Content uploaded by Karin Frey
Author content
All content in this area was uploaded by Karin Frey on Mar 10, 2016
Content may be subject to copyright.
Effects of a school-based social–emotional competence program:
Linking children’s goals, attributions, and behavior
Karin S. Frey
a,b,
*
, Susan Bobbitt Nolen
a
,
Leihua Van Schoiack Edstrom
b
, Miriam K. Hirschstein
b
a
College of Education, University of Washington, United States
b
Research and Evaluation Department, Committee for Children, Seattle, Washington, United States
Abstract
This study examined the effects of the Second Step social–emotional learning program and addressed the
relations between social cognitions and prosocial and antisocial behavior. Children (N = 1,253) in intervention and
control groups were assessed by teacher ratings, self report, and observation in two conflict situations. Intervention
children were more likely to prefer prosocial goals and give egalitarian reasons for satisfaction than control
children. Intervention children also required less adult intervention, and behaved less aggressively and (among
girls) more cooperatively. Teacher ratings of social behavior showed improvement over time. Individual and
dyadic behavior varied as a function of goals, hostile attributions, and attitude concordance within dyads. Findings
are discussed with respect to social-cognitive models of aggression and prosocial behavior.
D 2005 Elsevier Inc. All rights reserved.
Keywords: Prosocial; Aggression; Intervention; Goals; Attributions
1. Introduction
Teachers spend a considerable amount of time mediating disputes between students (Johnson &
Johnson, 1996). Students’ conflicts often elicit aggressive, oppositional behaviors directed toward peers
and adults. Such behaviors compromise the learning environment and are associated with later conduct
0193-3973/$ - see front matter D 2005 Elsevier Inc. All rights reserved.
doi:10.1016/j.appdev.2004.12.002
* Corresponding author. Educational Psychology, Box 3600, University of Washington, Seattle, WA 98195-3600, United
States.
E-mail address: kfrey@cfchildren.org (K.S. Frey).
Applied Developmental Psychology 26 (2005) 171–200
problems, substance use, and school failure (Coie, Lochman, Terry, & Hyman, 1992; Miller-Johnson,
Lochman, Coie, Terry, & Hyman, 1998; Pepler & Rubin, 1991). A growing body of evidence suggests
that programs that focus on promoting protective factors and mitigating risk can reduce multiple problem
behaviors (see reviews by Durlak, 1995; Greenberg, Domitrovich, & Bumbarger, 2001).
Like antisocial behavior, prosocial behavior has important implications for the social and school-
related adjustment of children. Whereas aggressive and oppositional behavior are associated with school
problems, prosocial behavior (e.g., empathic, socially responsible behavior) has been linked to higher
grades and test scores (see reviews by Wentzel, 1996; Wentzel & Wigfield, 1998), peer acceptance
(Wentzel & Erdley, 1993), and high status (Coie, Dodge, & Kupersmidt, 1990). Cognitively-based
models of social–emotional learning (Bandura, 1986; Crick & Dodge, 1994; Kendall, 1993) posit that
similar cognitive and affective processes underlie positive and negative social behaviors. Like aggression
(see review by Coie & Dodge, 1998), prosocial behavior shows considerable stability from preschool to
young adulthood (Eisenberg et al., 1999), suggesting that early intervention might provide long-term
benefits. Elementary school programs under various labels (e.g., social competence, violence prevention,
conflict resolution training and character education) have emerged to promote socially responsible
behavior and discourage aggressive behavior.
1.1. Promoting prosocial behavior: The role of motivation and social cognition
Researchers have long realized that behavior deficits may be due to either an inability to perform
competently or insufficient motivation to do so (Bandura, 1986). When applied to interventions, this
suggests that effecting behavioral change may require alterations in motivation, in addition to promotion
of social skills. Social cognitions such as goals, beliefs, and attributions appear to play a strong
motivational role with respect to behavior.
1.1.1. Social goals
Erdley and Asher (1999) have argued that teaching cognitive, affective, and behavioral skills will be
insufficient to promote positive behavior unless training also promotes the adoption of prosocial goals.
Few investigations have addressed the socialization of goals, however (see review by Wentzel, 1996),
and there is no consensus regarding the likelihood of changing children’s goals through intervention.
Working with adolescents, Lochman, Wayland, and White (1993) conclude that social goals are
enduring characteristics that are not amenable to change by psychoeducational interventions. In contrast,
Jarvinen and Nicholls (1996) suggest that changing beliefs about the behaviors that foster successful
relationships will lead to changes in social goals. Research is needed to determine whether school-based
interventions can encourage the adoption of positive goals, and whether those goals are related to
positive school behavior.
Goal preferences are associated with important, real-world outcomes. Children who endorse prosocial
goals in conflict situations have higher levels of social functioning (Murphy & Eisenberg, 1996) and
more friends than children who endorse retaliation goals (Rose & Asher, 1999). Control, domination,
and hostile goals are typical of aggressive children (Boldizar, Perry, & Perry, 1989; Erdley & Asher,
1996, 1999; Slaby & Guerra, 1988) and predictive of substance use, self-reported crime, peer rejection,
depression, and inattentive behavior (Lochman et al., 1993). In addition, Stein and Albro’s (2001) goal-
based model of argument provides a framework for understanding how social goals can influence
children’s approaches to interpersonal conflict. An arguer who believes that maintenance of a
K.S. Frey et al. / Applied Developmental Psychology 26 (2005) 171–200172
relationship is more important than bwinningQ should be motivated to compromise or submit. When
relationship goals are weaker, the arguer’s actions are likely to emphasize instrumental gains or the
establishment of dominance. Indeed, a preference for relationship maintenance goals is associated with a
preference for prosocial strategies to resolve conflict, whereas self-interested, controlling, and retaliatory
goals are linked to hostile or aggressive strategies (Chung & Asher, 1996; Delveaux & Daniels, 2000).
These findings suggest that social goals are important targets for intervention.
1.1.2. Attributions of hostile goals
Social attributions are another cognitive construct believed to motivate behavior. There is strong
evidence that the beliefs individuals have about other people’s goals and intentions are important
predictors of emotions and behavior. Individuals who attribute others’ actions to hostile intent tend to
react with blame, anger, and retributive aggression when they are injured, compared to those who infer
benign intent or withhold judgment (see Crick & Dodge, 1996 for a review; and Orobio de Castro,
Veerman, Koops, Bosch, & Monshouwer, 2002 for a meta-analysis). Interventions have shown promise
in reducing hostile attributions and aggression among boys. Sixth-grade aggressive boys placed in a
frustration-inducing communication task were less likely to infer hostile intent, less angry, and less
verbally aggressive if they had been in a social attribution retraining group. They were rated by teachers
as less aggressive, although not more socially skilled than their counterparts in the control group (Hudley
& Graham, 1993). A subsequent study with third- to sixth-grade aggressive and non-aggressive boys
also showed immediate post-intervention reductions in attributions of hostile goals compared to controls,
although these improvements were not maintained at a 12-month follow-up. Teacher ratings of social
skills showed no group differences (Hudley et al., 1998). Research is needed to determine the efficacy of
social attribution retraining with girls and effectiveness of a school-implemented program with both boys
and girls.
1.2. The present study
The first goal of the study was to evaluate the impact of the Second Step program (Committee for
Children, 1992, 1997), a universal social–emotional intervention, on students’ behavior, social
cognitions, and affect. The second goal of the study was to test the conceptual basis of the program
by examining relations among behavior and motivational constructs. The program is based on the
understanding that behaviors are influenced by goals, beliefs, and emotions, as well as information-
processing and performance skills (Crick & Dodge, 1994; Lemerise & Arsenio, 2000). The Second
Step program is designed to both decrease aggressive behavior and increase empathic, socially
responsible behavior by (a) fostering children’s cognitive, emotional, and behavioral skills; (b)
reducing maladaptive beliefs about aggression (Slaby & Guerra, 1988); and (c) promoting positive
social goals and values. The basic methodology has been translated into a developmentally
sequenced set of activities for preschool through middle school.
Studies using trained observers of preschool classrooms (McMahon, Washburn, Felix, Yakin, &
Childrey, 2000) and elementary school playgrounds (Grossman et al., 1997) have found decreased levels
of physically aggressive or disruptive behavior in children participating in the Second Step program. The
playground observations were undertaken as part of an experimental study and also showed more
positive social behaviors among Second Step participants relative to those in control schools. In contrast,
teacher reports of social competence showed no improvements in either of these studies. Finally, middle
K.S. Frey et al. / Applied Developmental Psychology 26 (2005) 171–200 173
and junior high school students reported that social–emotional skills were easier to perform after
intervention (Van Schoiack-Edstrom, Frey, & Beland, 2002). Moreover, their normative beliefs
(Huesman & Guerra, 1997) were less supportive of physical, verbal, or relational aggression than those
of students in the comparison group.
These studies have been limited by their exclusive focus on either behavior or beliefs. Without
measuring both simultaneously, researchers were unable to address conceptual questions about the re-
lationship of social cognitions to behavior. The present study (a) tests the specific effects of the
intervention on social goals, attributions, satisfaction, and behavior in several contexts; and (b) examines
the relations among goals, attributions, and prosocial and antisocial behavior, both individually and
in dyads.
1.2.1. The intervention
The Second Step: program, consists of commercially available curriculum materials (Committee
for Children, 1997a,b), professional staff training, and staff training materials. It translates
techniques developed by cognitive behavior therapists and others (e.g., Kendall, 1993; Spivack &
Shure, 1974) into an educational framework consisting of classroom activities and generalization
efforts throughout the day (Frey, Hirschstein, & Guzzo, 2000). Lessons lasting 25–40 minutes
(depending on grade level) are presented by classroom teachers. Using suggested lesson scripts,
teachers introduce key concepts through questions stimulated by photo cards or videotaped stories.
Performance-based instruction and cognitive interventions are used to foster development in
thoughts, feelings, and behavior (Kendall, 1993). Questions are designed to promote perspective-
taking and, as the lesson progresses, elicit specific strategies for dealing with the illustrated
situations. Teachers and videotapes provide models of the key skills. Children practice specific self-
regulatory strategies and behavioral skills with role-playing and other classroom activities. Strategies
for cueing, coaching, and acknowledging the targeted behaviors are included in program materials,
as well as suggestions for integrating content with the academic program.
There are three units in the program: Empathy Training, Impulse Control and Problem Solving, and
Anger Management. Empathy lessons teach children to notice and interpret relevant contextual and
expressive cues (Halberstadt, Denham, & Dunsmore, 2001), including those indicative of benign as
opposed to hostile intent (Dodge & Newman, 1981). Emotional understanding, prediction, and
communication are taught as core skills. In the second unit, children repeatedly practice generating and
evaluating solutions to social problems. Positive goals such as safety, fairness, efficacy, and the social–
emotional benefits of mutually rewarding interaction serve as evaluation criteria. Discussions help
identify the behaviors that help children sustain enjoyable play and those that interfere. Cognitive-
behavioral techniques such as self-talk (Kendall, 1993; Meichenbaum, 1977) and attention control
(Derryberry & Rothbart, 1988; Metcalfe & Mischel, 1999) are emphasized in the Anger Management
unit. In all units, children practice specific behavioral skills that are meant to serve as building blocks for
social problem solving (e.g., resisting negative peer pressure, apologizing, showing appreciation).
1.2.2. Measurement strategy
We took a multilevel, multi-informant approach to assessing the effect of the intervention. We
measured behavior with teacher reports, self reports, and direct observations, the first two focusing
on individual behavior, the last one on dyadic behavior. Measurement of social cognitions and affect
employed hypothetical vignettes and in-the-moment interviews during structured conflicts.
K.S. Frey et al. / Applied Developmental Psychology 26 (2005) 171–200174
1.2.2.1. Observing behavior during structured conflicts. The use of direct observation is an important
addition to the more typical measures of program effectiveness (e.g., Reid, Eddy, Fetrow, &
Stoolmiller, 1999; Stoolmiller, Eddy, & Reid, 2000). In the present study, blinded observers coded
the interactions of children during structured conflict. Unlike conflicts in schoolyards, those elicited
by structured situations offer more challenges to the participants by restricting opportunities to leave
(Hartup, French, Laursen, Johnston, & Ogawa, 1993), thus increasing the frequency of typically low
base-rate behaviors such as aggression. Explicitly prosocial behaviors are also relatively infrequent
and can be difficult to distinguish from routinely sociable behaviors in a fast-paced environment like
a school playground (e.g., Grossman et al., 1997).
The two conflict situations used in the current study were structured to elicit either competition or
cooperation over the distribution of resources. The first involved negotiations between partners
regarding the choices they would make jointly during a bprisoner’s dilemmaQ game. Engaging in
mutually cooperative strategies during a dilemma is associated with friendship quality and positive
emotional tone in four-year-olds (Matsumoto, Haan, Yabrove, Theodorou, & Carney, 1986). The second,
more naturalistic task required the negotiated division of four bthank youQ gifts. Questions regarding the
control of resources elicit social dominance concerns in some children (Johnson & Johnson, 1996;
Shantz & Hobart, 1989), presenting a good opportunity to observe individual differences in aggression
and cooperative behavior—differences that we expected to favor the intervention group.
1.2.2.2. Measuring cognitions and affect in the context of structured conflicts. Structured conflicts
enable investigators to examine linkages between observed behavior, cognition, and affect. In their classic
monograph (Dodge, Pettit, McClaskey, & Brown, 1986), Dodge and colleagues used a staged provocation
to predict children’s aggressive responses from hostile attributions. Hudley and Graham (1993) used a
structured conflict to examine both the effects of attribution retraining and the applicability of attribution
theory (Weiner, 1985) to psychoeducational intervention. As predicted by the model, hostile attributions
were associated with greater anger and more hostile behavior. Negotiation paradigms offer a particularly
rich opportunity to examine complex interaction patterns and test conceptual models simultaneously.
In the current study, social cognitions were measured in conjunction with the observations through
individual interviews regarding goals, empathic reasoning, and satisfaction with the outcomes of the
prisoner’s dilemma game and prize division.
1.3. Hypothesized intervention effects
We expected that posttest comparisons between the intervention and control groups would show
benefits of the program with regard to behavior, cognitions, and affect. Table 1 summarizes the measures
and specific intervention hypotheses to be described below.
1.4. Hypothesized relationships among social cognitions, behaviors, and satisfaction
1.4.1. Goals and behavior
The current study tests the conceptual underpinnings of the Second Step program and extends
previous research by examining links between social cognitions and observed behavior. In line with the
previous discussion, we expected prosocial goals to predict less aggression, more cooperative behavior,
and more satisfaction in the observed conflict situations. It was less clear whether goals measured in the
K.S. Frey et al. / Applied Developmental Psychology 26 (2005) 171–200 175
context of a prisoner’s dilemma task would also predict teacher-reported social competence or student
behavioral intentions.
1.4.2. Goals and satisfaction
Goals help organize an individual’s emotional responses to a social outcome by determining which
behavioral outcomes are valued and which circumstances prove satisfying or disappointing (Crick &
Dodge, 1994; Jarvinen & Nicholls, 1996; Rubin & Krasnor, 1986). Children with prosocial and
egalitarian goals may have more opportunities to experience satisfaction than those operating in the zero-
sum context of dominance goals. Prosocial goals and behavior may also elicit reciprocal behavior from
peers, offering greater avenues for this kind of social reward. Thus, we expected a positive relationship
between situation-specific prosocial goals and subsequent satisfaction.
1.4.3. Hostile beliefs and behavior
Based on social-cognitive models and previous research, we anticipated that children with hostile
beliefs would be more antisocial and less socially competent, as rated by teachers; have more aggressive
intentions in response to a hypothetical provocation; and display more aggression during behavior
observations.
1.4.4. Shared goals, beliefs, and dyadic behavior
A further extension of the work on goals and beliefs is the examination of observed dyadic behavior.
Research by Coie et al. (1999) has demonstrated that dyad effects are as predictive of aggression as actor
or target effects. The current study allowed us to examine joint decision making and negotiation behavior
Table 1
Study variables, measurement context, and hypothesized intervention effects
Variables Measurement context Hypothesized effects
Classroom behavior
Social competence Teacher report More competent
Antisocial behavior Teacher report Less antisocial
Structured conflicts: beliefs
Goal preference Prisoner’s dilemma More prosocial
Expected satisfaction Prisoner’s dilemma Lower, self-high outcome
Higher, equal outcome
Post-game satisfaction Prisoner’s dilemma Higher satisfaction
Satisfaction reasoning Prisoner’s dilemma More egalitarian
Outcome satisfaction Prize division Higher satisfaction
Structured conflicts: behavior
Cooperative choices Prisoner’s dilemma More frequently
Cooperative negotiation Prize division More frequently
Coercive negotiation Prize division Less frequently
Adult intervention Prize division Less frequently
Hypothetical social cognitions
Attributions of intent Student survey Less hostile
Intended behavior Student survey Less aggressive, more competent
K.S. Frey et al. / Applied Developmental Psychology 26 (2005) 171–200176
when the participants shared goals or beliefs, or when they differed. To our knowledge, dyadic processes
have not been examined with respect to goals and attributions.
Prisoner’s dilemma games conducted between individuals show that shared cooperative goal
preferences predict cooperative decision making (McClintock & Liebrand, 1988). We predicted that
pairs who shared prosocial goals or less hostile attributions would exhibit more cooperative and less
aggressive behavior than pairs who shared self-interested goals or largely hostile attributions. It was not
clear, however, how divergent goals would affect links between goals and behavior—whether a person
with prosocial goals or benign beliefs would be able to appeal successfully to his or her partner’s bbetter
side,Q or whether the person with self-interested goals or hostile beliefs would heighten self-interested
concerns in his or her partner.
2. Methods
2.1. Schools
Fifteen elementary schools (seven K-5th grade and eight K-6th grade) from three cities in
western Washington were recruited to participate in the study. Conditions for inclusion were that
(a) school personnel agreed to random assignment to either the intervention or control group, and
(b) school personnel agreed to refrain from implementing any new social competence programs
for the duration of the study. The schools were located in urban districts of two moderately-sized
cities, two suburban districts contiguous to the urban districts, and a small city adjacent to a
naval base.
Schools in the intervention and control groups did not differ with respect to ethnic makeup or
percentage of students receiving free and reduced lunch (ts b 1). Three intervention and three
control schools had under 20% of their population receiving free or reduced lunch. Four
intervention and three control schools ranged from 20% to 50% receiving free or reduced lunch.
One intervention and one control school had more than 75% of their population receiving free or
reduced lunch. The school populations ranged from 52% to 89% European-American. Asian-
American and African-American comprised the next largest ethnic groups (18% and 12%,
respectively).
Schools in this area had ready access to research-based programs, and early school contacts revealed
that schools were reluctant to agree to a wait-list control design. Therefore, schools assigned to both
intervention and control groups received program materials, teacher training, and substitute teachers
during training, albeit for different grades. Control schools received these benefits for classrooms that
were not in the study. Fig. 1 presents a schematic of the implementation and data collection schedule.
2.2. Design and randomization
This study used data taken from a larger study that investigated the Second Step intervention over
time and with two different cohorts. The data in the present study come from children in the second
intervention cohort and in the control group, collected simultaneously.
Due to necessity rather than design, school recruitment extended over two years, the pre-study
year and the initial year of data collection for the first intervention cohort. Eleven of the schools
K.S. Frey et al. / Applied Developmental Psychology 26 (2005) 171–200 177
were recruited in the pre-study year and randomly assigned to the intervention (two-thirds of
sample) or control group (one-third of sample). The four other schools recruited in the first year of
cohort 1 data collection agreed to random assignment, although all were assigned to the control
group. Functional equivalence between the control schools was examined in preliminary analyses.
Post hoc analyses specific to the truly randomly assigned schools further explored the study’s
internal validity.
2.3. Students
Second- and fourth-grade teachers sent letters home with students requesting parental permission for
study participation. Active consent for two years of study participation was obtained for 63% of students
(n = 1,253). These served as participants in the study, although the program was administered to all
students in the intervention classrooms.
1
Participants were between the ages of seven and eleven and were roughly evenly divided by sex
(48.2% female) and grade level (54.6% in second grade) with proportions equivalent in the two groups.
There were 620 participants in the intervention group, and 615 in the control group. Attrition over the
two years, 25.5% in the intervention group and 28.8% in the control group, yielded samples of 462 and
436, respectively.
Participation
Year 1 Year 2
Group
Fall
data
a
Program Spring
data
a
Fall
data
b
Program Spring
data
c
Intervention
Study grades 2 2 2 3 3 3
4 4 4 5 5 5
Control
Study grades 2 2 3
3
4 4 5 5
Non-study grades K 1
1 2
Student
Fig. 1. Data collection and program implementation schedule by grade. Arrows denote chronological time.
a
Teacher ratings and
student surveys collected.
b
Teacher ratings collected.
c
Observations, interviews, teacher ratings, and student surveys collected.
1
One student did not participate due to parent request.
K.S. Frey et al. / Applied Developmental Psychology 26 (2005) 171–200178
2.4. Program implementation
Prior to program implementation, 47 second- and fourth-grade teachers participated in an initial two-
day training from Committee for Children trainers. They were introduced to program content, and they
practiced teaching lessons and exchanged ideas regarding modeling and support of skills throughout the
school day. At the beginning of the second year of implementation, 48 third- and fifth-grade teachers
received the same training (their first year with the program). Program consultants met with teachers
twice monthly to discuss and document program implementation.
Teachers typically taught one or two program lessons per week. The number of lessons varied from 18
to 27, depending on grade. The proportion of program lessons taught during the first year ranged from
42% to 100% (M = 79%, SD = .17). The mean was 83.3% in year 2, ranging from 0% to 100%
(SD = .20). A teacher who completed 50% of the lessons typically had completed the empathy unit and
the lessons that taught the basic problem-solving method. A teacher who completed 80% of the lessons
(46.3% of teachers in year 1, 67.3% in year 2) would have completed the empathy and impulse-control
units, and the basic techniques of emotion regulation. Although a minority of the students received the
complete program, we used an intent-to-treat model, including all experimental classrooms in the
intervention group.
2.5. Overview of study procedures
As shown in Fig. 1, students were followed through two school years. Each year, program
implementation started in late October, and was scheduled to finish in mid-March. Teachers rated student
social behavior each October over two years. Surveys of attributions and intended responses to
aggression were group-administered in classrooms in October of the first year. Teacher rating forms and
student surveys were re-administered in April of each year.
During May and June of year 2 (two to six weeks following survey administration), pairs of children
participated in tasks designed to create conflicts between self-interested and prosocial goals. A prisoner’s
dilemma game was used as a vehicle to assess children’s goals, expected outcome satisfaction, joint
cooperative choices, outcome satisfaction, and reasoning. A task requiring negotiation over the division
of four prizes was used to observe need for adult intervention, coercive behavior, and cooperative
behavior, and to assess outcome satisfaction. The dyadic decision-making processes in the prisoner’s
dilemma and responses in the prize division task were audio taped for later coding.
2.6. Measures and tasks
2.6.1. Teacher-reported social behavior
Merrell’s (1993) School Social Behavior Scale (SSBS) asks teachers to report how frequently students
engage in each of 32 antisocial and 33 socially competent behaviors. Both of the subscales have high
internal consistency (alphas ranged from .91 to .96 for both scales). Test–retest reliability is adequate,
with rs ranging from .76 to .82 for social competence and .60 to .73 for antisocial behavior. Stability of
scores across raters measured with Pearson product–moment correlations ranged from moderate for
social competence (rs from .72 to .83), to low for antisocial behavior (rs from .53 to .71). This suggests
that comparisons across school years (and raters) will be less comparable than comparisons within
school year.
K.S. Frey et al. / Applied Developmental Psychology 26 (2005) 171–200 179
The SSBS has been shown to discriminate between the behavior adjustment of children classified as
emotionally–behaviorally disordered, learning disabled, average, and gifted-talented. Scores on the
SBSS correlate highly with other rating scales of school-based social behavior.
Scales were used in analyses if no more than one item was missing per subscale. In such cases,
subscale means were substituted for the missing item.
2.6.2. Survey of beliefs and intentions
Surveys administered to students in their classrooms assessed attributions of hostile goals from
hypothetical vignettes of ambiguous provocations. Four vignettes were adapted from items developed by
Dodge (1980). After hearing themselves described as suffering harm from a peer, students indicated
whether the peer’s actions (which were ambiguous) were (a) due to unknown or benign intent; (b) bkind
of meanQ; or (c) breally mean on purpose.Q Ratings were summed across the four vignettes to create a
measure of hostile beliefs.
After rating the intentions of the vignette characters, students rated how likely they were to respond
with physical aggression, verbal aggression, and socially competent behavior on a five-point scale.
Students also evaluated responses to two additional vignettes in which the intent of the peer was
unambiguous (teasing or bullying behaviors). Two forms of the survey, each having four ambiguous and
two unambiguous vignettes, were counterbalanced across schools. At each school, forms were alternated
with administration (e.g., A, B, A).
2.6.3. Structured conflict: Prisoner’s dilemma game
Four same-sex, same-grade children from two different classes were randomly selected and
escorted to an unused room on the school grounds. Experimenters explained to each pair of children
that they would earn money for their class party while the other pair of children earned money for
their own class party (provided by experimenters at the end of the year). For each of eight trials,
each pair would choose either a cooperative or exclusively self-interested strategy. When both pairs
had made their decisions, they revealed their choices on the experimenter’s cue by simultaneously
holding up a color-coded card. As shown in Table 2, the earnings for each pair were determined
jointly by their selection and that of the other pair. Charts displayed overall and for each trial how
much children would earn given the four possible combinations (equal-high, self-high, self-low, and
equal-low). Any suggestion of initial confusion (13% of the sample) was followed by clarifying
instructions and examples. After each trial, an experimenter sitting with each of the pairs stated the
trial outcome and informed the children of the accumulated earnings of each pair.
2.6.3.1. Outcome preference, reasoning, and expected satisfaction. Prior to the eight decision trials,
children were audio taped as they indicated their first and second preferences among the four possible
outcomes and explained why they preferred their first choice to their second. With the first two outcomes
covered, children ranked the remaining two outcomes. These two measures were later combined to form
the prosocial goal variable.
To form the four expected satisfaction variables, children rated the satisfaction they expected with each
of the four outcomes on a five-point scale that ranged from 1 (DISAPPOINTED!) to 5 (PLEASED!).
2.6.3.2. Joint decision making and outcome satisfaction. The decision-making process of each pair
was audio recorded as they jointly selected a cooperative or self-interested strategy on each of eight
K.S. Frey et al. / Applied Developmental Psychology 26 (2005) 171–200180
trials. After completing the eight trials and hearing the amounts of money earned for their class and the
other pair’s class, children rated their overall satisfaction on a five-point scale (DISAPPOINTED! to
PLEASED!). They were then asked to explain why they were pleased or disappointed with the outcome.
2.6.4. Structured conflict: Negotiated prize division
In a second, more naturalistic task, each pair of children was offered four bthank-you giftsQ that varied
in attractiveness (a magnetic travel game, a stylish pop-a-point pencil, a bhappy face Q eraser, and an
undistinguished paper sticker). The children were told they could divide them any way they liked, as
long as both children in the pair agreed. Negotiations were audio taped. Experimenters did not comment
unless the negotiations stalled with the pair unable to resolve conflict over the division. In that case,
experimenters intervened in a set sequence, starting with a single open-ended question, bAny other
ideas?Q and becoming progressively more directive if negotiations did not resume.
After the prize negotiation, students rated their satisfaction with the outcome on a five-point scale.
They were then escorted back to their classrooms.
2.7. Coding and data summary
2.7.1. Responses to interviews during structured conflicts
2.7.1.1. Reasoning about goals and satisfaction. Four research assistants who were blind to condition
coded the reasons students gave for their preferred outcome prior to engaging in the prisoner’s dilemma
game, and their satisfaction (or dissatisfaction) with the outcome after engaging in the game. For each
measure, a coding system designed for this study used three ordered categories to represent degree of
prosocial reasoning: (a) desire to beat or dominate the other pair (e.g., bWe try to beat Ms. Chase’s
classQ) as a reason for preferred outcome and bAt least we got more than those guys,Q as a reason for
outcome satisfaction; (b) desire to earn absolute highest amount for one’s own class (bThat way we’ll get
a lot for our party;QbI wanted to get more than thatQ;) or (c) desire for equality or to earn the most
possible for both classes. (bThat way there’s more for the school;QbAt least we both got the sameQ).
Interrater agreement was 86%. Correcting for chance agreement, Cohen’s kappa was j = .80.
2.7.1.2. Prosocial goals. As shown in Table 2, the self-high outcome would be preferred by those who
wanted the highest possible absolute gain and those who wanted a competitive advantage over the other
pair. In order to help determine the motivating goal, children’s outcome rankings and reasoning were
Table 2
Possible trial outcomes for the joint contingency game
Pair B choice Pair A choice
Cooperative Self-interested
Cooperative $0.40 ($0.40) $0.50 ($0.20)
Equal-high Self-high
Self-interested $0.20 ($0.50) $0.10 ($0.10)
Self-low Equal-low
Note: Other pair’s choice is in parentheses.
K.S. Frey et al. / Applied Developmental Psychology 26 (2005) 171–200 181
used to form four ordinal categories reflecting prosocial goals in the prisoner’s dilemma task; the
categories were the following: (a) dominating—children selected the self-high outcome as most
preferred and explained their preference as the desire to dominate the other pair; (b) individualistic—
children selected the self-high outcome as most preferred and explained their preference as the desire to
earn the largest possible amount; (c) egalitarian—children selected the equal-high outcome as most
preferred and the self-high outcome as their second choice; and (d) altruistic—children selected the
equal-high outcome as most preferred and the self-low or equal-low outcome (rare) as their second
choice. All choices fell into these four categories.
2.7.2. Behavior coding
2.7.2.1. Joint decision making. The purpose of having children decide their choices jointly on the
prisoner’s dilemma game was to observe negotiation strategies and spontaneous reasoning (e.g., Frey &
Ruble, 1985, 1987). These plans failed as the children gestured and whispered to avoid having their
decisions overheard by the other pair. Our ability to transcribe the conversation accurately was severely
compromised and we abandoned coding efforts for the game.
Table 3
Prize negotiation codes, mean frequencies, examples, and intraclass correlation coefficients
Code Description
Cooperative
Check preference M = .48 Checks partner’s desires without constraining partner ’s choices (bDo you want the
backgammon game?QbWhat do you want?Q). r = .843
Prompt M = .74 Prompts partner to lead or act on wishes; may be submissive (bIf you want this one, it’s
okay.QbGo ahead.Q). r = .916
Check agreement M = .93 Checks partner’s desires after stating own desire or offering proposal (bI want the game.
What do you want?QbYou could have this, ok?Q). r = .776
Coercive
Aggression M = 1.23 Interrupts negotiation with aggression: insults, threats to cheat, refusal to talk (pointedly
ignores a question or says, bThat’s a stupid idea!Q). r = .823
Demand/whine M = 2.22 Tries to impose own preferences without considering the other: demands, bribes, rejection
of ideas, whining (bThat’s mine.QbNo way!Q). r = .898
Convincing
State wants M = 1.99 States desires (bI want the game.QbI don’t want that sticker!Q). r = .810
Process M = 1.12 Suggests a process for decision making (bI think the game and eraser should go together.Q
bLet’s put it away and decide on these three.Q). r = .900
Persuade M = 2.33 Uses persuasive comments that indicate perspective-taking: explanations, clarifications,
offering compensatory proposals (bYou could have all three;QbI could use the game
because I’m going on vacationQ). r = .934
Additional codes
Toy interest M = 1.37 Positive or interested comments about prizes (bHow does that work?QbCool!Q). r = .806
Miscellaneous M = 7.93 All other utterances, including expressive noises and comments to the experimenter
(bEarth to Bob;QbCould we get two pencils?Q). r = .930
K.S. Frey et al. / Applied Developmental Psychology 26 (2005) 171–200182
2.7.2.2. Need for adult intervention in the prize division. Four individuals, blind to group assignment,
transcribed and coded negotiation outcome and strategies. The outcome of the prize negotiation was
coded as (a) requiring adult intervention for resolution; (b) resolved independently by students (no
intervention was required beyond a single prompt, bAny other ideas?Q); or (c) not codable due to
premature adult intervention (e.g., omitting the non-directive prompt prior to issuing directive
statements). Generalizability was acceptable, ICC r = .73.
2.7.2.3. Observed negotiating strategies. All phrases uttered by the partners while negotiating the
prize division were coded using a simplified version of a coding system developed by DeVries and
colleagues (DeVries, Reese-Learned, & Morgan, 1991). Behaviors fell into one of the ten mutually
exclusive and exhaustive categories shown in Table 3. These behaviors included three categories of
cooperative strategies, three convincing strategies, and two coercive strategies. The aggressive behavior
category differed slightly from other codes in that it included non-responses in which one student
pointedly ignored or refused to respond to a statement made by the other student. Two additional
categories were toy interest and miscellaneous. Intraclass correlation coefficients for the ten codes
ranged from .78 to .93.
3. Results
In this section, we examine sample equivalence and present our analytic strategy for the outcome
evaluation. We then test the effects of intervention for the entire sample and the randomly assigned sub-
sample. Finally, we examine intra-individual relations between behavior and motivational constructs:
social goals and attributions of hostile intent.
3.1. Group differences
3.1.1. Sample equivalence
3.1.1.1. Attrition and pre-intervention differences. A 2 (group) by 2 (attrition) multivariate analysis of
variance (MANOVA) was performed on all time 1 (baseline) variables: hostile beliefs, behavior
intentions, and teacher ratings. Neither the multivariate, F(6, 830) = 1.47, ns, nor the univariate tests
showed a group by attrition interaction. Students who did not complete the study were more likely to give
socially competent responses to hypothetical problems at baseline than those who participated for two
years, F(1, 835) = 9.05, p b .01. A significant baseline group difference was found for teacher ratings of
antisocial behavior, higher for intervention group students, F(1, 835) = 5.14, p b .05. Means suggesting
lower teacher ratings of student social competence for the intervention relative to the control group did not
reach significance ( p b .10), but also indicated a need to use baseline covariates as appropriate.
3.1.1.2. Within control group equivalence. Preliminary univariate analyses (ANOVA) investigated
possible outcome differences between the two sets of control schools, those recruited simultaneously
with the first intervention cohort (who were randomly assigned and brought into the study the following
year) and those recruited a year later. One-way ANOVAs indicated that only 2 of the 46 analyses showed
significant differences. Those recruited first made more hostile attributions at Time 2 (spring 1) than
K.S. Frey et al. / Applied Developmental Psychology 26 (2005) 171–200 183
those in the second recruitment cohort. At Time 4 (spring 2) the pattern was reversed. Since these two
differences were not consistent, and the number of differences did not exceed that expected by chance,
control schools were treated as a single group.
3.1.2. Analytic strategy
3.1.2.1. Omnibus tests. Whenever possible, we used omnibus multivariate analyses of covariance
(MANCOVA), grouping conceptually related variables in order to decrease the number of analyses and
chance of spurious results. Univariate analyses were undertaken only if multivariate results were
significant at p b .05. Participant sex, grade, and baseline ratings of social competence or antisocial
behavior were entered as covariates, yielding a 2 (group) analysis of covariance. Divergences from this
design are noted, as when preliminary analyses indicated an interaction of a covariate with group.
3.1.2.2. Level of analysis. We adopted three different analytic strategies based on our conceptualiza-
tion of the variables. Children within classes were rated by the same individuals who observed children’s
behavior in the context of the classroom. Multivariate analyses were therefore followed by confirmatory
multilevel modeling with individuals nested within classroom.
Observed behavior reflected the dyadic relationship and joint decision making, requiring analyses on
the combined behavior of the pair. Preliminary partitioning of the variance components confirmed that
dyad accounted for 22.7% to 78.7% of the variance in negotiation behaviors, with classroom accounting
for virtually none.
Goals and satisfaction measured during the dyadic session reflected the individual’s expectations,
goals, and interpretation of the events as influenced by knowledge of the partner’s identity and that
partner’s subsequent behavior. Partitioning showed significant variance at the individual and pair levels.
Confirmatory multilevel modeling addressed subject nesting within pairs.
3.1.3. Teacher-reported social behavior
Because students changed teachers in the second year of program implementation, ratings were
confounded with reporters across years. Previous research indicates that teachers complete scales in
idiosyncratic ways (Coie et al., 1999), making it difficult to compare ratings across years/reporters. We
therefore examined change within each implementation year, entering baseline ratings as covariates. An
overall MANCOVA revealed significant group differences, F(4, 731) = 11.31, p b .001, in teacher-
reported social behavior. The discovery of a significant group by covariate interaction, however,
necessitated a separate analysis for antisocial behavior.
In order to look at meaningful differences, we divided baseline (Time 1) antisocial ratings at one
standard deviation above the mean. We analyzed changes in antisocial behavior during the 2 years by
means of a 2 (group) by 2 (antisocial baseline) MANCOVA, with sex and grade as covariates. As shown
in Table 4, the significant multivariate findings for group, F(2, 749) = 16.65, p b .001, and the group by
baseline interaction, F(2, 749) = 6.17, p b .01, were due to changes in the first year of program
implementation. Among those with high baseline ratings, the intervention group showed greater declines
in antisocial behavior than the control group ( p b .001, ES = .25). Significant but smaller differences also
favored the intervention group among students with low baseline scores ( p b .05, ES = .17). Intervention
students with low baseline scores showed no change in antisocial behavior, whereas control students
increased. There were no group differences in antisocial behavior change in Year 2.
K.S. Frey et al. / Applied Developmental Psychology 26 (2005) 171–200184
A two-group MANCOVA of social competence with sex, grade, and baseline social competence
as covariates indicated a significant group effect, F(2, 746) = 19.04, p b .001. Changes in the first
intervention year showed significant intervention group gains in social competence relative to the
control group. The intervention group also showed relative gains in the second year. As shown in
Table 4, these were of smaller magnitude ( p b .01, ES = .10), than the first-year gains ( p b .001,
ES = .20).
Using HLM to control for nesting in teacher reports, fall baseline ratings and sex were entered at the
individual level and grade and group at the classroom level. To investigate the differential intervention
effects by grade and baseline ratings, group was also entered on the baseline rating slope at the
classroom level.
Individual Level
Y
ij
¼ b
0j
þ b
1j
SEXðÞþb
2j
BASELINEðÞþr
0j
Classroom Level
b
0j
¼ c
00
þ c
01
GRADEðÞþc
02
GROUPðÞþu
0j
b
1j
SEXðÞ¼c
10
b
2j
BASELINEðÞ¼c
20
þ c
21
GROUPðÞ
HLM analyses confirmed the group difference and group by baseline ratings interaction for antisocial
behavior ( ps b .001) in Year 1. Separate HLM analyses showed group differences for both the high- and
low-baseline antisocial groups ( ps b .05). HLM also confirmed the group difference for social
competence in year 1 ( p b .01) but not for year 2.
Table 4
F-values and adjusted mean yearly changes in antisocial behavior and social competence ratings
Sub-groups by year Group F
Intervention Control
Group by baseline effects
Year 1 antisocial F(1, 750) = 10.72***
High baseline 14.57
a
2.26
a
Low baseline 0.73
b
4.27
b
Year 2 antisocial F b 1
High baseline 3.87 3.30
Low baseline 2.33 2.12
Cell sizes
High baseline 61 65
Low baseline 277 354
Group effects
Year 1 competence 7.92 0.29 F(1, 745) = 30.16***
Year 2 competence 2.72 0.26 F(1, 745) = 7.71**
Cell sizes 339 411
Means with the same subscripts differ at p b .001, and p b .01, respectively.
**p b .01. ***p b .001.
K.S. Frey et al. / Applied Developmental Psychology 26 (2005) 171–200 185
3.1.4. Prisoner’s dilemma game
3.1.4.1. Children’s understanding of contingencies. All children chose either the equal-high or self-
high outcome as their first choice. Further evidence that children understood the game contingencies
comes from a 4 (outcome) repeated measures ANOVA of the expected satisfaction ratings. The mean
ratings for equal-high and self-high outcomes were between bsatisfiedQ and bvery satisfiedQ (4.19 and
4.28, respectively), whereas those for equal-low and self-low outcomes fell between bdissatisfiedQ and
bin betweenQ (2.92 and 2.68, respectively). Since no hypotheses were formed for the self-low and equal-
low outcomes, these were not examined further.
3.1.4.2. Prosocial goals and expected satisfaction. There was a significant multivariate effect for
group, F(3, 741) = 5.00, p b .05, ES = .17. As predicted, the univariate analyses summarized in Table 5
indicate that goals chosen by the intervention group were more prosocial than those chosen by the
control group. As shown in Table 6, the mean differences appear due primarily to a smaller percentage of
the intervention group choosing a dominating goal and a larger percentage choosing the egalitarian goal.
Table 5 indicates that children in the control group expected greater satisfaction for the self-high
outcome than those in the intervention group. Post hoc comparisons showed that control group children
expected greater satisfaction from the self-high outcome than from the equal-high outcome ( p b .05).
The groups did not differ in the satisfaction expected following an equal-high outcome.
Table 5
F-values and adjusted group means for goals and satisfaction in the prisoner’s dilemma game and prize division
Group F
Intervention Control
Prosocial goals 2.52 2.30 F(1, 743) = 11.19**
Expected satisfaction
Self-interested 4.21 4.38 F(1, 743) = 6.96**
Cooperative 4.21 4.22 F b 1
Actual satisfaction
Game outcome 4.12 3.95 F(1, 784) = 3.92*
Prize division 4.49 4.25 F(1, 784) = 11.52***
Egalitarian reasons 1.07 0.95 F(1, 632) = 4.33*
*p b .05. **p b .01. *** p b .001.
Table 6
Frequencies of goal choices by group
Goals Intervention Control
nP(%) nP(%)
Dominating 55 14.0 96 23.4
Individualistic 127 32.4 130 31.6
Egalitarian 170 43.4 146 35.5
Altruistic 40 10.2 39 9.5
N 392 411
K.S. Frey et al. / Applied Developmental Psychology 26 (2005) 171–200186
Hierarchical linear modeling (HLM) of prosocial goals and expected satisfaction included baseline
social competence scores (grand mean centered) and student sex at the individual level. In order to verify
the multivariate analyses, we entered group at the pair level intercept. Analyses with this design
confirmed all group differences obtained with MANCOVA ( ps b .05).
Individual Level
Y
ij
¼ b
0j
þ b
1j
SEXðÞþb
2j
BASELINEðÞþr
0j
Pair Level
b
0j
¼ c
00
þ c
01
GRADEðÞþc
02
GROUPðÞþu
0j
b
1j
SEXðÞ¼c
10
b
2j
BASELINEðÞ¼c
20
3.1.4.3. Joint decision making. Because this was a measure of the dyad’s cooperative choices on the
prisoner’s dilemma game, the sum of the pair’s baseline ratings for social competence was entered as a
covariate along with sex and grade. The proportion of cooperative choices made by each pair on the
eight trials was subjected to univariate analysis of covariance. Contrary to predictions, the two groups
did not vary in the proportion of cooperative choices, F b 1, M = 37.1%.
3.1.5. Observed negotiation strategies during prize division
3.1.5.1. Negotiation context. The average number of utterances made during the negotiation task was
20.34. Judging from the high level of student interest in the prizes, the negotiation task was well-suited
for testing children’s ability to resolve conflicts amicably. Positive or interested comments about the
prizes were common among pairs of students and did not vary by group (M = 1.38).
2
3.1.5.2. Need for adult intervention. Coders identified 6 out of 427 sessions in which the experimenter
failed to follow protocol and intervened prematurely to assist in resolving the conflict (1.4%). The
remaining 421 sessions were coded as requiring adult intervention or resolving negotiation successfully.
A 2 (group) by 2 (adult intervention) Chi-square analysis showed the predicted group difference in need
for adult intervention, v
2
(1) = 6.66, p b .05, with pairs of students in the intervention group less likely to
require adult intervention (n = 16, 7.6% of 210) than those in the control group (n = 28, 12.9% of 217).
3.1.5.3. Observed negotiating strategies. The behavior observed during the prize negotiation provides
insight into the differential rates of conflict resolution found between the intervention and control group
dyads. As predicted, significant group effects were found among the most and least competent strategies
(cooperative and coercive). Pairs of children in the intervention and control groups were equally likely to
try to convince their partners to change position.
Coercive strategies were analyzed with the sum of the pairs’ baseline antisocial ratings, sex, and grade
entered as covariates. Multivariate analyses showed that group assignment predicted coercive behavior,
F(2, 397) = 4.14, p b .05, ES = .14, during negotiations. As shown in Table 7, the intervention group was
less likely than the control group to behave aggressively while negotiating for prizes. The groups did not
differ in demanding, whiney behavior.
2
Negative comments about the toys occurred so infrequently that coders were not able to code them reliably.
K.S. Frey et al. / Applied Developmental Psychology 26 (2005) 171–200 187
Preliminary analyses of cooperative strategies indicated the need to include sex as a factor with
group, yielding a 2 (group) by 2 (sex) MANOVA with grade and baseline social competence as
covariates. Multivariate results showed a significant group by sex interaction, F(3, 396) = 2.62, p b .05.
Inspection of the significant univariate findings showed that pairs of girls in the intervention group
were more likely to check preferences of their partners than girls in the control group ( p b .05,
ES = .17). A marginally significant group by sex interaction ( p b .10, ES = .14) suggested that boys in
the intervention group were more likely than their cohorts in the control group to ask for their
partner’s agreement.
3.1.6. Outcome satisfaction
There were no group differences in the outcome of the prisoner’s dilemma game. However, univariate
tests following the finding of a significant multivariate group effect, F(2, 783) = 7.17, p b .001, ES = .14,
showed, as predicted, that intervention students were more satisfied with the outcomes of the prisoner’s
dilemma game and prize division than control students (see Table 5).
The reasons children gave for their level of satisfaction had a high proportion of missing data. In order
to preserve sample numbers for evaluating outcome satisfaction, the reasons were analyzed separately
with 2 (group) ANCOVAs with sex, grade, and baseline as covariates. Students in the intervention group
gave more egalitarian reasons for their satisfaction or dissatisfaction than students in the control group,
F(1, 632) = 4.33, p b .05, ES = .10.
Table 7
F-values and means for coercive and cooperative negotiating strategies
Strategy Group F
Intervention Control
Group effects
Aggression 0.88
a
1.52
a
F(1, 398) = 6.44*
Demand, whine 2.36 2.07 F b 1
Group by gender effects
Check preference F(1, 398) = 4.57*
Girl pairs 0.78
b
0.49
b
Boy pairs 0.34 0.39
Check agreement F(1, 398) = 3.40
y
Girl pairs 0.89 1.25
Boy pairs 0.92
c
0.69
c
Prompt partner F b 1
Girl pairs 0.86 0.89
Boy pairs 0.60 0.66
Cell sizes
Girl pairs 87 110
Boy pairs 98 108
Means with the same subscripts differ at p b .05.
y
p b .10. * p b .05.
K.S. Frey et al. / Applied Developmental Psychology 26 (2005) 171–200188
Using the previous two-level (individual and pair) design, individual HLM analyses confirmed the
significant findings for satisfaction with the prize division and egalitarian reasoning for satisfaction or
dissatisfaction with the outcome of the prisoner’s dilemma game ( ps b .05), but not satisfaction with the
game outcome.
3.1.7. Survey of hostile attributions and intentions
In preliminary analyses, behavioral intentions showed nearly identical patterns of responses to the
ambiguous and non-ambiguous provocations. Aggressive intentions were uniformly higher for non-
ambiguous provocations, but there were no interactions of provocation-type with group, student sex, or
grade. Responses for each type of behavior were therefore combined across the six vignettes.
Cronbach’s alphas were .84, .82, and .63, respectively, for physically aggressive intentions, verbally
aggressive intentions, and socially competent intentions. The alpha value for the measure of hostile
attributions was .61.
A 2 (group) by 3 (assessment time) by 3 (response type) multivariate repeated measures test with sex and
grade as covariates was used to analyze hostile attributions and behavioral intentions. Contrary to
predictions, there were no significant differences between- or within-groups. The data suggest a possible
floor effect. The percentages of subjects at baseline who scored higher than the midpoints for hostile
attributions, verbal aggression, and physical aggression were only 19.0%, 10.0%, and 11.8%, respectively.
3.1.8. Random subset analyses for program effects
In order to further explore internal validity, we performed post hoc analyses of our significant findings
with just the randomly assigned sample. Post hoc 2-group ANCOVAs were computed with the students
assigned to the intervention (n = 440) and control (n = 165) groups, using sex, grade, and baseline
teacher ratings as covariates. Year 1 teacher ratings again showed positive effects for social competence,
F(1, 600) = 30.18, p b .0001, and antisocial behavior, F(1, 599) = 16.78, p b .001. A significant group by
baseline rating interaction, F(1, 599) = 8.14, p b .01, indicated the most potent effects occurred among
children with initially high antisocial behavior. Intervention effects on Year 2 ratings of social
competence did not reach significance, F(1, 600) = 3.48, p b .07.
The observed dyadic behavior showed positive intervention effects on aggressive negotiating
strategies, F(1, 263) = 4.19, p b .05, and joint effects of group and sex on checking partner preferences,
F(1, 263) = 3.96, p b .05.
Social-cognitive measures showed the predicted effects for prosocial goals, F(1, 504) = 6.36, p b .05,
and expected satisfaction with a self-interested outcome F(1, 504) = 4.01, p b .05. Unlike the results from
the larger sample, satisfaction ratings did not reach significance in the random sample [game outcome
F(1, 530) = 2.02, p b .16], showing only a marginal effect for the prize division, F(1, 530) = 2.99, p b .09.
Children in the intervention group were again found to offer egalitarian reasons for their level of
satisfaction, F(1, 436) = 3.90, p b .05, more frequently than those in the control group.
3.2. Cognition–behavior-satisfaction linkages
In order to analyze cognitive–behavioral relationships within individuals and within dyads, we grouped
children based on their goals and attributions. Children who identified the equal-high outcome as most
preferred (altruistic or egalitarian goal) constituted the cooperative goal group. Children who preferred the
self-high outcome (self-interested or dominating goal) constituted the self-interested goal group.
K.S. Frey et al. / Applied Developmental Psychology 26 (2005) 171–200 189
There were a limited number of possible scores for the mean of the attribution responses. Unable to
split scores at the median, we reasoned that behaviorally significant differences were more likely to
occur between more extreme high scores and lower scores. We therefore split the scores between the
lower 65% (up to and including .5) and the upper 35% (.75 and above). This yielded 2 (goals)
2
(attribution) ANOVAs for analyzing the contribution of social cognitions to individual satisfaction,
teacher-reported behavior, and behavior intentions.
3.2.1. Satisfaction
As predicted, analyses of the satisfaction variables showed a significant effect for goal preferences,
F(3, 581) = 4.65, p b .05. Satisfaction with the prize division was higher if students preferred cooperative
goals F(1, 583) = 5.23, p b .05, than if they preferred self-interested goals (M = 4.51 and M = 4.34,
respectively). Cooperative goals were also associated with a tendency to provide prosocial reasons for
one’s satisfaction or dissatisfaction with the results of the prisoner’s dilemma game F(1, 583) = 9.08,
p b .01 (M = 2.10 v. M = 1.93 for those with self-interested goals). Contrary to predictions, goal
preferences were not related to actual satisfaction with the game outcome. Attributions were not
predictive of satisfaction.
3.2.2. Teacher-reported social behavior
Analyses of teacher-reported social competence and antisocial behavior showed significant effects for
goals, F(2, 745) = 8.07, p b .001, and attributions, F(2, 745) = 7.00, p b .001. The goal by attribution
interaction approached significance, F(2, 745) = 2.90, p b .06, necessitating an evaluation of the main
effects (shown in Table 8). Children with cooperative goal preferences were rated more socially
competent, F(1, 746) = 16.93, p b .001, and less antisocial than those with self-interested goal
preferences, F(1, 746) = 12.40, p b .001. Non-hostile attributions were also associated with greater
social competence, F(1, 746) = 13.87, p b .001 and less antisocial behavior, F(1, 746) = 9.05, p b .01.
Examination of the goal by attribution interactions for both social competence F(1, 746) = 3.95, p b .05,
and antisocial behavior F(1, 746) = 5.69, p b .05, showed that hostile attributions predicted the behavior
ratings of children with self-interested, but not cooperative goal preferences, ps b .001 (See Table 8).
Table 8
Teacher- and self-reported behavior as a function of goal preferences and attributions
Behavior Non-hostile attributions Hostile attributions
Cooperative goals Self-interested goals Cooperative goals Self-interested goals
Teacher ratings
Social competence 130.70
a
127.80
b
127.15
c
117.55
abc
Antisocial behavior 51.49
d
53.30
e
52.48
f
61.87
def
N 220 260 129 142
Intended behavior
Verbal aggression 1.57
g
1.60
h
2.10
gh
2.47
gh
Physical aggression 1.39
i
1.44
j
1.99
ij
2.38
ij
Socially competent 3.89 3.84 3.88 3.71
N 230 270 130 139
Means with the same subscripts differ at p b .001.
K.S. Frey et al. / Applied Developmental Psychology 26 (2005) 171–200190
Those children with self-interested goals and hostile attributions were significantly less socially
competent and more antisocial than other children.
3.2.3. Intended behavior
Analysis of intended behavioral responses to the hypothetical conflict scenarios showed the expected
close relationship to attributions, F(3, 763) = 40.64, p b .0001. They were also related to the goals
children preferred in the prisoner’s dilemma game, F(3, 763) = 3.82, p b .05, and to the goal by
attribution interaction, F(3, 763) = 2.79, p b .05. Both verbally aggressive, F(1, 765) = 7.39, p b .01, and
physically aggressive intentions, F(1, 765) = 9.64, p b .001, were lower among children who espoused
cooperative goals than those who espoused self-interested goals. Hostile attributions were linked to
higher levels of verbally aggressive, F(1, 765) = 95.46, p b .0001, and physically aggressive intentions,
F(1, 765) = 113.04, p b .0001. As shown in Table 8, there were significant goal by attribution
interactions for both verbal, F(1, 765) = 5.82, p b .05, and physical aggression, F(1, 763) = 8.37, p b .01.
These reflected larger attribution group differences found in the self-interested goal group, p b .0001,
than in the cooperative goal group, p b .001.
3.2.4. Cognitions and dyadic interaction
Within dyads, there were three patterns of goal concordance: (a) both children preferred the
cooperative goal (altruistic or egalitarian pattern); (b) both children preferred the exclusively self-
interested goal (self-interested or dominating pattern); or (c) children preferred different goals. There
were also three groups corresponding to pair attributions: (a) both children espoused non-hostile
attributions; (b) both espoused hostile attributions; and (c) one espoused non-hostile and the other
espoused hostile attributions. These yielded 3 (pair goals) by 3 (pair attributions) MANOVAs for
analyzing the contribution of social cognitions to dyadic behavior.
3.2.4.1. Joint decision making. Joint decision making on the prisoner’s dilemma task showed that
cooperative choices varied as a function of pair goals, F(2, 298) = 5.71, p b .001. As shown in Table 9,
pairs in which both children shared cooperative goals made significantly more cooperative choices than
Table 9
Dyadic behavior as a function of within-pair goal concordance
Pair behaviors Pair goals
Shared Unshared
Cooperate Self-interest
Game choices
Cooperative 0.42
a
0.37
a
0.33
a
Prize negotiation
Aggression 1.00 1.56 1.03
Demand, whine 1.14
bc
2.28
b
2.24
c
Check preference 0.62
d
0.55
e
0.30
de
Prompt 0.85 0.80 0.56
Ask agreement 0.79 1.02 0.87
N of pairs 91 104 112
Means with the same subscripts differ at p b .05 (subscripts b, c, and e) or p b .01 (subscripts a and d).
K.S. Frey et al. / Applied Developmental Psychology 26 (2005) 171–200 191
pairs in which children shared self-interested goals, p b .01, or disagreed in their preference, p b .001.
Pairs of children who disagreed made the fewest cooperative choices, significantly lower than children
who shared a preference for the self-interested goal, p b .01.
3.2.4.2. Negotiating strategies. Coercive negotiating strategies were also related to pair goals,
F(4, 668) = 2.58, p b .05. Examination of the significant results for demanding, whining behavior,
F(2, 333) = 3.48, p b .05, showed that pairs who shared cooperative goals were less demanding and
whiney than pairs who shared self-interested goals or pairs who did not agree (both ps b .05).
Cooperative negotiating strategies were related to pair goals, F(6, 666) = 2.33, p b .05, and pair
attributions, F(6, 666) = 2.28, p b .05. Goals predicted the frequency of checking preferences with one’s
partner, F(2, 333) = 5.22, p b .01. Pairs of children who shared cooperative goals checked partner
preferences more than pairs who disagreed on goals, p b .01. Pairs who shared self-interested goals also
checked partner preferences more than pairs who disagreed on goals, p b .05. (See Table 9). Pair
attributions predicted the frequency of asking for partner’s agreement, F(2, 333) = 4.88, p b .01. Pairs in
which both children made non-hostile attributions asked for agreement more frequently than pairs who
made hostile attributions, p b .05, or disagreed, p b .01 (means = 1.25, .86, and .67, respectively).
4. Discussion
This study extends prevention research and research on social development in three ways: first by
demonstrating that a universal intervention can influence goals and observed behavior in conflict
situations; second, by linking goals and attributions to observed behavior; and third, by examining the
links between dyadic behavior and shared or unshared social cognitions. The discussion first addresses
program-related changes and then the linkages between social cognitions and behavior that provide a
conceptual basis for the Second Step program.
4.1. Group differences
Participation in the Second Step program was associated with significant benefits in student behavior,
goals, and social reasoning for the sample as whole, and for the smaller, randomly-assigned sub-sample.
The specific effects of the intervention varied according to the type of measurement and analysis.
Consistent differences across analyses and sample were found in aggression and the need for adult
intervention during prize division, and in motivational constructs such as goals, expected satisfaction,
and reasoning about satisfaction. Group differences in teacher-reported behavior were robust for the first,
but not second, year of intervention. Program effects on higher-level negotiation strategies were limited
to intervention girls, and group differences in post-task satisfaction varied by analysis. No group
differences were found for the survey measures, attributions and behavioral intentions, or for joint
cooperative choices in the prisoner’s dilemma game.
4.1.1. Developmental and contextual factors in conflict behavior
Consistent with naturalistic observations (Grossman et al., 1997), children in the intervention group
displayed less aggression than those in the control group. Children in the two groups, however, made an
equal number of demands when negotiating the division of their bthank-youQ prizes. Meta-analyses of
K.S. Frey et al. / Applied Developmental Psychology 26 (2005) 171–200192
other observational studies indicate that typically developing children use predominately coercive
negotiating strategies (Laursen, Finkelstein, & Townsend Betts, 2001). The Second Step program
appears to have reduced coercive strategies that fall into the aggressive, antisocial end of the continuum,
but not the more developmentally typical bpower plays.Q
Children who use higher-level negotiation strategies have better school adjustment than their peers
(Yeates, Schultz, & Selman, 1991). In this study, girls in the intervention group were most likely to use
such a strategy. Asking for a partner’s preferences is particularly useful, as well as prosocial, because
elementary students typically fail to state what they want in conflict situations (Johnson, Johnson,
Dudley, & Acikgoz, 1994). The intervention did not result in the same findings for boys, although
intervention boys were marginally more likely to check for agreement than their control group
counterparts. Girls, who generally display higher levels of prosocial behavior than boys (Eisenberg &
Fabes, 1998) may have found it relatively easy to adopt a collaborative strategy, whereas boys responded
with marginal increases in cooperative, but not explicitly prosocial, behavior.
Previous research using a toy division paradigm shows that failure to reach agreement is associated with
destructive negotiating strategies and negative relationships between participants (Ram & Ross, 2001).
The present study extends this line of research to classmates, who often have emotionally-charged disputes
over access to resources. (Johnson et al., 1994). These can last several days, interfering with opportunities
to learn and teach. During toy negotiations, program participants required adult intervention 41% fewer
times than those in the control group. Although the overall level of intervention was low, even small
differences in student self-sufficiency can be critical to teachers. Reduced aggression and need for adult
intervention potentially can free up teacher time, as well as reduce the potential for learning disruption.
In contrast to effects on negotiating strategies and success, the intervention had no impact on joint
cooperative choices in the prisoner’s dilemma game. Other research using both a game format and
resource negotiation task found only the negotiation to be predictive of peer relationships (Fonzi,
Schneider, Tani, & Tomada, 1997). It is possible that the game-like conditions of the current procedure
elicited too much competitive spirit to be sensitive to group differences. The greater variability in goals
and behavior that are elicited by negotiations may also transfer better to open-field situations, such as the
playground, than to a highly structured game context, where, typically, the goal is to bwin.Q
4.1.2. Teachers’ perceptions of changes in student social behavior
The findings showed some convergence between teacher-reported and directly observed behavior.
Teachers in the first year of student program participation reported clear increases in social competence
and decreases in antisocial behavior relative to the control group. Decreases in antisocial behavior were
largest among intervention children initially rated as highly antisocial, but also significant for those rated
low in antisocial behavior. Reports of intervention group improvement in the second year were, at best,
equivocal for social competence and absent for antisocial behavior. This pattern is similar to results of a
two-year study of the bI Can Problem SolveQ program (Shure & Spivak, 1982). The current study does
not allow us to discern whether acceleration of social development occurred only in the first year, or
whether teachers failed to notice continued improvement.
4.1.3. Motivational factors in conflict situations
Program participants were explicitly taught norms related to fairness, cooperation, and compassion for
others. Perhaps these norms helped direct children’s attention (Metcalfe & Mischel, 1999) to the
prosocial aspects of the conflict situations. In support of this interpretation, intervention students were
K.S. Frey et al. / Applied Developmental Psychology 26 (2005) 171–200 193
more likely to espouse prosocial goals and state that mutual, rather than unilateral, good fortune in the
prisoner’s dilemma game contributed to their satisfaction with the outcome. This perspective may allow
more opportunities for enjoyment than an exclusively self-interested one, and elicit prosocial behavior
from others. Adolescents with prosocial goals experience more satisfaction with social relationships than
those with dominance goals (Jarvinen & Nicholls, 1996). Consistent with those findings, cooperative
goals, but not attributions, were associated with greater satisfaction with the prize division and more
references to others’ well-being when explaining satisfaction. Further research may indicate that social
satisfaction is an important aspect of intervention success. Reinforcement provided by mutual goodwill
may help maintain positive behavior change and commitment to prosocial goals (Erdley & Asher, 1999).
4.1.4. Attributions and intended responses to hypothetical situations
In contrast to positive effects of the intervention on motivation variables during the structured conflict
situations, we found no program benefits with regard to attributions and intentions. Our method of
assessment may have suffered from a floor effect. Unlike goals and reasoning, which were measured in
the context of a real interaction, hostile beliefs and intentions relied on multiple-choice answers to
hypothetical situations, a format less likely to uncover individual differences (Orobio de Castro et al.,
2002). Alternatively, this may indicate a need to strengthen the part of the program that teaches children
to question hostile beliefs.
4.2. Linking goals and beliefs with aggressive and prosocial behavior
Previous research indicates that social cognitions predict behavior in competitive situations (Lochman
& Dodge, 1998). In the current study, converging evidence from observer-, teacher-, and self-reports
show linkages between social cognitions and behavior. Cooperative negotiating strategies and social
competence were positively related to cooperative goals; antisocial behavior and aggressive intentions
were negatively related. Attributing hostile goals to hypothetical others predicted more aggressive
intentions, lower social competence, and greater antisocial behavior, as rated by teachers.
Examination of the goal by attribution interactions for teacher ratings and aggressive intentions
suggests that having cooperative goals may partially buffer children from the influence of hostile biases
in attribution. Having self-interested goals may bprimeQ children to adopt a bwin–loseQ perspective on
social interaction and increase their reliance on subjective and indirect means of understanding others’
behavior. Conversely, children who did not evince hostile biases were less likely to display negative
behavior, even if their goals were exclusively self-interested.
Previous research indicates that aggressive beliefs formed in the primary grades predict later
aggressive behavior and account for considerable stability in aggression (Burks, Dodge, Price, & Laird,
1999; Huesmann & Guerra, 1997). The current findings suggest that a combination of self-interested
goals and hostile biases may erect significant barriers to positive social development as well. The links
found between goals and a wide spectrum of maladaptive and competent behaviors support arguments
for increased focus on promoting positive social goals (Erdley & Asher, 1999).
4.3. Predicting interactive behavior from dyadic goals and beliefs
As predicted, children who shared cooperative goals and non-hostile attributions were more
cooperative and less coercive during the prisoner’s dilemma and prize negotiation tasks. Thus, the
K.S. Frey et al. / Applied Developmental Psychology 26 (2005) 171–200194
cooperative negotiating strategy of asking for partner agreement occurred at low levels if even one of the
partners had attributed hostile goals to others. Similarly, demanding, whiney strategies increased if one
or both partners had self-interested goals.
The behaviors associated with less prosocial goals and beliefs cannot be entirely explained as due to
one person’s bbad influence,Q however. While shared cooperative goals were associated with cooperative
choices in the prisoner’s dilemma game and inquiring as to partner’s toy preference during prize
division, unshared goals were associated with the least cooperative behavior. Even pairs of children who
shared self-interested goals behaved with greater cooperation than those with unshared goals, suggesting
that dyadic conflict may have heightened the salience of self-interested behavior. These results await
replication. Future research that compares the individual and dyadic behaviors of children differing in
social cognitions could begin to identify when dyadic or group mechanisms contradict or extend those
identified in individual analyses.
4.4. Implications for practice
For a school of 600 students, class materials, professional staff training, and training materials are
estimated to be less than $9.00 per student, with no additional costs in subsequent years. This is the
second field trial of the Second Step program to show beneficial effects on unbiased observations of
aggressive and positive social behavior (Grossman et al., 1997). The effects on observed and teacher-
reported behavior were modest, particularly for children that did not exhibit high levels of antisocial
behavior. We cannot determine whether program contamination in control schools or the lack of the
recommended whole-school implementation may have reduced group differences. As Rosenberg,
Powell, and Hammond (1997) have nonetheless demonstrated, small changes in the behavior of
individual students have an important impact on a classroom environment when those changes are
reflected in the behavior of 25 individuals.
The current results also show group differences in positive goals, offering more hope for program-
related change than Lochman and colleagues (1993) derived from their study of aggressive adolescent
boys. A key difference in the two studies is age of intervention. Huesmann and Guerra (1997) found that
social cognitions are unstable and particularly open to influence in elementary school. Their
developmental learning model of how beliefs are formed corresponds in several ways to practices
found in the Second Step program. First there is direct tuition in values and goals. A second source of
learning is positive modeling by teachers, a more potent influence than exhortation alone (Grusec, Saas-
Kortsaak, & Simutis, 1978). A third strategy is reinforcement and instruction bin the momentQ when real
world conflicts occur. Such events help shape later social beliefs and goals (Huesmann & Guerra, 1997).
Students’ behavior, and the reinforcement and attributions they receive for that behavior, appear to be
highly influential (Grusec, Kuczynski, Rushton, & Simutis, 1978) as children construct beliefs about
themselves and others.
Another influence on positive goal change may stem from improvements in cognitive, affective, and
behavioral skills fostered by social–emotional learning programs. Tesser’s (1986) model of self-
enhancement maintenance and Eccles’ expectancy-value theory (Eccles et al., 1983) suggest that social
skill deficits will lead to a devaluing of prosocial behavior and goals while perceived expertise in
aggression will increase the perceived value of hostile goals. In this model, self-perceived competence in
prosocial domains would mediate the link between skills and goals (Ruble & Frey, 1991; Wigfield &
Eccles, 1992), with both dimensions necessary to support socially responsible behavior.
K.S. Frey et al. / Applied Developmental Psychology 26 (2005) 171–200 195
4.5. Caveats
We obtained the agreement of all fifteen schools to assign randomly, but were only able to do so with
the eleven schools recruited in the first year. The random and non-randomly assigned control schools
appear functionally equivalent in comparisons. More importantly, the duplication of results in the
random-only subset for observed negotiation behavior, teacher-reported behavior, goals, reasoning, and
expected satisfaction strongly support a causal interpretation of findings in the larger sample.
The longitudinal design of this study allowed us to look for effects after two years of participation in a
universal program. The period covered did not allow us to assess long-term transfer of training, nor to
investigate fully the possible effects of development on the likelihood that students would retain or
expand their prosocial goals and strategies. Further research with follow-up assessments that reach into
early adolescence is needed to explore these possibilities. Having both pre- and posttest measures of
students’ goals, beliefs, and strategies in actual conflict situations would also facilitate a transactional
analysis of behavior and attitudinal change.
5. Conclusions
The current study used multi-informant outcome measures to examine observable and phenomeno-
logical aspects of prosocial and antisocial behavior. Positive goals and beliefs, in combination with skills
(e.g., perspective-taking, emotion regulation) may be viewed as protective characteristics. In order to act
in socially responsible ways, children must possess both the relevant skills and the motivation to use
those skills. The concordance found between behavior, satisfaction, and cognitions is consistent with
social-cognitive models of aggression and social competence (e.g., Crick & Dodge, 1994; Huesman,
1988) that provide much of the conceptual foundation of the Second Step program. Goals and hostile
attributions were independently predictive of a wide range of outcome variables, including the use of
prosocial behaviors during conflict. Interventions that foster positive motivation and multiple social-
cognitive, emotional, and behavior skills are likely to be more successful than single component
approaches.
For a relatively low cost, universal programs such as the Second Step program may strengthen the
resilience of b at-riskQ children without stigmatizing individuals (Greenberg et al., 2001). Although a
universal program that attempts change only in school is unlikely to change strongly maladaptive
behavior patterns, it can provide a stable social base for more intensive interventions by providing
consistent school expectations, a shared vocabulary, and complementary behavioral repertoires.
Equally important, universal programs can provide benefits to children not identified as bat riskQ
(Durlak, 1995) by reducing classroom disruption and building skills that help children meet life’s
challenges.
Acknowledgement
Versions of this paper were presented at the meeting of the Society for Research in Child
Development, Minneapolis, April, 2001, and at the meeting of the Society for Prevention Research,
Washington, D.C., June, 2001. Three of the authors are employed by Committee for Children. The
K.S. Frey et al. / Applied Developmental Psychology 26 (2005) 171–200196
opinions expressed herein are the authors’ and do not necessarily reflect those of Committee for
Children. The authors would like to thank the staff, students, and parents of the participating schools.
They are grateful to A. Alexander, C. Blais, J. Gerber, A. Mackey, C. Parnell, C. Santos, A. Siqveland,
and P. Yerkovitch for assistance with coding and program implementation, and to C. Keelan and D.
Cooper for help with manuscript preparation. Thanks are also due to M. Greenberg for reviewing an
earlier manuscript and to K. Beland for a program that marries empirically-based content with teacher
needs.
References
Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory. Englewood, NJ 7 Prentice Hall.
Boldizar, J. P., Perry, D. G., & Perry, L. C. (1989). Outcome values and aggression. Child Development, 60, 571–579.
Burks, V. S., Dodge, K. A., Price, J. M., & Laird, R. D. (1999). Internal representational models of peers: Implications for the
development of problematic behavior. Developmental Psychology, 35, 802–810.
Chung, T. -Y., & Asher, S. R. (1996). Children’s goals and strategies in peer conflict situations [Special issue]. Merrill-Palmer
Quarterly, 42, 125 – 147.
Coie, J. D., Cillessen, A. H. N., Dodge, K. A., Hubbard, J. A., Schwartz, D., Lemerise, E. A., et al. (1999). It takes two to fight:
A test of relational factors and a method for assessing aggressive dyads. Developmental Psychology, 35, 1179– 1188.
Coie, J. D., & Dodge, K. A. (1998). The development of aggression and antisocial behavior. In W. Damon (Series Ed.) & N.
Eisenberg (Vol. Ed.), Handbook of child psychology: Vol. 3. Social, emotional, and personality development (pp. 779–862).
New York: Wiley.
Coie, J. D., Dodge, K. A., & Kupersmidt, J. B. (1990). Peer group behavior and social status. In S. R. Asher, & J. D. Coie
(Eds.), Peer rejection in childhood (pp. 17–59). New York7 Cambridge University Press.
Coie, J. D., Lochman, J. E., Terry, R., & Hyman, C. (1992). Predicting early adolescent disorder from childhood aggression and
peer rejection. Journal of Consulting and Clinical Psychology, 60, 783 – 792.
Committee for Children. (1997a). Second Step: A violence prevention curriculum, grades 1–3 (2nd ed.). Seattle, WA7 Author.
Committee for Children. (1997b). Second Step: A violence prevention curriculum, grades 4–5 (2nd ed.). Seattle, WA7 Author.
Crick, N. R., & Dodge, K. A. (1994). A review and reformulation of social information-processing mechanisms in children’s
social adjustment. Psychological Bulletin, 115, 74–101.
Crick, N. R., & Dodge, K. A. (1996). Social information-processing mechanisms on reactive and proactive aggression. Child
Development, 67, 993–1002.
Delveaux, K. D., & Daniels, T. (2000). Children’s social cognitions: Physically and relationally aggressive strategies and
children’s goals in peer conflict situations. Merrill-Palmer Quarterly, 46, 672 – 692.
Derryberry, D., & Rothbart, M. K. (1988). Affect, arousal, and attention as components of temperament. Journal of Personality
and Social Psychology, 55, 958 – 966.
DeVries, R., Reese-Learned, H., & Morgan, P. (1991). Sociomoral development in direct-instruction, eclectic, and constructivist
kindergartens: A study of children’s enacted interpersonal understanding. Early Childhood Research Quarterly, 6, 473 – 517.
Dodge, K. A. (1980). Social cognition and children’s aggressive behavior. Child Development, 51, 162 – 170.
Dodge, K. A., & Newman, J. P. (1981). Biased decision-making processes in aggressive boys. Journal of Abnormal Psychology,
90, 375–379.
Dodge, K. A., Pettit, G. S., McClaskey, C. L., & Brown, M. M. (1986). Social competence in children. Monographs of the
Society for Research in Child Development, 51, 1–85.
Durlak, J. (1995). School-based prevention programs for children and adolescents. Thousand Oaks, CA7 Sage.
Eccles, J. S., Adler, T. F., Futterman, R., Goff, S. B., Kaczala, C. M., Meece, J., et al. (1983). Expectancies, values and academic
behaviors. In J. T. Spence (Eds.), Achievement and achievement motives (pp. 75 – 146). San Francisco7 Freeman.
Eisenberg, N., & Fabes, R. A. (1998). Prosocial development. In W. Damon (Series Ed.) & N. Eisenberg (Vol. Ed.), Handbook
of child psychology: Vol. 3. Social, emotional, and personality development (pp. 701 –778). New York7 Wiley and Sons.
Eisenberg, N., Guthrie, I. K., Murphy, B., Shepard, S., Cumberland, A., & Gustavo, C. (1999). Consistency and development of
prosocial dispositions: A longitudinal study. Child Development, 70, 1360–1372.
K.S. Frey et al. / Applied Developmental Psychology 26 (2005) 171–200 197
Erdley, C. A., & Asher, S. R. (1996). Children’s social goals and self-efficacy perceptions as influences on their responses to
ambiguous provocation. Child Development, 67, 1329–1344.
Erdley, C. A., & Asher, S. R. (1999). A social goals perspective on children’s social competence. Journal of Emotional and
Behavioral Disorders, 7, 156 – 167.
Fonzi, A., Schneider, B. H., Tani, F., & Tomada, G. (1997). Predicting children’s friendship status from their dyadic interaction
in structured situations of potential conflict. Child Development, 68, 496 – 506.
Frey, K. S., Hirschstein, M. K., & Guzzo, B. A. (2000). Preventing aggression by promoting social competence. Journal of
Emotional and Behavioral Disorders, 8, 102–112.
Frey, K. S., & Ruble, D. N. (1985). What children say when the teacher is not around: Conflicting goals in social comparison
and performance assessment in the classroom. Journal of Personality and Social Psychology, 48, 550–562.
Frey, K. S., & Ruble, D. N. (1987). What children say about classroom performance: Sex and grade differences in perceived
competence. Child Development, 58, 1066–1078.
Greenberg, M. T., Domitrovich, C., & Bumbarger, B. (2001). The prevention of mental disorders in school-aged children:
Current state of the field. Prevention & Treatment, 4 (Article 1. Retrieved December 16, 2002, from http://journals.apa.org/
prevention/volume4/pre0040001a.html).
Grossman, D. C., Neckerman, H. J., Koepsell, T. D., Liu, P. Y., Asher, K. N., Beland, K., et al. (1997). Effectiveness of a
violence prevention curriculum among children in elementary school: A randomized controlled trial. Journal of the
American Medical Association, 277, 1605–1611.
Grusec, J. E., Kuczynski, L., Rushton, J. P., & Simutis, Z. M. (1978). Modeling, direct instruction, and attributions: Effects on
altruism. Developmental Psychology, 14, 51 – 57.
Grusec, J. E., Saas-Kortsaak, P., & Simutis, Z. M. (1978). The role of example and moral exhortation in the training of altruism.
Child Development, 49, 920–923.
Halberstadt, A. G., Denham, S. A., & Dunsmore, J. C. (2001). Affective social competence. Social Development, 10, 79 – 119.
Hartup, W. W., French, D. C., Laursen, B., Johnston, K. T., & Ogawa, J. R. (1993). Conflict and friendship relations in middle
childhood: Behavior in a closed-field situation. Child Development, 64, 445 – 454.
Hudley, C., Britsch, B., Wakefield, W. D., Smith, T., Demorat, M., & Cho, S. -J. (1998). An attribution retraining program to
reduce aggression in elementary school students. Psychology in the Schools Special Issue: Addressing Youth Anger and
Aggression in School Settings, 35, 271–282.
Hudley, C., & Graham, S. (1993). An attributional intervention to reduce peer-directed aggression among African-American
boys. Child Development, 64, 124–138.
Huesmann, L. R. (1988). An information processing model for the development of aggression. Aggressive Behavior, 14,
13–24.
Huesmann, L. R., & Guerra, N. G. (1997). Children’s normative beliefs about aggression and aggressive behavior. Journal of
Personality and Social Psychology, 72, 408 – 419.
Jarvinen, D. W., & Nicholls, J. G. (1996). Adolescents’ social goals, beliefs about the causes of social success, and satisfaction
in peer relations. Developmental Psychology, 32, 435 – 441.
Johnson, D., & Johnson, R. (1996). Conflict resolution and peer mediation programs in elementary and secondary schools: A
review of the research. Review of Educational Research, 66, 459– 506.
Johnson, D. W., Johnson, R., Dudley, B., & Acikgoz, K. (1994). Effects of conflict resolution training on elementary school
students. Journal of Social Psychology, 134, 803 – 817.
Kendall, P. C. (1993). Cognitive-behavioral therapies with youth: Guiding theory, current status, and emerging developments.
Journal of Counseling and Clinical Psychology, 61, 235 – 247.
Laursen, B., Finkelstein, B. D., & Townsend Betts, N. (2001). A developmental meta-analysis of peer conflict resolution.
Developmental Review, 21, 423 – 449.
Lemerise, E. A., & Arsenio, W. F. (2000). An integrated model of emotion processes and cognition in social information
processing. Child Development, 71, 107–118.
Lochman, J., & Dodge, K. (1998). Distorted perceptions in dyadic interactions of aggressive and nonaggressive boys: Effects of
prior expectations, context, and boys’ age. Development and Psychopathology, 10, 495 – 512.
Lochman, J. E., Wayland, K. K., & White, K. J. (1993). Social goals: Relationship to adolescent adjustment and to social
problem solving. Journal of Abnormal Child Psychology, 21, 135 – 151.
Matsumoto, D., Haan, N., Yabrove, G., Theodorou, P., & Carney, C. C. (1986). Preschoolers’ moral actions and emotions in
Prisoner’s Dilemma. Developmental Psychology, 22, 663–670.
K.S. Frey et al. / Applied Developmental Psychology 26 (2005) 171–200198
McClintock, C. G., & Liebrand, W. B. G. (1988). Role of interdependence structure, individual value orientation, and
another’s strategy in social decision making: A transformational analysis. Journal of Personality and Social Psychology,
55, 396 – 409.
McMahon, S. D., Washburn, J., Felix, E. D., Yakin, J., & Childrey, G. (2000). Violence prevention: Program effects on urban
preschool and kindergarten children. Applied and Preventive Psychology, 9, 271 – 281.
Meichenbaum, D. (1977). Cognitive-behaviour modification: An integrative approach. New York7 Plenum Press.
Merrell, K. (1993). School social behavior scales. Brandon, VT7 Clinical Psychology Publishing Company.
Metcalfe, J., & Mischel, W. (1999). A hot/cool-system analysis of delay of gratification dynamics of willpower. Psychological
Review, 106, 3 – 19.
Miller-Johnson, S., Lochman, J. E., Coie, J. D., Terry, R., & Hyman, C. (1998). Comorbidity of conduct and depressive
problems at sixth grade: Substance use outcomes across adolescence. Journal of Abnormal Child Psychology, 26,
221–232.
Murphy, B. C., & Eisenberg, N. (1996). Provoked by a peer: Children’s anger-related responses and their relations to social
functioning [Special issue]. Merrill-Palmer Quarterly, 42, 103– 124.
Orobio de Castro, B., Veerman, J. W., Koops, W., Bosch, J. D., & Monshouwer, H. J. (2002). Hostile attribution of intent and
aggressive behavior: A meta-analysis. Child Development, 73, 916 – 934.
Pepler, D. J., & Rubin, K. H. (Eds.). (1991). The development and treatment of childhood aggression. Hillsdale, NJ7 L. Erlbaum
Associates.
Ram, A., & Ross, H. S. (2001). Problem-solving, contention, and struggle: How siblings resolve a conflict of interests. Child
Development, 72, 1710–1722.
Reid, J. B., Eddy, J. M., Fetrow, R. A., & Stoolmiller, M. (1999). Description and immediate impacts of a preventive
intervention for conduct problems. American Journal of Community Psychology, 27, 483–517.
Rose, A. J., & Asher, S. R. (1999). Children’s goals and strategies in response to conflicts within a friendship. Developmental
Psychology, 35, 69 – 79.
Rosenberg, M. L., Powell, K. E., & Hammond, R. (1997). Applying science to violence prevention. Journal of the American
Medical Association, 277, 1641–1642.
Rubin, K. H., & Krasnor, L. R. (1986). Social-cognitive and social behavioral perspectives on problem solving. In M.
Perlmutter (Ed.), Cognitive perspectives on children’s social and behavioral development (Vol. 18, pp. 1–68). Hillsdale, NJ7
Lawrence Erlbaum Associates.
Ruble, D. N., & Frey, K. S. (1991). Changing patterns of comparative behavior as skills are acquired: A functional model of
self-evaluation. In J. Suls, & T. A. Wills (Eds.), Social comparison: Contemporary theory and research (pp. 79 – 113).
Hillsdale, NJ7 Lawrence Erlbaum Associates.
Shantz, C., & Hobart, C. (1989). Social conflict and development: Peers and siblings. In T. J. Berndt, & G. Ladd (Eds.), Peer
relationships in child development (pp. 71–94). New York, NY7 John Wiley and Sons.
Shure, M. B., & Spivak, G. (1982). Interpersonal problem-solving in young children: A cognitive approach to prevention.
American Journal of Community Psychology, 10, 341 – 356.
Slaby, R. G., & Guerra, N. G. (1988). Cognitive mediators of aggression in adolescent offenders: I. Assessment. Developmental
Psychology, 24, 580 – 588.
Spivack, G., & Shure, M. B. (1974). Social adjustment of young children: A cognitive approach to solving real-life problems.
Oxford, England7 Jossey-Bass.
Stein, N. L., & Albro, E. R. (2001). The origins and nature of arguments: Studies in conflict understanding, emotion, and
negotiation. Discourse Processes Special Issue: Argumentation in Psychology, 32, 113–133.
Stoolmiller, M., Eddy, J. M., & Reid, J. B. (2000). Detecting and describing preventive intervention effects in a universal
school-based randomized trial targeting delinquent and violent behavior. Journal of Consulting and Clinical Psychology, 68,
296–306.
Tesser, A. (1986). Some effects of self-evaluation maintenance on cognition and affect. In R. M. Sorrentino, & E. T. Higgins
(Eds.), Handbook of motivation and cognition (pp. 435–464). New York7 Guilford.
Van Schoiack-Edstrom, L., Frey, K. S., & Beland, K. (2002). Changing adolescents’ attitudes about relational and physical
aggression: An early evaluation of a school-based intervention. School Psychology Review, 31, 201– 216.
Weiner, B. (1985). Attributional theory of achievement motivation and emotion. Psychological Review, 92, 548 – 573.
Wentzel, K., & Wigfield, A. (1998). Academic and social motivational influences on students’ academic performance.
Educational Psychology Review, 10, 155 – 175.
K.S. Frey et al. / Applied Developmental Psychology 26 (2005) 171–200 199
Wentzel, K. R. (1996). Social goals and social relationships as motivators of school adjustment. In J. Juvenon, & K. R. Wentzel
(Eds.), Social motivation: Understanding children’s school adjustment (pp. 226 – 247). New York7 Cambridge University
Press.
Wentzel, K. R., & Erdley, C. A. (1993). Strategies for making friends: Relations to social behavior and peer acceptance in early
adolescence. Developmental Psychology, 29, 819 – 826.
Wigfield, A., & Eccles, J. S. (1992). The development of achievement task values: A theoretical analysis. Developmental
Review, 12, 265 – 310.
Yeates, K. O., Schultz, L. H., & Selman, R. L. (1991). The development of interpersonal negotiation strategies in thought and
action: A social-cognitive link to behavioral adjustment and social status. Merrill-Palmer Quarterly, 37, 369–405.
K.S. Frey et al. / Applied Developmental Psychology 26 (2005) 171–200200
