prosocial behavior and peer expectations (see Table 4) met the conditions for mediation ([4]). A follow‐up [38] test indicated that the indirect effect of sex on prosocial behavior by way of perceived peer expectations was significant (z=3.58, p<.001). Similarly, the decreasing beta weight for race at Step 2 combined with significant relations between race and peer expectations (see Table 3), and between prosocial behavior and peer expectations (shown in Table 4) also met the conditions for mediation. A follow‐up Sobel test indicated that the indirect effect of race on prosocial behavior by way of peer expectations was significant (z=−2.52, p<.01).

Finally, the two significant interaction terms supported our moderation hypotheses. Slopes for depressive affect and prosocial behavior differed significantly for boys and girls (β=.15, t=1.95, p<.05). Analysis of simple slopes within groups ([1]) indicated that depressive affect was a significant predictor of behavior for boys (β=−.28, t=3.51, p<.001) but not for girls (β=−.08, t=−0.17). Further examination of the data indicated a curvilinear relation between depressive affect and behavior for boys, such that behavior scores of boys with depressive affect scores.5 SD above the mean were significantly lower than those of boys with lower depressive affect scores; a linear trend was evident for girls. In addition, slopes for expectations from peers and prosocial behavior were significantly different for Caucasian and African American students (β=−.14, t=−1.90, p<.05). The simple slope was significant for Caucasian (β=.28, t=2.67, p<.001) but not for African American students (β=.06, t=−0.21).

Predictors of prosocial goal pursuit. We next regressed prosocial goal pursuit on the demographic variables (Step 1), depressive affect, perceived competence, perspective taking, empathy, and teacher and peer expectations (Step 2), and reasons for behavior (Step 3). Because the Sex × Depressive Affect and Race × Peer Expectation interaction terms were significant predictors of prosocial behavior, we included them as possible predictors of prosocial goal pursuit at the last step (see Table 5). Perceived competence, perspective taking, peer expectations, and internal, other‐focused, and external reasons for behavior were significant predictors of prosocial goal pursuit, indicating that these variables were related to prosocial behavior indirectly by way of their relation to goal pursuit. The interaction terms were not significant predictors of prosocial goal pursuit.

5 
 Predictors of Prosocial Goal Pursuit: Multiple‐Regression Analyses

5 Note.n=263. Sex was coded such that 0=male and 1=female; race was coded such that 0=Caucasian and 1=African American. 
 *p<.05 ; **p<.01 ; ***p<.001.

Our mediation hypotheses for sex and race were confirmed in that the decreasing beta weight for sex at the second step combined with significant relations between sex and empathy, perceived competence, perspective taking, and peer expectations (see Table 3), and between goal pursuit and empathy, perceived competence, perspective taking, and peer expectations (see Table 5) met the conditions for partial mediation. Sobel tests indicated significant indirect effects of sex on goal pursuit by way of empathy (z=3.60, p<.001), perceived competence (z=2.43, p<.05), and peer expectations (z=4.59, p<.001). Also, the decreasing beta weight for race at Step 2 combined with significant relations between race and empathy, perceived competence, and peer expectations (see Table 3), and between goal pursuit and empathy, perceived competence, and peer expectations (shown in Table 5) met the conditions for partial mediation. Sobel tests indicated significant (p<.01) indirect effects of race on goal pursuit by way of empathy (z=−3.33), perceived competence (z=−4.37), and perceived peer expectations (z=−2.69).

Predictors of reasons for prosocial behavior. Finally, each reason for behavior was regressed on the demographic variables, self‐processes, and contextual cues (see Table 6). Perceived social competence, empathy, perspective taking, and teacher and peer expectations for behavior were significant predictors of internal reasons for prosocial behavior; depressive affect, perspective taking, empathy, and perceived expectations from peers were significant predictors of self‐focused reasons; depressive affect, perceived expectations from peers and sex, race, and grade level were significant predictors of other‐focused reasons; and grade level, race, depressive affect, and perceived expectations from teachers were significant predictors of external reasons.

6 
 Predictors of Reasons for Prosocial Behavior: Multiple‐Regression Analyses

6 Note.n=263. Sex was coded such that 0=male and 1=female; race was coded such that 0=Caucasian and 1=African American; and grade was coded such that sixth grade=1 and eighth grade=2. 
 *p<.05 ; **p<.01 ; ***p<.001.

These findings provide evidence that reasons for behavior provide a pathway that links self‐processes and social cues to prosocial goal pursuit. Specifically, depressive affect was related to prosocial goal pursuit indirectly, by way of its significant relations with external and other‐focused reasons; perspective taking, empathy and teacher and peer expectations by way of significant relations with internal reasons; and peer expectations by way of other‐focused reasons. Also as predicted, empathy was related most strongly to internal reasons for behavior and unrelated to external reasons, depressive affect was related to external but not internal reasons for behavior, expectations of peers was related to internal but not external reasons, and expectations of teachers was related to external reasons. Contrary to predictions was a significant relation between expectations from teachers and internal reasons for behavior.

Mediation effects also were evident. For self‐focused reasons, the decreasing beta weight for sex at the second step of the model combined with significant relations between sex and empathy, perspective taking, and peer expectations (see Table 3), and between self‐focused reasons and these same variables (see Table 7) met the conditions for partial mediation. Follow‐up Sobel tests indicated significant (p<.001) indirect effects of sex on self‐focused reasons by way of empathy (z=3.39), perspective taking (z=4.59), and perceived peer expectations (z=3.78). Similarly, for internal reasons, the decreasing beta weight for sex at the second step combined with significant relations between sex and empathy, perceived competence, perspective taking, peer expectations, and teacher expectations (see Table 3), and between internal reasons and these same self‐processes and social cues (shown in Table 7) met the conditions for partial mediation. Sobel tests indicated indirect effects (p<.01) of sex on internal reasons by way of empathy (z=3.75), perceived competence (z=2.44), perspective taking (z=5.19), perceived peer expectations (z=4.28), and teacher expectations (z=4.25).

Finally, for self‐focused reasons, the decreasing beta weight for race at the second step of the model combined with significant relations between race and empathy and peer expectations (see Table 3), and between self‐focused reasons and empathy and peer expectations (shown in Table 7) met the conditions for partial mediation. Sobel tests indicated significant (p<.01) indirect effects of race on self‐focused reasons by way of empathy (z=−3.10) and perceived peer expectations (z=−2.47). For internal reasons, the decreasing beta weight for race at the second step combined with significant relations between race and empathy, perceived competence, and peer expectations (see Table 3), and between internal reasons and these same variables (shown in Table 7) also met the conditions for partial mediation. Sobel tests indicated indirect effects (p<.001) of race on internal reasons by way of empathy (z=−3.42), perceived competence (z=−4.09), and perceived peer expectations (z=−4.32).

The current study was designed to address two general questions: How are goals (to help, share and cooperate), supporting self‐processes (reasons for behavior, empathy, distress, perspective taking, and perceived competence), and contextual cues (perceived expectations from teachers and peers) related to prosocial behavior? And, how are sex, race, and grade level related to prosocial behavior? Our findings concerning each of these questions are discussed in turn.

Our conceptual model is based on the premise that prosocial behavior is motivated by a complex set of psychological processes including goal pursuit, reasons for behavior, empathy, depressive affect, perspective taking, and perceived competence, and contextual cues in the form of perceived expectations of teachers and peers. In support of this model, we found that pursuit of prosocial goals is an important mechanism that links prosocial actions to other psychological processes; pursuit of these goals was a direct predictor of prosocial behavior and the remaining self‐processes and contextual cues were related to behavior indirectly by way of their relations with goal pursuit. In line with previous work, perceived competence and perspective taking appeared to play a particularly important role in adolescents' pursuit of prosocial goals (see also, [9], [17]). The significant role of peer expectations in predicting goal pursuit also supports the notion that peers play a central role in motivating adolescents' displays of positive behavior at school ([52]). Finally, our findings also confirm the often‐cited belief ([18]) that individuals display prosocial behavior for altruistic and intrinsically motivated reasons but also for reasons extrinsic to the self; adolescents were likely to pursue prosocial goals not only because they thought it was important to do so but also because of the social costs associated with not doing so.

The role of personal goals has not received widespread attention in studies of prosocial behavior. Therefore, our results extend the literature on prosocial goal setting and behavior in several ways. At a general level, our findings suggest that an examination of adolescents' social goals can provide important insights into their behavior at school. Indeed, prosocial goal pursuit proved to be a significant predictor of prosocial behavior, even when taking into account a larger set of self‐processes and social cues relevant for understanding social competence. In addition, more global goals as expressed in identified, external, and other‐focused reasons for behaving in a prosocial manner were significant predictors of prosocial goal pursuit. These multiple reasons for goal pursuit also served as pathways by which other processes were related to goal pursuit, with each reason for goal pursuit being related to a unique set of predictors.

As expected, reports of internal reasons for behaving in a prosocial manner were predicted by empathy and perspective taking, processes traditionally believed to promote altruistic forms of prosocial behavior, and depressive affect was related to external, other‐focused, and self‐focused reasons. In addition, perceived competence was related only to internal reasons for behavior. It is possible that external reasons for behavior are explained to a greater degree by beliefs that prosocial actions are governed by outside forces than by beliefs about one's competence, whereas internal reasons reflect a sense of personal control and therefore, a belief that one's abilities will influence behavior ([11]). Another possibility is that the measure used to assess perceived competence taps beliefs about what students actually do (e.g., "Some teenagers usually do the right thing") rather than beliefs about ability to behave prosocially. Inclusion of measures of perceived control (e.g., [11]) as well as alternative measures of perceived competence in future studies would address this issue.

Contextual cues also were related differentially to reasons for prosocial behavior suggesting that to some extent, goal setting is situation specific and subject to social influence. As predicted, perceived expectations of teachers for prosocial behavior (i.e., expectations from more powerful individuals) were related to reasons based on perceived threats of punishment and retribution. Unexpectedly, perceived expectations of teachers also were related to internal reasons. In this case, it might be that students who hold internalized values for prosocial behavior interpret social expectations for such behavior, whether from teachers or from peers, as merely reflecting their own values. In contrast, peer expectations predicted internal, self‐focused, and other‐focused reasons for behavior. These findings support the notion that getting along with peers presents a strong social motive for behaving in positive ways, but also that reactions to peer expectations for positive behavior are often likely met with empathy and internalized motives to help others who are similar to oneself ([6]).

Future work in this area might profit from the inclusion of additional reasons for prosocial behavior, such as to gain resources or to control others (see [23]). In addition, consideration of ways in which goal hierarchies reflecting the content of adolescents' goals (i.e., what they would like to achieve) in relation to more global goals (i.e., reasons why they want to achieve them) is rare in the literature on social competence. However, this perspective on social goal setting also might provide valuable insights into the motivational underpinnings of other aspects of social functioning such as the quality of students' relationships with peers and teachers. On a methodological note, the nearly identical factor structures for the four reasons across three types of prosocial behavior provide further confirmation of the validity of this approach for assessing adolescents' reasons for their prosocial actions.

Relations of sex, race, and grade level to prosocial behavior and its predictors also were examined in this study. We found few grade‐level differences although as predicted, sixth graders reported having external reasons for prosocial behavior more often than did eighth graders. Grade level also was a significant predictor of external and other‐focused reasons for behavior (Table 6), such that sixth graders reported these reasons as being more frequent than did older students. Together, these findings suggest that sixth graders perceive stronger social influences on their behavior than do older students, and they are in line with other reports that teachers of sixth graders often have a heightened concern with controlling students' classroom behavior when compared with other teachers ([16]). In general, therefore, it appears that few changes in self‐processes, perceptions of the classroom social environment, and prosocial behavior take place across the middle school years. However, further elaboration of our model to include developmental processes that might influence moral reasoning and other processes related to prosocial behavior ([18]) would provide impetus for longitudinal studies of children as they grow into the adolescent years.

In addition, mean differences in prosocial behavior, goal pursuit, self‐processes, and contextual cues were found as a function of sex and race. In terms of overall effect sizes, the differences between boys and girls were consistently stronger than those between Caucasian and African American students. As noted by other researchers, these group differences are likely to reflect in part, measurement issues such as social desirability, how items are interpreted, and the source of information ([20]). With regard to the latter, peer nominations were made by same‐sex informants to insure that familiarity effects and gender‐specific styles of cooperation and helping ([15]) did not influence scores. Owing to limitations of sample size, we could not follow this procedure for race; the number of peer nominators for some students would have been reduced to one or two students. However, the potential importance of informant characteristics is borne out in our post hoc analyses of race differences in prosocial behavior. Our findings indicate that the composition of classrooms with regard to majority and minority status of peer nominators can have a significant impact on behavior scores of African American students.

Although these findings require replication, a clear implication is that unless within‐group nomination procedures are employed when studying racially diverse populations, valid inferences concerning significant group differences cannot be made (see also [28]). Indeed, researchers must consider the possibility that significant group differences are the result of measurement strategies and the unique perspectives on social behavior that students of various cultural backgrounds bring to the classroom (e.g., [5]). In this regard, the potential for school‐level factors to influence student behavior ([46]) also deserves further attention. Finally, we did not identify students who might have come from multiracial backgrounds. The extent to which exposure to multiple cultures and worldviews might influence students' social experiences at school also deserves further study.

Our findings concerning mediation and moderation also provide compelling evidence that the development of research hypotheses and designs aimed at identifying more precisely the social and psychological underpinnings of significant group differences as a function of sex and race are needed. As predicted, significant relations between prosocial outcomes and sex and race could be explained in part by other variables. In most cases, the strength of relations between sex, race, and prosocial outcomes decreased significantly when self‐processes and social cues were taken into account, although the relations remained statistically significant. However, significant relations between sex and race and internal reasons for goal pursuit became nonsignificant when other psychological variables were introduced into the model. In short, these findings support the notion that significant differences in prosocial behavior, goal pursuit, and certain reasons for goal pursuit as a function of sex and race cannot be attributed solely to group membership but to other mechanisms associated with psychological functioning.

Counter to predictions, relations between sex and race and other‐focused reasons for behavior and the significant relation between race and external reasons for behavior were not explained by other variables. On the one hand, it is possible that adolescent boys and African American students are socialized to seek social approval more frequently than girls and Caucasian students, and that external rewards or punishments to students are dispensed differentially as a function of students' race. On the other hand, other variables not assessed in this study might explain these significant relations. In particular, perceptions of control ([11]), social identity ([35]), and attitudes toward authority and compliance ([39]) might be important variables to examine in this regard.

In addition, our predictions concerning moderation were supported (albeit weakly) in that relations between depressive affect and prosocial behavior differed for girls and boys, and relations between peer expectations and behavior differed for African American and Caucasian students. In the case of depressive affect and prosocial behavior, our findings suggest that whereas for girls, depressive affect was related to prosocial behavior in linear fashion, the relation for boys was curvilinear. It has been argued that significant relations between girls' depressive affect and prosocial behavior might reflect girls' heightened concerns with compliance (e.g., [27]). Although speculative, it is possible that difficulties with emotional regulation in boys are manifest more systematically in externalizing types of behavior such as aggression and acting out and that only in extreme cases are they borne out in decreased levels of prosocial behavior. Both of these possibilities, along with a focus on specific socialization processes that might contribute to them deserve further study.

The fact that perceived expectations from peers was related to prosocial behavior for Caucasian, but not African American students also raises intriguing issues about the impact of adolescent peer groups on prosocial behavior at school. As suggested by [41], our findings might reflect the fact that African American students who attend schools in which they are a racial minority do not experience the same differentiated crowd structures as do Caucasian students and therefore, are not influenced by peer norms to the same degree or in the same ways. Research in schools where African American students represent the majority population and where they form norm‐based peer groups would be the next step in determining social‐contextual factors that might explain further the role of peer expectations in promoting positive classroom behavior. In addition, theoretical perspectives that address racial minority students' social experiences at school more directly (e.g., [34], [39], [51]) would provide a more focused foundation for future work in this area.

Limitations of the current study provide clear directions for future research. First, our data are cross sectional. Therefore, research that follows individual students over the course of middle school is needed to determine the extent to which the effects of contextual factors in sixth grade are fleeting or have a significant impact on later prosocial actions. In a related vein, the possibility that the positive social feedback associated with prosocial behavior might influence the development of self‐processes and the interpretation of contextual cues should not be discounted. To illustrate, it is likely that positive feedback in the form of social approval and acceptance from classmates is likely to increase the degree to which students are empathic toward peers, take their perspective, experience a positive sense of emotional well‐being, and perceive positive expectations for continued prosocial behavior. Longitudinal and experimental work that can identify specific causal mechanisms is clearly needed in this regard.

Several assumptions concerning the nature of prosocial behavior also warrant discussion. Specifically, we adopted the perspective that adolescent students display prosocial behavior in consistent fashion and therefore, that it occurs in large part as a function of other fairly stable self‐processes. However, we also acknowledged the role of context‐specific expectations for behavior that are likely to differ from classroom to classroom due to norms and values communicated by specific teachers and groups of peers. Our findings underscore the importance of these cues in predicting adolescents' behavior and warrant further investigations concerning how these expectations are communicated and enforced in individual classroom settings.

In line with our classroom‐specific approach to understanding prosocial behavior, we did not consider the potential impact of parental influences on adolescents' behavior. Social developmental perspectives suggest that parents who encourage perspective taking and evoke empathic responses to the distress of others, in conjunction with threats of punishment or love withdrawal, are likely to promote the internalization of altruistic values in their children ([24]). Assuming that inductive reasoning occurs frequently with respect to peer‐oriented behavior, it follows that children who have internalized prosocial values as a result of these parenting practices will be especially sensitive to the consequences of their behavior on their peers when deciding on a prosocial course of action. However, the degree to which adolescents' prosocial behavior is the outcome of ongoing parental influence or a situation‐specific reaction to peers remains an important, unanswered question for future research.

We also assumed that reasons for behavior would not be stable characteristics representing levels of internalization (cf., [37]) but rather, that students could have multiple reasons for behavior that are subject to social influence. In this regard, it would be informative to examine individual differences in the degree to which students adopt specific reasons for their actions on a consistent basis. Follow‐up analyses of our data indicated that subsets of students reported primarily internal (n=32) or primarily external (n=28) reasons for each type of behavior with high frequency. However, the majority of students did not respond with this degree of consistency. It is logical to assume that self‐reports of reasons for behavior are the most valid way to assess this process. However, alternative strategies that assess students' reasons for behavior under varying circumstances (e.g., presenting them with a range of situational cues that might cause them to change their reasons for behavior) might provide additional insights into the consistency with which students adopt specific reasons for behavior.

In conclusion, the current study provides new insights into the social and psychological processes related to young adolescents' prosocial behavior at school. Our findings provide support for the notion that prosocial behavior is motivated by a complex set of self‐processes and contextual cues, with goal pursuit providing a pathway that links these processes to prosocial behavior. Our results also indicate that goal pursuit is motivated by external, other‐focused, and internalized reasons, with variables typically studied in relation to prosocial behavior such as empathy and perspective taking being related significantly only to internalized reasons for behavior. Finally, the effects of sex and race on prosocial outcomes appear to be complex, explained in part by methodological factors and specific aspects of psychological functioning. Continued examination of adolescents' multiple reasons for prosocial behavior and their motivational antecedents, further clarification of the effects of sex and race on prosocial behavior, and employment of longitudinal and experimental designs would extend the current findings in interesting and important directions.