The Relationship between Children’s Self-perceptions and Behavior in Context: Implications for the Study of Behavior Change in At-Risk Youth

By Hallie Caroline Spierings van der Wolk

INDEPENDENT RESEARCH THESIS April, 2009

Submitted in partial fulfillment of the requirements for the degree of Bachelor of Science in the Department of Psychology at Brown University

Children’s Self Perceptions

Page 2 of 75

Acknowledgments

I do not know where to start in thanking Jack Wright for his support and guidance throughout this project. Jack, your enthusiasm for creative thought, probing questions, and analytic thinking has inspired me throughout this process to ask questions and explore ideas, even when no clear answer is in sight. Your passion for what you do is evident always and I am a better student, investigator, and thinker as a result of the excitement that you bring to the table. Thank you for letting me think and explore, while never failing to guide me when I don’t have the tools to get where I want to go. I would like to thank Audrey Zakriski for her support during summer data collection. Your presence at Saturday Night Research Parties is always a breath of fresh air. To the research team and the dedicated staff at Wediko, thank you for all of the time and effort spent collecting data and interviewing children. You work tirelessly to effect change, both immediate and long term, and for this I and the children whose lives you have touched thank you. Thank you to my family for all of the ways in which you support me. Finally, a huge thank you is in order to all of my friends who now know more about Wediko than they may ever have wanted to, yet never fail to share in my enthusiasm for it. I am completely humbled by your support for me during this project, I cannot thank you enough.

Children’s Self Perceptions

Page 3 of 75

Table of Contents

Acknowledgements

2

Table of Contents

3

Abstract

4

Introduction

5

Method

15

Results

24

Discussion

33

References

42

Tables

48

Figure Captions

51

Figures

52

Appendices

55

Children’s Self Perceptions

Page 4 of 75

Abstract My project explores the ways in which children’s self-perceptions of behavior relate to behavioral observations in a clinical treatment setting for at-risk youth. This study examined the relationship between children’s self-perceptions of their behavior in “aversive” conflict situations with peers and adults, and extensive field observations of their behavior. Both context-specific behavioral observations of children in conflict situations and uncontextualized observations of children’s overall behavioral frequencies were collected. Analysis of aggressive, withdrawn, and prosocial behavior demonstrated clear disagreement between perceptions and observations of behavior frequencies in response to aversive events. Children’s self-perceptions showed strong agreement with global behavior rates, highlighting the possibility that children view their behavior in overall trends rather than in response to specific situations. Analysis of behavior change over time demonstrated a significant increase in aggression over all three measurement types. Prosocial behavior, on the other hand, showed distinct patterns of change for each measurement type. Several interpretations of this finding are offered, highlighting the possibility that some behaviors (aggression) may be easier to perceive than others (prosociality). Further examination of behavior change over time suggests that when grouped according to low, medium, and high levels of aggression, patterns of change differ by aggressive child type. Not only do these findings lend support for iatrogenic effects of treatment, they suggest that children are well aware of these effects. In sum, this study highlights the importance of understanding children’s self-perceptions of behavior as a possible factor in the outcome of treatment.

Children’s Self Perceptions

Page 5 of 75

Introduction Emotion regulation is a topic of enduring interest in the study of normative child development (Hubbard & Coie, 1994) and in the study of children with emotional and behavioral disorders (EBD) (Zeman, Cassano, Perry-Parrish, & Stegall, 2006). EBD children struggle to receive and interpret emotional cues from others, pay less attention to both situational detail and emotional stimuli, and have more difficulty recognizing and interpreting their own affective experience (Halberstadt, Denham, & Dunsmore, 2001). Improving our understanding of the maturation of skills necessary for emotion regulation has undeniable developmental and clinical significance (Cole, Martin, & Dennis, 2004; Halberstadt et al., 2001). Social Competence Attempts to conceptualize emotion regulation have clarified connections between emotion regulation, cognitive development (Cook, Greenberg, & Kusche, 1994; Lewis & Todd, 2007), social competence, and psychological functioning (Zeman et al., 2006). Halberstadt et al. (2001) defined the construct of emotion regulation as affective social competence, or “the efficacious communication of one’s own affect, one’s successful interpretation and response to others’ affective communications, and the awareness, acceptance, and management of one’s own affect” (p. 80). The breadth of this definition points to the difficulty in quantifying its many components and how they contribute to a child’s social competence. The multifaceted nature of the construct of social competence has challenged researchers in this field for some time. During the past 30 years, a common goal in many social development studies has been to compare the experience of an emotion with the individual’s self-reported awareness of it (Halberstadt et al., 2001; Hubbard et al., 1994). In doing so, investigators sought to clarify the relationship between the two defining aspects of emotion regulation: emotion

Children’s Self Perceptions

Page 6 of 75

understanding and emotion expression (Southam-Gerow & Kendall, 2002). Yet the difficulty in studying emotion regulation continues to be apparent (Zeman et al., 2006). Difficulties in Studying Emotion Hubbard and Coie (1994) define social competence as the ability to reach social goals, specifically, being liked by peers. Developmentally, this skill is acquired by the child’s increasingly acute and context-specific ability to recognize, predict, and respond to emotion in others. Initially, children are only able to understand the emotions that others are feeling, by distinguishing basic facial expressions (Odom & Lemond, 1972). With maturity comes the ability to connect situations with emotions that are often experienced in them. In this way, Hubbard and Coie (1994) equate the proficiency with which the child can gauge complex emotions in peers with social competence, ostensibly because the ability to gauge emotions is a key requirement for being liked by peers. For example, whether a child can successfully enter an unfamiliar peer group is in part a reflection of that child’s ability to accurately gauge the social intentions and emotions of their peers. Studies of sociometric status lend support to Hubbard’s hypothesis regarding the relationship between social competence and acceptance by peers. Cassidy, Parke, Butkovsky, and Braungart (1992) and Halberstadt et al. (2001) suggest that children who are able to understand their own emotions are better able to maintain positive peer relationships. Emotional expressiveness within families (thought to contribute to children’s understanding of emotions) has been shown to relate to children’s behavioral and emotional functioning (Cassidy et al., 1992). These researchers suggest several factors that would cause parental expressiveness to positively affect children’s emotional awareness, including modeling and the opportunity to understand the effects of one’s own behaviors. Though it is reasonable to conclude that exposure to social interactions and emotional expressiveness might contribute to a child’s socio-emotional

Children’s Self Perceptions

Page 7 of 75

skill set, without extensive controls for children’s environments, other social interactions, and personality characteristics, the presumed correlation between parent expressiveness and child’s functioning must be seen as a tentative one. Others investigators (e.g., Pettit, Dodge, & Brown, 1987; Wright, Giammarino, & Parad, 1986) question the view that social competence should be defined as the ability to be well liked, citing similarity models of interpersonal attraction and the context-dependence of peer-group composition. For example, children with a given personality trait (e.g., aggressiveness) may be accepted by peers who share that trait, yet rejected by peers who do not. Thus, using sociometric status as a marker of maturation potentially oversimplifies both the ‘social competence’ construct and the meaning of social status. Because emotion regulation has clear implications for an individual’s overall development and especially for EBD children (Southam-Gerow et al., 2002; Zeman et al., 2006), other methods and measures are needed in order to better evaluate its development, without depending on sociometric outcomes. Self-Perception: An Alternative Method for Studying Emotion Regulation Quantifying emotion regulation in normative and EBD populations has been pursued by Casey (1993, 1996) by comparing children’s self-perceptions and video recorded observations of emotional expressiveness in response to emotionally provoking situations. Using this approach, Casey (1993) examined the extent to which children’s self reports of emotion corresponded to their immediate facial expressions within carefully controlled laboratory paradigms. In this study, predominantly White, middle class children without serious emotional problems were exposed either to positive or negative feedback from a peer within a lab setting. In the positive scenario, the peer stated that he or she would like to play with the subject child and complimented both the subject’s appearance and the way the child had played the game. In the

Children’s Self Perceptions

Page 8 of 75

negative scenario, the peer stated that he did not want to play with the subject child and criticized both the subject’s appearance and the way he or she had played the game (Casey, 1993). Subjects were then videotaped for about 95 seconds, beginning at the start of the feedback session. Following this, a structured interview asked subjects to describe their experience and assess their own reactions. The recall portion of the interview had 4 parts, each of which was scored as expressing an emotion that was either consistent or inconsistent with observer’s impressions of their initial facial expressions. In the self awareness portion of the interview, children were asked three questions regarding how they knew that they felt their reported feeling(s). The sophistication of the response was rated (e.g., ability to explain or express own feelings or make general statements about the type of situation that elicits an emotion). Results of this study, compared with results from a similar study for EBD populations (Casey, 1996), suggest that both child (normative versus EBD) as well as situation (positive versus negative) affect the relationship between children’s self-perceptions and adult coders’ observations of their emotions. Casey (1993) reports that when presented with negative feedback from a peer, older children’s coded expressions were negatively correlated with their self-report of fear and anger. When presented with positive feedback, on the other hand, children’s report of emotion was positively correlated with facial expression (Casey, 1993). In other words, children’s report of emotional experience was not consistent with observed emotion in negative feedback situations, but the two were consistent in positive feedback situations. A related study of EBD children (Casey, 1996) found that during positive feedback situations, children with behavioral disorders were less (rather than more) able to accurately reflect on their emotional expression. Negative situations evoked more negative responses for highly externalizing children but more positive emotions for internalizing children. These studies

Children’s Self Perceptions

Page 9 of 75

highlight interesting patterns of self-perception in positive and negative situations for both normative and EBD populations. Confounding Factors in Self-Perception Studies of Emotion: Display Rules Display rules are believed to affect the way in which emotions are expressed (Saarni, 1979). As a result, the extent to which the individual is conscious of the effect that display rules have on their personal expression of emotion, in any given situation, will influence the accuracy of their self-reports of emotion. Accuracy of self-report, therefore, is not a measure of emotional awareness, but rather an indication of one’s understanding of the ways that social rules affect the expression of emotion. Further, there are conflicting reports on whether display rules diminish the expression of emotion or whether they enhance the expression of emotion (Casey, 1993; Parker, Hubbard, Ramsden, Relyea, Dearing, Smithmyer, & Schimmel, 2001). Describing one possible effect of display rules, Casey (1993) explains that “it is possible that hearing the negative stimulus may have increased children’s concern about controlling their facial display, and there may have been more concern in reporting negative expressive reactions in the negative condition” (p. 127). In this example, the relationship between observed and self-reported emotion would be inconsistent because even though the individual might report feeling sad, his expression might have been altered in order to reflect social rules concerning sadness. Parker et al. (2001) point out that while knowledge of display rules may indicate socially competent functioning, it is also probable that in situations where negative emotions are considered to be acceptable, display rules might not diminish expression. In this scenario, socially competent functioning would be marked by a consistent relationship between selfreported and observed emotions because an understanding of display rules would enhance the expression of angry feelings. Parker et al. (2001) state that if “the expression of felt emotion was

Children’s Self Perceptions

Page 10 of 75

considered an appropriate response, […] correspondence between self-reports of emotional experience and observations of emotion expression may be indicative of socially competent functioning” (p. 530). Thus, based on the individual’s understanding of the appropriateness of an emotional response, display rules may cause enhanced or diminished expressions of emotion. As a result, both a correspondence as well as a divergence between self-reported and observed emotional reactions could indicate socially competent functioning. Summary. Available evidence suggests that the expression of emotion is affected by social norms, referred to as display rules (Saarni, 1979). Studies comparing emotional expression and self-report are subject to various methodological and interpretive problems. This suggests that additional methods and measures may be needed to clarify the relationship between children’s self reports and others’ observations of their behaviors. Behavioral Awareness Research on children’s perceptions often deals with their perceptions of other’s behavior, rather than perceptions of their own (e.g., Wright & Mischel, 1988). Observer’s perceptions of behavior are often considered to be more accurate than children’s perceptions of their own behavior (DiBartolo & Grills, 2006). However, in a study of cross-informant agreement, DiBartolo et al. (2006) found that in some respects, children were better able to reflect on their experience than were observer’s, saying “children’s reports predicted their own anxious feelings and, to some degree, their behaviors, […] whereas other informants’ reports did not” (p. 630). This suggests that further examination of children’s self-perceptions of behavior may reveal information regarding their behavior. Comparing children’s self-perceptions of behavior with adult’s observations may reveal important characteristics of cross-informant agreement and children’s self-knowledge.

Children’s Self Perceptions

Page 11 of 75

Self-perceptions of Context-Specific and Global Behavior The study of behavior emphasizes important differences in the measurement of contextspecific versus global behaviors (Zakriski, Wright, & Parad, 2007). Specifically, these two measures can reveal different patterns of behavior for a given target child. For example, children who improved least during short-term residential treatment showed increased aggression according to a global behavior frequency measure, indicating an adverse or “iatrogenic” effect of treatment. Such iatrogenic effects are thought to stem from aggregating problem children, and have been documented in other clinical settings, particularly for children who are “marginally deviant” upon entry into treatment (Dodge, Dishon, & Lansford, 2006). Silver and Eddy (2006) report that “association with deviant peers is a significant predictor of a variety of interrelated youth problem behaviors, including student misconduct, academic failure, delinquency, substance abuse, and early sexual behavior” (p. 254). Although Zakriski et al. (2007) found clear adverse changes when overall behavior measures were used, context-specific observations revealed a more complex picture: these children had become more aggressive towards peer provocation but simultaneously becoming less aggressive towards non-aversive social stimuli (e.g. being talked to by adults) (Choukas-Bradely et al., 2008). Thus, changes in context-specific reactions can better qualify iatrogenic effects of treatment. In light of past studies examining children’s ability to reflect on their own affective experiences in emotionally arousing situations (e.g., Casey, 1993, 1996), it may be that examining children’s self-perceptions of their behavior will further our understanding of their behavior. Though not the focus of this study, minimization of the iatrogenic effects of treatment for EBD children could hinge on a better understanding of both behavior in context and the accuracy with which children are able to reflect on their own behavior in context. As in studies of children’s emotion regulation, the factors that mediate a treatment outcome are thought to be

Children’s Self Perceptions

Page 12 of 75

“social-cognitive in nature and depend heavily on information processing and individual interpretations of an intervention and subsequent reactions” (Dishion et al., 2006). In the present study, children’s aggressive, withdrawn, and prosocial behaviors were assessed using three measurement methods: children’s self-perceptions of their reactions to aversive events, context-specific observations of their reactions to aversive events, and global observations of behavior. There were four main features of analysis. First, correlational analyses examined the temporal stability of measurement types as well as the concurrent validity of perceptions and observations of behavior. Second, the main effects of behavior were assessed for each measurement type and results were compared over perception and observation data. Third, changes in behavior over time were assessed for each measurement method and similarly compared across perception and observation data. Finally, trends in children’s self-perceptions and observations of behavior were examined for children demonstrating high, medium, and low levels of aggression. Hypotheses. My hypotheses were based on past research showing that emotional awareness was not correlated with observed emotional experience (Parker et al., 2001) and describing the difficulty experienced by children with serious emotional and behavior disorders in identifying their affective experience (Halberstadt et al., 2001). Guided by this work, I hypothesized that children’s self-perceptions of behavior, even when reliably assessed, would not be consistent with either context-specific behavioral observations that matched the contexts in which children reported, or global measures that assessed their behavior more broadly. Secondly, as my method focused on children’s self-perceptions in response to aversive social situations, I hypothesized that aggressive behavior would most frequently be reported while prosocial behavior would be least likely to be reported. Third, in light of work by Dodge et al. (2006) describing observed iatrogenic effects of the aggregation of problematic children in

Children’s Self Perceptions

Page 13 of 75

intervention programs, I hypothesized that children characterized as being least aggressive would display the most profound iatrogenic effects. This would include the greatest increases in aggression and greatest decreases in withdrawal and prosocial behavior, measured both by context specific and global observations, though not by children’s own self-perceptions.

Children’s Self Perceptions

Page 14 of 75

Methods Data for this project were collected at Wediko, a short-term (45-day) residential treatment center for children with serious emotional and behavioral disorders. Children are referred primarily through the Boston Public School System and by social workers, therapists, and parents. At Wediko, children are assigned to cabins of about 10 same-sex, developmentally comparable peers. On a daily basis, children take part in activities such as fishing, archery, or art, two hours of level appropriate classroom work, and group therapy with their cabin group. Child participants. A final sample of 124 children was used. Children were placed into age groups that were consistent with those determined by Achenbach and Rescorla (2001) (described below). Age and gender groups were as follows: young girls (N = 15, Myears = 9.8, SD = 1.27), young boys (N = 35, Myears = 9.63, SD = 1.35), older girls (N = 22, Myears = 14.23, SD = 1.5), and older boys (N = 52, Myears = 13.94, SD = 1.68). An overall aggression score, based the aggregation of three administrations of the Teacher Report Forms (TRF) (described below), was calculated (M = 69.85, SD = 8.32). Children were grouped in three groups based on overall aggression (described below) and are as follows: low aggression group (N = 44, MT-score = 62.10, SD = 3.4), moderate aggression group (N = 39, MT-score = 68.56, SD = 1.7), and high aggression group (N = 41, MT-score = 79.39, SD = 5.99). Both children and their parents signed consent forms prior to the start of the summer (see Appendix A). Exclusions. Children who missed either the first or the second administration of the “Reactions to Situations Interview” (see below) were excluded from analysis. Children who were coded using the PALM system less than 85 times over the course of the summer were also

Children’s Self Perceptions

Page 15 of 75

excluded. Children who switched living groups did so within the first week and as a result their data were retained. Staff participants. During the summer of 2008, 158 counselors were employed at Wediko. Most were college and graduate students, who primarily held residential counselor and activity counseling positions. Social workers, psychologists, and special education teachers held lead staff and supervisory positions. Counselors who arrived late were excluded from TRF administrations for two weeks as each TRF asked the counselor to respond based only on the previous two week period. Behavioral Assessments Reactions to Situations Interview (RTS). This measure evaluated children’s perceptions of their own behavior. To do this, children were read descriptions of two situations, which were selected in order to sample “aversive” social interactions involving conflict with peers and adults. The first was a scenario in which a peer teased him/her, and the second was one in which an adult gave a warning. In the peer-tease situation, the scenario read “Imagine that you are with your group at an activity that you enjoy. You are working really hard and feeling proud of what you are doing. You show your project to another kid in your group and s/he starts making fun of it and teasing you about it.” In the adult-warn situation, the scenario read “Imagine you’re with your group at an activity you like and you are excited to get started. You say something to another kid while an adult is explaining the rules. The adult comes over to you and warns you that if you don’t stop talking you will lose your turn.” The wording of these scenarios was crafted specifically to avoid individual interpretation of the situation. In other words, both situations include terms (“making fun of and teasing” and “warns”) defining the target interaction. A full copy of the Reactions to Situations protocol can be found in Appendix B.

Children’s Self Perceptions

Page 16 of 75

In response to each these situations, children were asked to judge how often they would respond in a variety of ways including “feel sad,” “argue or talk back,” and “ask someone else for help” (Appendix B). A response scale of “never, “sometimes,” “often,” or “very often” was provided. In addition, two final items assessed how hard it would be and how long it would take to recover from each situation. These items were not used in the present study as it will primarily focus on behavioral awareness rather than recovery from aversive situations. These items can be found in Appendix B. Procedure. During the second week of the program, counselors were trained to administer this measure. They were asked to interview children privately, away from the group and in a quiet place. A sample question was read to the child, and possible “never,” “sometimes,” “often,” “very often” responses were given. Children were then asked if they had any questions. The assessment was administered at two points doing the summer, at the end of week 3 and again at the end of week 5. For consistency across setting, each child completed the questionnaire with same 1 to 4 day period, at both time points. Teacher Report Form (TRF). The Teacher Report Form (TRF) (Achenbach & Rescorla, 2001) is one of the most widely used in school systems as a measure of problem behavior. It uses 118 items, grouped into 8 narrow band scales, to quantify behavior. These narrow-band scales include aggression, delinquency, withdrawal, somatic complaints, anxiety, thought disorder, attention problems, rule breaking, and social problems. A broad-band measure of internalization aggregates withdrawal, somatic complaints, and anxiety narrow-band scales. The broad-band measure of externalization aggregates aggression and rule-breaking. Raters are asked to respond to specific prompts such as “argues a lot” and “complains of loneliness” with 3 possible responses: “not true, “somewhat or sometimes true,” or “very true or often true.” Children’s scores are reported as T-scores, which are normed against the national sample for age and gender

Children’s Self Perceptions

Page 17 of 75

groups. Children between 6 and 11 years of age are categorized as “young,” while children between the ages of 12 and 18 years of age are categorized as “old.” Once normed, the TRF reports a mean score of 50 and a standard deviation of 10 for all children. A T-score of 67 is often considered to be “borderline clinical,” while a score of 70 or higher is considered to be clinical. Reliability information can be found in Achenbach and Rescorla (2001). The TRF was administered for every child at 3 points during the summer, at two-week intervals. Raters were asked to fill out the TRF based only on the preceding 2 weeks (since the children arrived or since the last administration). Rater-child pairs were determined based on the extent of the rater’s exposure to the target child. Because all staff filled out two TRF forms at each administration, some, but not all, children were rated twice. As a result, when more than one TRF was available for any child per administration, the rater who would have spent more time with the child, as determined by their residential versus activity staff status, was selected and the other TRF was not used. If both raters spent approximately the same about of time with the child, one was chosen at random. Field observations of behavior. Context-specific and global field observations were collected using hand held Palm Pilots (“Palm”© Z-22s). A full copy of the Palm coding manual can be found in Appendix C. Six days a week, 83 Palms were distributed to staff throughout the setting. All Palms were distributed in the mornings and collected at dinner. Overnight they were docked such that data were removed, the next day’s schedule was uploaded, and palms were charged for the next day. In the morning classrooms were given one palm, every activity (e.g., archery, fishing) was given 1 or 2, and every group carried 3 throughout the day. Using SatForm© software created for this project, coding sessions for every child were specific to the child’s movements throughout the day as well as the staff and children who would be with the child at any given activity. Coding sessions were composed of two parts, the

Children’s Self Perceptions

Page 18 of 75

combination of which aimed at encoding behavioral tendencies (global measures of behavior) as well as context-specific behavior (reactions to specific social events). Every child was ‘coded’ based on his or her behavior up to 6 times a day. Mean observations for each child was M = 142 over the course of the summer. To begin a coding session, staff logged in with a designated username and password. They then chose the activity to be coded. Counselors were asked to code children towards the end of the period (1 hour long), but not exceeding 30 minutes after the activity ended. Having chosen the target activity, counselors were prompted to choose a target child from a list of children at that activity. Counselors were first asked to rate how often the child displayed 8 behaviors during the preceding hour (since arriving at the activity), thus constructing a measure of overall behavior during the period. Behaviors included “withdrew/isolated,” “look sad/cried,” “whined/complained,” “argued/quarreled,” “teased/bossed,” “hit/pushed/ attacked,” “talked prosocially,” and “complied/cooperated,” each of which required a rating on a given scale of “not at all,” “somewhat,” “moderately,” or “a lot.” (Additional details can be found in the Palm Coding Manual in Appendix C.) Next, staff were prompted to evaluate how often the child experienced different the types of social events during the activity. To do this, the same frequency scale described earlier was used. The events were: “peer tease/boss,” “peer argue/quarrel,” “peer ask/tell,” “peer talked prosocially,” “adult disciplined/punished,” “adult warn/reprimand,” “adult gave instructions,” and “adult talked prosocially.” Details on these code categories and their definitions can be found in the Palm Coding Manual, included in Appendix C. Finally, the coding session collected further information about children’s responses to events that occurred. To do this, two events that were coded as having happened in the previous

Children’s Self Perceptions

Page 19 of 75

coding step were chosen at random for further probing. For each probed event, the coder was asked to identify the peer or adult who had most recently performed the event being probed. Then s/he was asked to identify the reaction that the child displayed and to what extent this behavior was displayed. Using the same behaviors as those probed in the initial, global behavior analysis (see above), the coder was instructed to indicate whether the response occurred “somewhat,” “moderately,” or “a lot”; otherwise the response field was left unchecked. Counselors were then prompted to rate how the peer or adult interactant reacted to the target’s behavior. Because this part of the behavior chain was not used in the current analysis, it will only be described briefly. Full details concerning this step can be found in the Palm Coding Manual included in appendix C. Different response options were given based whether the event involved a peer or adult interaction. After this final step in the coding process, the coder was finished with the child’s code for that activity period. Preliminary data analysis In order to assess change in behavior rating over time, two time periods were defined. RTS interviews were administered between days 14 and 18 and then again between days 35 and 39. Thus, PALM data from the two weeks preceding both RTS administrations, days 1-14 and days 22-35, were used. Three behavior scales were created based on children’s self-perceptions. These included aggression, withdrawal, and prosociality. Using a varimax rotated factor analysis, items on the RTS scale were grouped to form 5 scales of response tendencies. These included “anger,” “aggression,” “talk,” “prosocial,” and “withdrawal.” The anger scale (Cronbach’s α=.87) was composed of 3 items: “feel upset,” “feel angry,” and “feel frustrated.” The aggression scale (α=.84) was made up of 3 items: “tease or boss,” “push or hit,” and “shout or yell.” The talk scale (α=.76) was composed of 2 items: “talk calmly” and “feel calm.” The prosocial scale

Children’s Self Perceptions

Page 20 of 75

(α=.67) was composed of “tell yourself what happened isn’t worth getting upset about,” “ask someone for help,” and “do something to forget about what happened.” Finally, the withdrawal scale (α=.85) was composed of “feel sad” and “look sad or cry.” Items were required to have a minimum loading of .5 to be considered a member of a scale. Aggression, withdrawn, and prosocial scales were chosen for further analysis because these behaviors were observable and could thus be measured by Palm coders. Scale scores for context-specific and global observations were composed to best match children’s self-perception scales (aggression, withdrawal, and prosocial behavior). The aggression scale for context-specific and global observations was the aggregate of “argued/quarreled,” “teased/bossed,” and “hit/pushed/attacked.” Similarly, the withdrawal scale was the aggregate of “withdrew/isolated” and “look sad/cried.” Finally the prosocial scale was composed of “talked prosocially.” All three behavior scales for all three measurement types, children’s self-perceptions, context-specific observations, and global observation’s, were then further aggregated over peer and adult situations based on correlations between children’s perceived responses to peers and their perceived responses to adults, both at time 1 and at time 2. (Details on these results are provided in Appendix D). In all cases, correlation coefficient r was significant (p < .01). Child Categorization. In order to examine hypothesized individual differences in children’s self-perceptions and behavior, three groups of children were formed based on TRF measures of aggression. To assess whether TRF aggression scores could be aggregated over the three time points, means and standard deviations of the AGG scores were examined at each time point. A repeated measure ANOVA revealed that there were neither time effects, F(2,242) = 1.55, p < .22 nor time by group interaction effects, F(4,242) = .805, p < .52. Over the three TRF time points Cronbach’s alpha was .82. TRF aggression scales were therefore averaged over the

Children’s Self Perceptions

Page 21 of 75

three administrations. The mean TRF aggression scores were then used to identify children who were reported as displaying low, moderate, and high levels of aggression. Cutoff points for levels of aggression were determined based on tertiales. Children with mean aggression scores lower than 65.67 were designated low, between 65.68 and 72.00 were designated as being moderately aggressive, and 72.01 and above represented the highly aggressive cohort.

Children’s Self Perceptions

Page 22 of 75

Results Temporal Stability of Perceptions and Behavior Though not the main focus of this study, analyses of temporal stability were performed to address possible alternative interpretations of other findings. For each behavior scale (aggression, withdrawal, and prosocial), correlations (Pearson’s r) between time 1 and time 2 were examined for all three types of behavior measures (children’s self-perceptions, contextspecific observations, and global observations of overall behavior rates). Table 1 reveals four main findings. First, the correlations over time for all behaviors and scales were significant (p < .01) with the exception of context-specific prosocial reactions (r = .17). Scatterplots examined the possibility of a floor effect for this exception, which would indicate that prosocial behavior in response to aversive situations was rarely, if ever, observed. Results revealed a slight floor effect, which would only partially explain the absence of a correlation over time for prosocial behavior. Second, children’s self-perceptions were significantly correlated over time for all behaviors (mean r = .63). Third, observations of context-specific reactions were correlated over time to a lesser extent, with a non-significant correlation for prosocial behavior. Overall, observations of context-specific reactions were less strongly correlated (mean r = .34). Fourth, observations of global behavior, like children’s selfperceptions, showed high correlations for all behaviors (mean r = .63). Though not the main focus of this project, additional analyses were performed to examine possible effects for age and/or gender. Temporal stability of children’s self-perceptions, contextspecific observations, and global observations were reasonably consistent across age and gender groups (see Appendix E). Consistent with Table 1, both children’s perceptions and observations of global behavior demonstrated strong correlations across age and gender lines, while contextspecific observations showed weaker correlations.

Children’s Self Perceptions

Page 23 of 75

Summary. These analyses showed that assessments for aggressive, withdrawn, and prosocial behavior were significantly correlated at time 1 and time 2, both for children’s selfperceptions and for global observations of behavior. Observed reactions to aversive situations showed significant but more modest correlations for aggressive and withdrawn behavior, and a low, non-significant correlation for prosocial behavior. Further implications for temporal stability will be addressed in the Discussion section. Concurrent Validity: Relations Between Perceptions and Behavior Context-specific reactions. Next, children’s self-perceptions were correlated with context-specific observations to aversive events and with global observations in order to assess the degree to which children’s perceptions were related to extensive field observations of their behavior. Table 2 highlights several differences between aggressive, withdrawn, and prosocial behavior. First, children’s self-perceptions and context-specific observations were significantly correlated (p < .01) for aggressive behavior at all time points. Second, withdrawal was correlated between these measurement types at time 2, though not at time 1. Third, children’s perceptions of their own prosocial reactions to aversive events were not at all related to staff’s observations of their reactions to these same events. Using the same approach described previously, these analyses were repeated for age and gender groups. Once again, the results for concurrent validity did not show clear age or gender effects. See Appendix F for details. Overall behavior rates. The preceding analyses examined the relation between children’s self-perceptions and context-specific observations; thus, the behavioral observations dealt with children’s responses to the same aversive events that were assessed in the Children’s SelfPerception Interview. In contrast, this section examines the linear relatedness of children’s perceptions of their behavior in response to aversive situations and global observations of their

Children’s Self Perceptions

Page 24 of 75

behavior (i.e., the overall frequency of their behavior, regardless of the specific context in which it might have occurred). The analysis of concurrent validity of children’s self-perceptions and global observations of behavior shows two main findings (see Table 3). First, aggressive responses were significantly correlated over measurement types (p < .01). Pearson’s r for aggression was highest at Time 1 compared to Time 2. Second, no significant relation for withdrawn or prosocial behaviors was found at either time point or for the aggregate over the time points. Appendix F and G show that there are no main effects for age or gender in this concurrent validity analysis for either children’s self-perceptions and context-specific observations (F) or children’s self-perceptions and global observations (G). Summary. Children’s perceptions of their own aggressive behavior in response to aversive events were significantly correlated with field observations of aggression (both contextspecific reactions to aversive events and global behaviors). Children’s perceptions of their own withdrawal behavior were significantly correlated with context-specific assessments only at time 2. Finally, there was no relation between children’s perceptions of their prosocial behavior and staff’s report of the same behavior, either for context-specific observations or for overall behavior rates. Children’s perceptions and staffs observations of change over time The preceding analyses of both temporal stability and concurrent validity assessed the linear relatedness of children’s self-perceptions and behavior observations, but this correlational approach does not examine possible mean-level changes over time. To address possible meanlevel changes, this section examines change over time of children’s self-perceptions, contextspecific observations, and global observations, using 2 (time) x 3 (behavior: aggression, withdrawal, prosocial) repeated-measures ANOVAs. To simplify the presentation, main effects

Children’s Self Perceptions

Page 25 of 75

for behavior are presented for each ANOVA before presenting the effects for time and the time x behavior interactions. Standardized data sets were used in order to present the later time and time x behavior effects clearly. Main Effects of Behavior Repeated-measures ANOVAs revealed a main effect of behavior for each measurement type: children’s self-perceptions, context-specific observations, and global observations. All results for repeated-measures effects and their interactions are based on Greenhouse-Geiser adjustments. For children’s self-perceptions, a main effect for behavior was obtained, F(2, 246) = 10.66, p < .001. As shown in Figure 1 (left panel), children perceived themselves to be more likely to show prosocial responses to aversive events than either aggressive or withdrawn responses. Pairwise comparisons reveal that effects of behavior were found to be significant for children’s self-perceptions of aggressive and prosocial behavior, F(1, 123) = 9.86, p < .01, as well as for withdrawn and prosocial behavior, F(1, 123) = 21.15, p < .001. No significant effect for aggression and withdrawal was found, F(1, 123) = 1.44, p < .23. Non-significant pairwise analyses will not be reported in future analyses. For children’s observed reactions to aversive situations, a main effect of behavior was found, F(2, 246) = 54.29, p < .001. As shown in Figure 1 (middle panel), children were most often observed being aggressive in response to aversive events, sometimes withdrawn, and least often prosocial. Pairwise comparisons reveal a significant effect for all three behaviors: aggression and withdrawal, F(1, 123) = 23.37, p < .001, aggression and prosociality, F(1, 123) = 116.94, p < .001, and withdrawal and prosociality, F(1, 123) = 29.59, p < .001. Thus, children’s observed reactions contrasted sharply with their self-perceptions. A main effect was also found for children’s overall behavior frequencies, F(2, 246) = 916.00, p < .001. In general, children displayed prosocial behavior much more often than either

Children’s Self Perceptions

Page 26 of 75

aggressive or withdrawn behavior. Withdrawn behavior was displayed more than aggressive behavior. Nonetheless, pairwise comparisons found a main effect for all three behaviors: for aggression and withdrawal, F(1, 123) = 64.96, p < .001, for aggression and prosociality, F(1, 123) = 1306.61, p < .001, and for withdrawal and prosociality, F(1, 123) = 817.43, p < .001. Effects of Time and Time x Behavior Interactions The preceding results demonstrate (sometimes large) main effects of behavior. It is possible that these effects could obscure the presentation of effects for time and the time x behavior interactions. In order to highlight the latter effects, the figures presented next were based on standardized dependent variables. First, the marginal means for behavior (i.e., the means over both time points) were removed, yielding deviations from the mean. Second, the standard deviation was computed for each behavior at each time point, and the pooled standard deviation for each behavior was then computed over the two time points. Finally, the data were rescaled by dividing the deviation scores by the pooled standard deviations for each behavior. Children’s perceptions of their behavior in response to aversive events showed a main effect of time, F(1,123) = 6.40, p < .01. The time x behavior interaction was not significant, F(2, 246) = 1.91, p > .15. As shown in Figure 2 (left panel), children gave higher frequency ratings for all three behaviors at time 2. For the aggressive reactions, a one-way ANOVA revealed a significant effect for time, F(1,123) = 10.70, p < .001. No significant effect of time was observed for either withdrawal, F(1,123) = .20, p < .66, or prosocial behavior, F(1,123) = 1.69, p < .20. For all subsequent analyses in this section, only significant effects for time or time x behavior interactions are reported. In contrast with children’s self-perceptions, context-specific observations to peer and adult aversive situations showed no effect of time but did show a time x behavior interaction, F(2,246) = 8.31, p < .001. Like children’s perceptions of behavior, context-specific observations

Children’s Self Perceptions

Page 27 of 75

also show a sharp increase in aggressive behavior, F(1,123) = 9.40, p < .001 and no significant change in withdrawn behavior. Unlike children’s perceptions, however, context-specific observations show a significant decrease in prosocial behavior, F(1,123) = 7.36, p < .01. For children’s overall behavior rates, a main effect for time, F(1,123) = 100.65, p < .001, and a significant time x behavior interaction was found, F(2,246) = 25.98, p < .001. As shown in Figure 2, children showed a significant increase in aggressive behavior, F(1,123) = 26.15, p < .001. Like children’s self-perceptions, observations of global behavior indicate an effect of time, F(1,123) = 100.65, p < .001, indicating an overall increase in behavior frequency (Figure 2, panel 3). Summary. These analyses reveal three main findings in regards to the frequency of behavior, the effects of time, and interactions between time and behavior. First, both children’s self-perceptions and global observations report that prosocial behavior was most frequent while aggression and withdrawal were less frequently reported. Context-specific observations, on the other hand, show a stark difference in behavior frequency, with aggression being the most frequent response, withdrawal less common, and prosociality least common. Second, both children’s self-perceptions and global observations show an increase in prosocial behavior over time. Once again, context-specific observations do not show change over time for this behavior. Third, both context-specific observations and global observations showed an interaction effect of time x behavior, while children’s self-perceptions did not. Individual Differences in Child Aggressiveness Current research suggests that aggregation of children who display problem behavior can lead to iatrogenic behavioral effects. Increased aggression in children’s self-perceptions, context-specific observations, and global observations support this idea. This section examines whether or not certain types of children understand or experience these iatrogenic effects

Children’s Self Perceptions

Page 28 of 75

differently. To do this, analyses were performed for children who displayed low, middle, and high levels of aggression, using aggression groups described in the Method section. Analyses are presented in two ways. First, the relationship between children’s self-perceptions and context-specific observations will be assessed. Relative levels of behavior and change over time will be noted. Second, examination of the magnitude of behavior change will assess relative iatrogenic effects of treatment for low, middle, and highly aggressive groups. For simplicity, the relationship between children’s self-perceptions and global observations of behavior based on aggression group will not be examined. It may be noted, however, that over all groups, global observations showed that aggression and prosocial behavior increased significantly over time while no change was observed for withdrawn behavior. Details can be found in Figure 3, row 3. Children’s Self-Perceptions and Context-Specific Observations For all aggression groups, the relationship between children’s self-perceptions and context-specific observations of behavior in response to aversive events shows striking similarity (Figure 3, rows 1 and 2). First, children who were low on aggression (Figure 3, column 1) displayed moderate and unchanging levels of withdrawn and prosocial behavior according to both self-perceptions and context-specific observations. As would be expected, this group showed low levels of aggression. Importantly, aggression increased significantly according to both measures. Second, children who showed moderate levels of aggression (Figure 3, column 2) displayed average (or close to average), unchanging levels of withdrawn and prosocial behavior on both perception and context specific observation scales. For both measurement types, aggression levels were low relative to withdrawn and prosocial behavior, but only children’s self-perceptions reported an increase over time. Third, highly aggressive children (Figure 3, column 3) reported low and unchanging levels of withdrawn and prosocial behavior, both in children’s self-perceptions and context-specific observations of behavior. As would be

Children’s Self Perceptions

Page 29 of 75

expected, aggression levels were high for both measures, though neither showed a significant change over time. Trends in Behavior Change According to both children’s self-perceptions and context-specific observations, aggressive behavior was the only behavior to show change over time. The greatest change (p < .001) was found in the lowest aggression group (Figure 3, column 1). For moderately aggressive children, significant change was found only in children’s perceptions of aggression (p < .05) but not according to context-specific observations. For both low and moderately aggressive groups, however, children’s self-perceptions of change were larger than context-specific observations. Finally, highly aggressive children showed no change in aggressive behavior for either measurement type. Summary. Separating children based on levels of aggression reveals an impressive likeness between children’s self-perceptions and context-specific observations of behavior in response to aversive events. No change was found for withdrawn or prosocial behavior. According to both measures, children who showed low levels of aggression increased the most in this behavior. Children who were moderately aggressive showed a steep increase in levels of aggression according to children’s perceptions only. Highly aggressive children demonstrated high, unchanging levels of aggression.

Children’s Self Perceptions

Page 30 of 75

Discussion The goal of this study was to examine behavioral awareness in children with serious emotional and behavioral disorders by comparing children’s self-perceptions of behavior with both context-specific and global observations of behavior. This research extends past work by bridging the gap between social development studies of emotion regulation and self report (Hubbard et al., 1994; Halberstadt et al., 2001) and personality assessment research that focuses on the cross-situational consistency of behavior (Wright, 2001). Past studies of the interconnectedness of children’s behavior and their social experience (Cassidy et al., 1992; Parker et al., 2001) provided the theoretical groundwork and called attention to the need to examine the relationship between behavioral self-perceptions and field observations of behavior. However, unlike previous social development studies, this work quantified the relationship between self-perceptions and observations of behavior, rather than emotion (Casey, 1993), which has proven to be difficult to measure (Halberstadt et al., 2001). Unlike previous behavioral studies that highlight the importance of both context-specific and global measures of behavior for assessing complex patterns of change (Choukas-Bradley et al., 2009), this study considered children’s own self-perceptions of behaviors as an important factor understanding behavior change. Several findings emerge from the present study that challenge past findings regarding self-perception, inform the study of the relationship between children’s self-perceptions of behavior in context, and suggest future directions of study that may have important developmental and clinical implications. This section will review each of these findings and address the ways in which each one is pertinent and important to the field. Temporal Stability High temporal stability (i.e., correlations over time) was found for children’s selfperceptions and global observations of behavior, and moderate correlations were found for

Children’s Self Perceptions

Page 31 of 75

context-specific behavioral observations. Stability over time for children’s self reports suggests that the reports were not random and that children were engaged in the self-assessment task. Factors such as inattention, fatigue, defensiveness, as well as genuine individual differences in behavior change would have caused low temporal stability; it is clear that this did not occur. This result, that children’s self-perceptions did not fluctuate randomly over time, lays the foundation for later findings. Past studies reviewed here assessed the correlation between selfreport (of emotion) and observed affective response at one time point and did not examine any stability either in children’s perceptions or expressions of emotion over time. Though investigators such as Hubbard and Coie (1994) acknowledge that children’s ability to perceive emotions changes over time as a result of maturation, stability of perceptions on a smaller time scale were not addressed. Of particular note are findings by Halberstadt et al. (2001) that children with severe emotional and behavioral disorders are presumably less able to reflect on their affective experience and tend to express emotions impulsively. This would suggest that children’s self-perceptions of their reactions to aversive situations might be erratic rather than temporally stable. The relative discrepancy between the temporal stability of children’s self-perceptions of their reactions to aversive situations versus the observations of their context-specific reactions to those situations should be interpreted with care. One possibility is that in reporting their selfperceptions, children are not sensitive to how their reactions to these situations may be changing over time. In this view, the temporal stability for the context-specific field observations is lower because multiple objective observers may be more objective and more sensitive to subtle changes in children’s behavior. Another possibility is that differences in reliability between the two measurements affected the results. For example, the frequency with which children’s behavior in response to aversive events was observed necessarily depends, in part, on the

Children’s Self Perceptions

Page 32 of 75

frequency with which children encountered these events. As a result, context-specific observations of aversive situations are made less frequently than global observations of behavior. A third possibility is that certain context-specific reactions showed a floor effect. For example, children may have rarely displayed prosocial behavior in aversive situations, in which case assessing the temporal stability of this variable using correlations looses meaning. However, scatterplots revealed that this floor effect was minimal, thus providing only a partial explanation of the low temporal stability that was found for prosocial reactions. Concurrent Validity Correlations between children’s self-perceptions and field observations showed that concurrent validity over varied over behaviors. For aggression, correlations between children’s self-perceptions and both context-specific and global observations were significant for all time points. For withdrawal, children’s self-perceptions showed relatively little accuracy either for context-specific or global observations of behavior. Nonetheless, correlations between selfperceptions and context-specific observations were relatively higher. Correlations between selfperceptions and global observations were essentially zero. For prosocial behavior, correlations between children’s self-perceptions and all observations were zero. Although past studies have examined the accuracy of self-perceptions, these findings extend past work because they examine more than a single behavior (Campos et al., 1989). Given that there was more agreement for aggressive behavior than for withdrawn or prosocial behavior, overarching conclusions about the accuracy of self-perceptions are problematic. Behavioral Effects Examination of mean -levels of behavior reveal that overall reports of behavior frequencies showed a striking similarity between children’s self-perceptions and global measures of behavior. Both children’s self-perceptions and global observations of behavior reported that

Children’s Self Perceptions

Page 33 of 75

the frequency with which children displayed prosocial behavior was significantly higher than aggressive or withdrawn behavior. Context-specific observations of behavior revealed an inverted affect: aggressive behavior was reported most frequently, withdrawn behavior was reported to a moderate extent, and prosocial behavior was very infrequent. This result is odd because both children’s self-perceptions and context-specific observations of behavior were assessed in response to aversive situations. One would predict that the frequency of aggressive, withdrawn, and prosocial behavior would be closest between these two measures, as they are precipitated by the same, negative events. However, children’s self-perceptions of behavior reflected those measured by global observations. The finding that children’s self-perceptions of their reactions to aversive situations most closely resemble global behavior observations seems to speak to the process of self-reflection. Context-specific and global observations show that children act very differently in response to aversive situations (mostly aggressively) than they do overall (mostly prosocially). One interpretation is that children’s report of high levels of prosocial behavior in response to aversive events may reveal a fundamental misconception that children make in reflecting on their behavior. Namely, it is possible that children are unaware of situational specificity, and see themselves most vividly in terms of global behavior. A second possible interpretation of this finding is that children felt the need to give “good” answers with respect to their behavior in aversive situations, leading them to be more hesitant in reporting aggressive behavior. The extent to which individuals perceived withdrawn behavior as being a desirable response would have affected whether children felt pressure to emphasize or ignore this behavior. A third interpretation is that children know more about their behavior than is revealed by contextspecific observations in a clinical treatment setting and use some of that insight in their selfperception reports.

Children’s Self Perceptions

Page 34 of 75

Change Over Time Diverse patterns of change over time in children’s self-perceptions, context-specific observations, and global observations of behavior indicate differences in perceptions and observations of change. In all cases, aggressive behavior increased significantly over time, suggesting an iatrogenic effect of treatment. Unlike this study, Choukas-Bradley et al. (2009) examined reactions to positive and negative events, finding that patterns of change can be entirely different depending on the nature of the situation. Specifically, a sample of “most improved” children were found to increase in aggression and decrease in prosocial behavior in response to aversive events while simultaneously decreasing in aggression and increasing in prosocial behavior in response to positive events (Choukas-Bradley et al., 2009). While contextspecific observations of behavior in this study replicated the findings of Choukas-Bradley et al. (2009) in regards to responses to aversive events, consideration of children’s self-perceptions of their behavior in response to these events may have important implications for iagtrogenic effects of treatment. Not only does aggression increase over time, but children seem to be aware of this shift. However, while children reported that they were aware of an increase in aggression, their self-perceptions of prosocial behavior suggest that they have less insight on this type of behavior. Prosocial behavior showed little consistency across measurement types; children’s self-perceptions of behavior reported no change while context-specific observations revealed a significant decrease in prosocial behavior. One interpretation for this apparent disconnect in the accuracy of children’s perceptions is that aggression is a behavior that is more readily perceived compared to prosocial behavior. Another interpretation is that children have a hard time recognizing their own behavior in a context-specific manner. This latter explanation is supported in this study’s findings for the main effects of behavior, in which overall frequencies of behavior according to children’s

Children’s Self Perceptions

Page 35 of 75

perceptions closely resembled those for global behavior observations. Yet, this explanation would suggest that there would be agreement between children’s self-perceptions and global observations of change. This is certainly the case for the sharp increase in aggressive behavior, but not at all for prosocial behavior, which shows a sharp and significant increase according to global observations. While many studies highlight the ways in which context-specific observations are necessary to understand complex patterns of change (Choukas-Bradley et al., 2009; Zakriski et al., 2007), children’s own perceptions of change highlight the ways in which these patterns are poorly understood. It is possible that children’s understanding of their own behavior change affects the outcome of treatment, specifically the transfer process to another environment. Given that the variability of change over time has been shown to be dependent on context (ChoukasBradley et al., 2009; Zakriski et al, 2007), an important direction for future research would be to examine children’s self-perceptions of their behavior in response to positive events, over the course of treatment. Aggressive Child Type Adverse effects of aggregating problem children have been well documented, particularly with respect to children who show low, initial levels of problem behavior (Dodge et al., 2006; Silver et al., 2006). This study lends support to the notion that exposure to deviant peers causes children who rarely display aggressive behavior to increase that behavior. However, by comparing children’s self-perceptions of behavior with context-specific observations of behavior in response to aversive events for low, moderate, and highly aggressive child groups, this study contributes to a better understanding of iatrogenic effects. Our results demonstrate that while context-specific observations show a significant increase in aggression for children in the low aggression group, children in both the low and

Children’s Self Perceptions

Page 36 of 75

moderately aggressive group tended to overestimate this iatrogenic effect. According to children’s self-perceptions in the low aggression group, increases in aggression reached a higher level of significance than context-specific observations did. According to children’s selfperceptions in the moderately aggressive group, aggression increased significantly, whereas context-specific observations of aggression showed no significant change. This suggests that children who experience iatrogenic effects as a result of clinical intervention may be hyperaware of their increased aggression. Evidence collected by this study (e.g. correlational analysis of concurrent validity) suggests that aggression may be more easily perceived than prosocial or withdrawn behavior. Therefore, if children who experience iatrogenisis are hyperaware of increases in aggression and are simultaneous less aware of any change in prosocial or withdrawn behavior, it may be that children who display adverse effects of treatment have an exceptionally negative view of their own behavior. Importantly, Dishion et al. (2006) suggests that an essential mediator in treatment outcome is the way in which the individual interprets its effects. This study, therefore, suggests the possibility that iatrogenic effects of treatment may not result only from the aggregation of deviant children, but also from the ways in which individuals interpret their own behavior. Conclusions This study proposes that understanding children’s self-perceptions of their behavior may contribute to our understanding of the effects of treatment for EBD children. Findings suggest that some behaviors are more readily perceived than others, that self-perceptions of behavior frequencies tended to be inaccurate, and that perceptions of change, when considered by aggressive child type, were not only correct, but possibly exaggerated. I suggest, therefore, that in the same way that context-specific observations have been shown to be necessary in revealing complex patterns of change (Zakriski et al., 2007), so too are children’s self-perceptions of

Children’s Self Perceptions

Page 37 of 75

behavior necessary to most clearly understand and predict the affects of treatment. An important future direction for the consideration of children’s self-perceptions as mediating factors in behavior change requires the collection of self-perception data in response to a wider variety of situations. Through the consideration of children’s perceptions in response to aversive events, this study demonstrates the importance of future examination of potential trends in children’s self-perceptions in response to positive events was well as neutral events.

Children’s Self Perceptions

Page 38 of 75

References

Achenbach, T. M., & Rescorla, L. A. (2001). Manual for ASEBA School-Age Forms & Profiles. Burlington, VT: University of Vermont, Research Center for Children, Youth, & Families.

Campos, J. J., Campos, R. G. & Barrett, K. C. (1989). Emergent themes in the study of emotional development and emotion regulation. Developmental Psychology, 25(3), 394-402.

Carroll, J. J., & Steward, M. S. (1984). The role of cognitive development in children's understanding of their own feelings. Child Development, 55, 1486-1492.

Casey, R. J. (1993). Children’s emotional experience: relations among expression, Self-Report, and Understanding. Developmental Psychology, 29(1), 119-129.

Cassidy, J., Parke, R. D., Butkovsky, L. & Braungart, J. M. (1992). Family-peer connections: the roles of emotional expressiveness within the family and children’s understanding of emotions. Child Development, 63(3), 603-618.

Choukas-Bradley, S., Banducci, A. N., Metcalfe, L. A., Wright, J. C., & Zakriski, A. L. (2008). Reassessing the assessment of change: Disentangling the social interactional processes that mediate behavior change in at-risk youth. Poster presented at the Eastern Psychological Association, Boston, Massachusetts.

Children’s Self Perceptions

Page 39 of 75

Choukas-Bradley, S., Metcalfe, L. A., Wright, J. C., Zakriski, A. L., & Cardoos, S. (2009). Toward a Contextual Approach to the Study of Change: Linking Impressions, Standardized Assessments, and Behavior in Context. Poster presented at the Society for Research on Child Development, Denver, CO.

Cole, P. M., Martin, S. E., & Dennis, T. A. (2004). Emotion regulation as a construct: methodological challenges and directions for child development research. Child Development, 75(2), 317-333.

Cook, E. T., Greenberg, M. T. & Kusche, C. A. (1994). The relations between emotional understanding, intellectual functioning, and disruptive behavior problems in elementaryschool-aged children. Journal of Abnormal Psychology, 22, 205-219.

Dearing, K. F., Hubbard, J. A., Ramsden, S. R., Parker, E. H., Relyea, N., Smithmyer, C. M., & Flanagan, K. D. (2002) Children’s self-reports about anger regulation: direct and indirect links to social preference and aggression. Merril-Palmer Quarterly, 48(3), 308-336.

Denham, S. A. (1986). Social cognition, prosocial behavior, and emotion in preschoolers: contextual validation. Child Development, 57, 194-201.

DiBartolo, P. M. & Grills, A. E (2006). Who is best at predicting children’s anxiety in response to a social evaluative task? A comparison of child, parent, and teacher reports. Journal of Anxiety Disorders, 20, 630-645.

Children’s Self Perceptions

Page 40 of 75

Dishion, T. J. & Dodge, K. A. (2006). Deviant peer contagion in interventions and programs: an ecological framework for understanding influence mechanisms. In Dodge, K. A., Dishion, T. J., & Lansford, J. E., Deviant peer influences in programs for youth (253-277). New York: The Guilford Press.

Dodge, K. A., Dishion, T. J., & Lansford, J. E. (2006). Deviant Peer Influences in Programs for Youth: Problems and Solutions. New York, The Guilford Press.

Halberstadt, A. G., Denham, S. A., & Dunsmore, J. C. (2001). Affective social competence. Social Development, 10(1), 79-119.

Hubbard, J., & Coie, J. (1994). Emotional correlates of social competence in children's peer relationships. Merrill-Palmer Quarterly, 40(1), 1-20.

Kenny, D. A., West, T. V., Cillessen, A. H. N., Coie, J. D., Dodge, K. A., Hubbard, J. A. & Schwartz, D. (2007). Accuracy in judgments of aggressiveness. Personality and Social Psychology Bulletin, 33(9), 1225-1236.

Odom, R. D., & Lemond, C. M. (1972). Developmental differences in the perception and production of facial expressions. Child Development, 43, 359-369.

Parker, E. H., Hubbard, J. A., Ramsden, S. R., Relyea, N., Dearing, K. F., Smithmyer, C. M. & Schimmel, K. D. (2001). Children’s use and knowledge of display rules for anger following

Children’s Self Perceptions

Page 41 of 75

hypothetical vignettes versus following live peer interaction. Social Development, 10(4), 528-537.

Rieffe, C., Oosterveld, P., Miers, A. C., Terwogt, M. M., & Ly, V. (2008). Personality and Individual Differences. Developmental Psychology, 45, 756-761.

Saarni, C. (1979). Children’s understanding of display rules for expressive behavior. Developmental Psychology, 15(4), 424-429.

Sani, F. & Bennett, M. (2009). Children’s Inclusion of the group in the self: evidence from a self-ingroup confusion paradigm. Developmental Psychology, 45(2), 503-510.

Shoda, Y., Mischel, W. & Wright, J. C. (1993). Links between personality judgments and contextualized behavior patterns: situation-behavior profiles of personality prototypes. Social Cognition, 11(4), 399-429.

Silver, R. B. & Eddy, J. M. (2006). Research-based prevention programs and practices for delivery in schools that decrease the risk of deviant peer influence. In Dodge, K. A., Dishion, T. J., & Lansford, J. E., Deviant peer influences in programs for youth (253-277). New York: The Guilford Press.

Southam-Gerow, M. A. & Kendall, P. C. (2002). Emotion regulation and understanding implications for child psychopathology and therapy. Clinical Psychology Review, 22, 189222.

Children’s Self Perceptions

Page 42 of 75

Wright, J. C., Giammarino, M. & Parad, H. W. (1986). Social status in small groups: individualgroup similarity and the social “misfit.” Journal of Personality and Social Psychology, 50(3), 523-536.

Wright, J. C., Lindgren, K. P. & Zakriski, A. L. (2001). Syndromal versus contextualized personality assessment: differentiating environmental and dispositional determinants of boy’s aggression. Journal of Personality and Social Psychology, 81(6), 1176-1189.

Wright, J.C., Mischel, W. (1988). Conditional hedges and the intuitive psychology of traits. Journal of Personality and Social Psychology, 55(3), 454-469.

Zakriski, A. L., Wright, J. C., & Parad, H. W. (2007). Contextual versus syndromal assessment of behavior change: Uncovering patterns of adaptation to clinical intervention. Poster presented at the Society for Research on Child Development.

Zeman, J., Cassano, M., Perry-Parrish, C., Stegall, S. (2006). Emotion regulation in children and adolescents. Developmental and Behavioral Pediatrics, 27(2), 155-168.

Children’s Self Perceptions

Page 43 of 75

Table 1. Temporal Stability of Self-perceptions and Behavioral Observations for All Children Behavior

Children’s self-

Context-specific

Global

perceptions

observations

observations

Aggression

.66**

.43**

.66**

Withdrawal

.66**

.42**

.63**

Prosocial

.57**

.17

.59**

Mean

.63

.34

.63

Note. * p <.05, ** p <.01.

Children’s Self Perceptions

Page 44 of 75

Table 2. Concurrent Validity of Children’s Self-perceptions and Context-specific Observations Behavior

Time 1

Time 2

Time 1 & 2

Aggression

.42**

.35**

.47**

Withdrawal

.15

.40**

.34**

Prosocial

.02

.01

.00

Mean Aggregate

.20

.25

.27

Note. * p < .05, ** p < .01.

Children’s Self Perceptions

Page 45 of 75

Table 3. Concurrent Validity of Children’s Self-perceptions Global Observations Behavior

Time 1

Time 2

Time 1 & 2

Aggression

.55**

.33**

.49**

Withdrawal

-.08

.076

.00

Prosocial

.12

-.070

.07

Mean Aggregate

0.20

0.11

0.19

Note. * p < .05, ** p < .01.

Children’s Self Perceptions

Page 46 of 75

Figure Caption Figure 1. Main effects of behavior for children’s self-perceptions, context-specific observations, and global observations.

Figure 2. Standardized data for children’s self-perceptions, context-specific observations, and global observations.

Figure 3. Standardized data for children’s self-perceptions, context-specific observations, and global observations, by aggression group.

Children’s Self Perceptions

Page 47 of 75

1.4

Self Perceptions

AGG **

1.0

Mean

1.2

PRO

0.8

WDR

1

2 Time

0.5

Observed Reactions

0.3

WDR

0.2

Mean

0.4

AGG **

0.1

PRO **

1

2 Time

Observed Globals

1.0 0.5

WDR AGG ***

0.0

Mean

1.5

PRO ***

1

2 Time

Children’s Self Perceptions

Page 48 of 75

0.2

Self Perceptions

0.0

PRO WDR

-0.2

Mean

0.1

AGG **

1

2 Time

0.0

AGG ** WDR PRO **

-0.4

Mean

0.2

0.4

Observed Reactions

1

2 Time

0.2

AGG ***

0.0

PRO ***

WDR

-0.4

Mean

0.4

Observed Globals

1

2 Time

Children’s Self Perceptions

Page 49 of 75

1

0.5

1

2

0.0

AGG

WDR PRO

-0.5

Mean

0.0

PRO AGG * WDR

-0.5

AGG ***

Mean

0.5

0.5 0.0

WDR PRO

-0.5

1

2

2

Time

Time

Time

Observed Reactions Low Aggressive

Observed Reactions Mod. Aggressive

Observed Reactions High Aggressive

1.0

1.0

1.0

Mean

Self Perceptions High Aggressive

Self Perceptions Mod. Aggressive

Self Perceptions Low Aggressive

0.0

PRO

-1.0

-1.0

1

2

WDR

-0.5

Mean

0.5 0.0

AGG PRO WDR

-0.5

Mean

0.5 0.0 -0.5

AGG *

-1.0

1

2

2

Time

Time

Observed Globals Low Aggressive

Observed Globals Mod. Aggressive

Observed Globals High Aggressive

2 Time

0.5

WDR

-0.5

PRO ***

-1.0

-1.0

1

AGG *

0.0

0.5

AGG ** WDR

-1.0

-0.5

WDR AGG ***

PRO ***

0.0

0.0

Mean

0.5

PRO ***

Mean

1.0

1.0

1.0

Time

-0.5

Mean

PRO WDR

1

Mean

0.5

AGG

1

2 Time

1

2 Time

Children’s Self Perceptions

Page 50 of 75

Appendix A: Consent Form The Transitions Project Research Consent Form At Wediko Children’s Services, we conduct research to improve our understanding of the children we serve and the ways counselors can help them. This summer we are cooperating with Dr. Audrey Zakriski of Connecticut College (principal investigator) and Dr. Jack Wright of Brown University (co-investigator) in the final year of a 3-year research project assessing children’s behavior change in response to treatment at Wediko. This consent form describes the project and invites you to give permission for your child to be a participant in this study. If you participated in 2006 or 2007 you will notice that the project is being streamlined this year, with no data collection from teachers outside of the standard Wediko application. Purpose and Description: The goal of this research is to understand how children’s behaviors change over the course of short-term residential treatment, how best to measure these changes, and what types of change are most likely to carry over into the following school year. If you and your child agree to participate in this project, data collected as part of the Wediko admissions process (including age, sex, special education status, life events, and family history) and additional behavioral ratings will be used to measure your child’s adjustment before the beginning of the program. Counselors will provide ratings of your child’s behavior at three points during the summer. They will also provide daily behavioral observations. These data will be used to track behavior change over the course of the summer. In the fall, we will contact you to obtain follow-up behavior ratings. These will be used to measure your child’s adjustment after treatment. Two types of measures will be used at each assessment point. One focuses on overall amounts of problem and prosocial behaviors, and the other focuses on how children respond to different types of interactions with peers and adults. As with all Wediko research, these data will be analyzed during the summer to help staff learn about the children. The data will also be studied after the program ends to answer the questions just described. Confidentiality of Records: The material collected about your child for this study will be kept confidential. During the summer, data will be shared with staff who know your child to help them understand and work with him/her. All Wediko staff sign confidentiality agreements stating that they will discuss information about the children only with other Wediko staff, and only for clinical purposes. No individual information from this study will be shared with anyone outside of Wediko including teachers or school administrators. For research use, all names will be removed from the data. Background information and pre-summer behavioral ratings will be assigned a unique identification number. This identification number will also be assigned to children’s summer assessment data when it is used for research purposes. The same identification number will be assigned to parent/guardian ratings of fall behavioral adjustment when they are returned by mail to the Wediko office. The identification number will allow us to match data over these different assessments, and will protect the confidentiality of your child’s data. Wediko staff will share identifying information about your child (parent/guardian & child name, address, phone number) with the principal investigator only for the purpose of sending you the fall follow-up forms. However, the principal investigator will not know which child has been assigned which identification number. In this way, the confidentiality of your child’s data is again protected. Once the

follow-up has been conducted, this list will be destroyed. The only master copy linking names and

Children’s Self Perceptions

Page 51 of 75

identification numbers will be kept at the Wediko Children’s Services office in Boston. The results of this research may appear in publications and be presented at professional conferences. These presentations/publications will report summary data only, and individual participants will not be identified. Possible Risks and Benefits: Parents/guardians might feel uncomfortable answering questions about their child’s behavioral difficulties. Children may be nervous about adults recording information about their adjustment. Parents/guardians might also be concerned about the privacy of personal information. To minimize these potential discomforts, staff working on this study will be trained to help people feel comfortable with these assessments, to be sensitive to when people are upset, and to answer any questions that arise. Staff will remind participants that they need not answer questions that make them feel uncomfortable, and that they can stop participation without penalty. Staff will also be trained to take every precaution to insure that all information is kept private. Because this research is being supported by a grant, we are able to pay you for taking the time to complete the additional behavior rating forms required for the study. Each of the two forms takes about 15-20 minutes to complete. You will be paid $20.00 from the grant for completing the pre-summer forms, and another $20.00 for completing the forms in the fall. Data collected for this study will be used to assess and monitor children’s responses to treatment during the summer. In addition, the fall follow-up will be monitored by Wediko senior staff who will be available at your request should questions arise about a child’s adjustment during the follow-up. Findings from this research will also help guide decisions about the best ways to assess children’s responses to treatment and to evaluate the effectiveness of treatment programs in general. Alternatives to Participation: Parents/guardians may choose not to have their child participate in this study. Because it is important to monitor children’s responses to treatment, and because Wediko does not want research participants to be treated differently from nonparticipants, summer assessment data will be collected on all children and used within the summer program. However, only children with research permission will be included in analyses of the data for research purposes. Also, additional pre-summer parent/guardian ratings and the parent/guardian fall follow-up assessments will be conducted only for children with research permission. To ensure that your decision about research participation will not affect Wediko admissions decisions, your research consent form will be shielded from research and clinical staff until after admissions decisions are made. Research Approval and Contact Information: This project has been approved by the Connecticut College Institutional Review Board, Dr. Ann Devlin (chair), and is subject to the guidelines for the ethical treatment of human subjects set forth by both the American Psychological Association, and the National Institutes of Health. Participants who want more information about their rights as participants or who want to report a research related problem may contact Dr. Ann Devlin at 860-439-2333. If you have any questions about this study or about research at Wediko, please contact Dr. Harry W. Parad, Executive Director, Wediko Children’s Services (617) 292-9200. You may also contact Dr. Audrey Zakriski at (860) 439-5134 or Dr. Jack Wright at (401) 863-3974 with specific questions about this study. We would very much like for you and your child to be a part of this project. Thank you for your consideration.

Children’s Self Perceptions

Page 52 of 75

Parent/Guardian Consent Agreement I understand the above description of the behavior change study (The Transitions Project) being conducted at Wediko in collaboration with Dr. Audrey Zakriski of Connecticut College and Dr. Jack Wright of Brown University. I believe that I have been fully informed about the project. I also certify that the participant is under eighteen and that I am the parent (guardian) of the participant. Complete and sign either A if you give permission OR B if you do not give permission for your child to participate in this research. A. I authorize Wediko, Dr. Zakriski, Dr. Wright, and authorized personnel to conduct the research as described above. This includes contacting me before the summer program and in the fall for behavioral adjustment ratings of my child. Name of child ____________________________________________________________ Name of Parent/Legal Guardian: _____________________________________________ Signature of Parent/ Legal Guardian __________________________________ Date:______________ Or B. I do not give permission for my child to participate in this research. Name of child ____________________________________________________________ Name of Parent/Legal Guardian: _____________________________________________ Signature of Parent/ Legal Guardian __________________________________Date:______________

Child Assent Agreement I understand the above description of the behavior change study (The Transitions Project) being conducted at Wediko in collaboration with Dr. Audrey Zakriski of Connecticut College and Dr. Jack Wright of Brown University. I believe that I have been fully informed about the project. Sign either A if you agree OR B if you do not agree to be a participant in this research. A. I authorize Wediko, Dr. Zakriski, Dr. Wright, and authorized personnel to conduct the research as described above. This includes contacting my parent/guardian before the summer program and in the fall for behavioral adjustment ratings of me. Signature of child _________________________________________________________ Or B. I do not agree to be a participant in this research. Signature of child _________________________________________________________

Boston Offices 72-74 East Dedham Street Boston, MA 02118 Phone: (617) 292-9200 Fax: (617) 292-9275 [email protected]

New Hampshire Campus

Wediko Children's Services, Inc. New Hampshire Summer Program www.wediko.org Tax ID #04 6002778

11 Bobcat Boulevard Windsor, NH 03244 Phone: (603) 478-5236 Fax: (603) 478-2049 [email protected]

Children’s Self Perceptions

Page 53 of 75

Appendix B: Reactions to Situations Interview Child/student’s name: _________________

Group: ________

Interviewer: ___________

Part 2 – Self-Reports of Reactions to Situations (Interviewer: Read the following.) Now I am going to ask some questions just about you – so we don’t need these pictures anymore. (Put photographs aside.) I am going to tell you about things kids and adults might say or do to you. Then I’m going to ask you what you might feel or do if those things happened. Here’s an example of how each question will work. Suppose I ask you, “Since Wediko started, how often was it noisy in the dining hall during dinner time?” You could answer by saying “never,” “sometimes,” “often,” or “very often.” Do you have any questions? Situation #1. Imagine you are with your group at an activity you enjoy. You are working really hard and feeling proud of what you are doing. You show your project to another kid in your group and s/he starts making fun of it and teasing you about it. If something like this happened to you, how often would you… (Interviewer: Repeat prompt and response options as needed.) (Circle one answer for each item)

Feel upset

Never

Sometimes

Often

Very often

Feel sad

Never

Sometimes

Often

Very often

Feel angry

Never

Sometimes

Often

Very often

Feel frustrated

Never

Sometimes

Often

Very often

Feel calm

Never

Sometimes

Often

Very often

Move away or get quiet

Never

Sometimes

Often

Very often

Tease or boss the other kid

Never

Sometimes

Often

Very often

Look sad or cry

Never

Sometimes

Often

Very often

Tell yourself what happened isn’t worth getting upset about

Never

Sometimes

Often

Very often

Push or hit the other kid

Never

Sometimes

Often

Very often

Ask someone else for help

Never

Sometimes

Often

Very often

Do something to forget about what happened

Never

Sometimes

Often

Very often

Talk to the other kid calmly

Never

Sometimes

Often

Very often

Shout or yell at the other kid

Never

Sometimes

Often

Very often

How hard would it be to get over being upset about a kid treating you this way?

Children’s Self Perceptions

Page 54 of 75

Easy

A little hard

Hard

Very hard

How quickly would you get over being upset about a kid treating you this way? Right away

Continued on next page...

In a few minutes

A long time

A very long time

Children’s Self Perceptions

Page 55 of 75

Situation #2. Imagine you’re with your group at an activity you like and you are excited to get started. You say something to another kid while an adult is explaining the rules. The adult comes over to you and warns you that if you don’t stop talking you will lose your turn. If something like this happened to you, how often would you… (Interviewer: Repeat prompt and response options as needed.)

(Circle one answer for each item)

Feel upset

Never

Sometimes

Often

Very often

Feel sad

Never

Sometimes

Often

Very often

Feel angry

Never

Sometimes

Often

Very often

Feel frustrated

Never

Sometimes

Often

Very often

Feel calm

Never

Sometimes

Often

Very often

Move away or get quiet

Never

Sometimes

Often

Very often

Argue or talk back to the adult

Never

Sometimes

Often

Very often

Look sad or cry

Never

Sometimes

Often

Very often

Tell yourself what happened isn’t worth getting upset about

Never

Sometimes

Often

Very often

Push or hit the adult

Never

Sometimes

Often

Very often

Ask someone else for help

Never

Sometimes

Often

Very often

Do something to forget about what happened

Never

Sometimes

Often

Very often

Talk to the adult calmly

Never

Sometimes

Often

Very often

Shout or yell at the adult

Never

Sometimes

Often

Very often

How hard would it be to get over being upset about an adult treating you this way? Easy

A little hard

Hard

Very hard

How quickly would you get over being upset about an adult treating you this way? Right away

In a few minutes

That’s the end of the interview. Thank you for your help.

A long time

A very long time

Children’s Self Perceptions

Page 56 of 75

Appendix C—Palm Coding Manual

Wediko Behavior Observation (WBOS) Coding Manual 2008 Palm Version

Children’s Self Perceptions

Page 57 of 75

Introduction to the 2008 Wediko Behavior Observation System Wediko is now in the third year of an NIH-funded project designed to improve our understanding of how children’s social interactions change over the summer, and how these changes are linked to their adjustment at home and school. The summer component of the project involves extensive observations of children’s interactions with peers and adults. Each day, over 70 hand-held computers (PDAs or “Palms”), programmed with the schedule for that day, will be distributed to staff members and Think City teachers. At the end of certain activities, staff will use the Palms to record information about children’s interactions during that period. The research team will collect the Palms each day, upload their data, and prepare them for the next day. Analyses of children’s behavior change will be provided to supervisors for possible review in clinical supervision meetings. The Wediko Behavior Observation System (or WBOS) has a long history. In the early 1980’s, staff completed weekly “checklists” on the children in their groups. This evolved into a system that focused on specific social events children experienced throughout the day (e.g., being teased) and how they responded to those events (e.g., withdraw, counterattack). A standardized version of the WBOS provided observations of over 600 children in a five-year study. Much of what we have learned and published about the behavior of aggressive, withdrawn, and “mixed” or comorbid children is based on these and related data collected elsewhere. The Palm version of the WBOS incorporates the strengths of the older system and avoids its weaknesses. At the end of an activity, the counselor provides information about events a target child encountered, how that child responded to those events, and how the other person involved in the interaction responded next. For example, suppose Tommy is the target child, and he was observed interacting with another child, Charlie. The interaction might look like this: Charlie teases Tommy  Tommy threatens Charlie  Charlie hits Tommy The Palm allows you to record this kind of information efficiently and quickly. It also obtains other information that is needed to interpret children’s behaviors. It does this by prompting you with the relevant questions and by providing the names of the children and counselors present in your activity. Your answers are recorded with simple “clicks” on the Palm screen and automatically stored for you. The best way to learn how to use the Palm is to practice with one. We will do this soon. In the meantime, this document provides an overview of the main elements of the system, including definitions of the codes and rating scales that will be used.

Children’s Self Perceptions

Page 58 of 75

Over view of Obser vation Schedule and Guidelines for Palm Use The observation schedule varies slightly over developmental programs. On Sundays with special activities, no data are collected. These and other details will be explained in program-specific meetings. For now, we provide a general overview of the schedule. 1. Pickup. At breakfast, designated counselors from each group or activity will pick up a WBOS fanny pack in Ames Hall. The pack will contain one or more PALMs. Each Palm has a unique ID number and each Palm will be assigned to a specific group or activity. The Palms cannot be exchanged. 2. Time of coding. Immediately following each “coded” activity, staff will complete a coding session for that period (see schedule below). It is important that the Palm coding not be done later in the day. If an emergency prevents you from completing the coding immediately, you should note this in the log book provided with the PALM. Do not try to “reconstruct” what happened later in the day, even if you think you remember. 3. Activities to be coded. A typical day is outlined at right. In general, children will be coded in two group activity periods, two sign-up periods, one Think City period, and in group therapy. Coding will be divided up among the adults present. This includes activity counselors, residential counselors (including lead staff), and Think City teachers. 4. Drop-off. By dinner, all Palms should be returned to the research station in Ames Hall. Some staff may prefer to drop off their Palms in the afternoon. Overnight, the Palms will be prepared for another day of coding. If you cannot return your Palm at dinner, you should contact a member of the research team. It is difficult for the team to prepare the Palms for the next day when they are returned later in the evening. 5. Guidelines for using PALMs. The Palms are valuable and difficult to replace. We have done everything we can to protect them, but we need your help. Please follow the guidelines at left. Report any problems you have with the Palms using the logbooks provided in your pack. The research team will

Children’s Self Perceptions

Page 59 of 75

respond to your question as soon as possible. Thank you in advance for your cooperation. 6. Answering children’s questions about PALMs. Children may ask questions about the Palms and may want to use them. If a child asks about what you are doing with a PALM, you should explain that we study how children interact with peers and adults to help us learn more about how children change over the summer. If the child wants more details, you could explain that you rate everyday behaviors like talking, arguing, teasing, and so on. If a child asks, you might also explain that the information in the Palm is used only to help us learn about children; it is never used to determine group or individual incentives, and is not part of children’s contracts or clinical checklists. All parents have been informed about the project and how the data will be used. Children must not be allowed to play with the PALMs. Games have been removed as far as possible and Palms are difficult to replace if damaged. 7. Features of the Palm Z-22 ™. Most of the features of the Palm that you will need are shown below. The graffiti pad allows you to “write” using special characters that the Palm recognizes. This will be demonstrated to you in small-group sessions.

Stylus (behind)

Home Keyboard Power

Graffiti Pad (you can “write” here)

Children’s Self Perceptions

Page 60 of 75

Pr ocedur e for Completing a WBOS Session The best way to familiarize yourself with coding is by practicing with a PALM. You will have an opportunity to do that later. This overview will get you oriented and provide a reference that you can use later. Please note the screens on the Palms may look slightly different from the reproductions below. The balloon callouts will not appear on your PALM.

Step 1: Starting a session. On your PALM, click the “W” icon to launch the WBOS program. Page 1 shows the beginning of a session. It shows the day the Palm is prepared to code. The Palm can only be used for the “sign up” date indicated. Please check to make sure the date is correct. On page 2, enter your user name and password, as shown below at left. The “graffiti” pad at the bottom of the screen also allows you to write your password more quickly. This will be demonstrated in small group sessions. Page 3 asks you to start a new activity or resume an activity that is already started. In the example shown, Think City coding has already been started; if you were in that activity, you could resume coding for it. To start coding a new activity, select “New” at the bottom. You will proceed to the next page.

Page 4 shows the activities you (the coder, or “Adam— Otters (A)” in this example) are expected to attend for this day. Simply select the activity you are going to code. If the activity you need does not appear in the list, select “Other” at the bottom. If you do this, you will be prompted for details about this activity on the next screen.

Children’s Self Perceptions

Page 61 of 75

Page 4b appears if you click “Other” on the previous page. Often, you will not need this page because you selected a scheduled activity on the previous screen. If you did pick “Other”, you will need to specify several fields using drop-down menus: the Program, Slot (time period of the activity), the Activity Type, and the specific Activity. The options that will appear in certain menus will depend on previous options you have selected. For example, the activity schedule for Crossroads is different from the one for Mastery.

Once you have selected an activity, on Page 5 you will select the child who will be the focus of your coding. This is known as the “target” child. The (C) indicates the person is a child. This helps you keep children separate from adults, who will be labeled with an (A). If the child you need to code does not appear in the list, select “Add” to add that child. Once a child has been coded and is marked “done” on this screen, s/he cannot be coded again for this activity.

Step 2: Overall assessment of target child’s behaviors. Page 6 is an instruction page; it does not require any information. Note that you will rate how often the target child showed certain behaviors during the activity period using the 0-3 scale shown. On Page 7 the name of the target child appears at the top and a list of 8 behaviors appears. Select the box that corresponds to the rating that best describes the child’s behavior during this period. The Palm will require an answer to each item. Note that the code labels are short due to the small screen size. It is important to use the complete code definitions that appear later in this document.

Children’s Self Perceptions

Page 62 of 75

Step 3: Overall assessment of social events that happened to the target child. Page 8 is an instruction page similar to the one you saw earlier. Page 9 asks you to rate how often 4 PEER events and 4 ADULT events happened to the target child. As before, the target child is identified at the top of the page. Remember, your ratings should describe events that occurred during this period. The code labels are short because the screen is small. It is essential to use the complete definitions of the events provided later in this manual.

The Palm will require a rating for each of the events (0, 1, 2, or 3). It will also require that at least 2 events get a rating of “1” or higher. You will note that this list includes events that commonly occur in children’s interactions with peers and adults. If you did not observe at least 2 of these events, the Palm will instruct you to abort the coding session for the target child and find another staff who can code him or her. If this is not possible, you will be instructed to make a note in your Palm logbook. The research team will contact you about these aborted sessions. Once you have made your ratings of the events, there may be a slight pause in the Palm while it prepares material for the next page. Please wait if this happens. Do not click with the stylus or turn the screen off.

Step 4: Behavior chains. The Palm will choose which event to probe based on your previous answers. It will then ask you to rate how the child responded to that event. Here, we assume the Palm selected being “Teased / bossed” by a peer. Page 10 asks you to indicate which peer “teased / bossed” the target. Again, note that this is a brief code label; full definitions appear later in this manual. If this event happened more than once, focus on the event you most clearly remember. Typically, this is the most recent event—that is, the one closest to the time of coding. Here, assume you selected Joey as the child that “teased / bossed” Allen. Page 11 is an instruction page that explains the scale used to rate the target child’s reactions.

Children’s Self Perceptions

Page 63 of 75

Page 12 asks you to rate how the target child (Allen) reacted when Joey “teased / bossed” him. “React” means the most salient behavior(s) the child displayed in response within a minute or so after the event occurred. So, if Allen is “teased / bossed” by Joey, then Allen “hit / pushed / attacked” Joey within a minute or so after the tease occurred, then you would record that reaction by Allen. If Allen pushed Joey 30 minutes after the tease occurred, this should not be coded as a reaction (even if you believe Allen was “getting back” at the other child). Note that a child may react with more than one behavior to an event. Also note that “0” does not appear. If a reaction occurred, give it a rating of 1, 2, or 3. If it did not occur, nothing is checked. Page 13 asks you to rate how the other child (Joey) responded after the target (Allen) showed the behaviors described on the last page. The rules are

the same as before: A response must occur within a minute or so after the target behaviors described on the last page. Also note that the scale is the same as the one used on the last page.

Step 4: Behavior chains, continued. The remaining pages in this session repeat the 4 preceding pages. In this example, the Palm has selected an adult event (“gave instructions”). You will then identify the adult who most recently gave instructions to the target. Next, the instruction page will appear again. As before, you will be asked to rate how the target child responded when the counselor showed the behavior in question. Again, if a particular response did not occur, nothing is checked for that response.

Children’s Self Perceptions

Page 64 of 75

As the final step in the chain, rate how the counselor responded to the behaviors of the target child. The coding rules for counselors’ responses are the same as the rules for children’s responses.

After two behavior chains have been coded, the Palm will return you to Page 5. The child you just completed will be marked as “done”, and you will either select another child or quit the coding session. When you quit, data you have entered are automatically stored in the Palm until it is docked by the research team.

Children’s Self Perceptions

Page 65 of 75

Code Definitions This section provides code definitions and examples. The following general points about the codes should be noted: 1. Some codes are used more than once during the coding process (see above description of behavior chains). The codes are grouped below based on where they are used in the coding system (e.g., target child behaviors, social event ratings, peer or adult responses to child in behavior chains). 2. Some codes are defined separately for peers and adults even though they are related. For example, to preserve role distinctions between adults and children, “ask / told” is used for peers, and “gave instructions” is used for adults. 3. Some codes have two or more parts, divided by a slash (e.g., “tease / boss”, “warn / reprimand”). For these codes, if any part of the code occurs, then code the behavior as an instance of the code. As noted earlier in this manual, code labels are short due to the small screen size of the PALM. It is important to use the code using the full definition provided below. For example, “tease / boss” not only covers the forms of verbal aggression that appear in the code label itself (tease, boss), but also covers threats (see definition below). 4. A single “burst” of behavior may be coded with one or more codes, depending on the behavior. For example, a child might say, “If you don’t shut up you’ll regret it!” and then pushes the other child, all within a few seconds. This includes a threat (so code as “tease / boss”) and a push (so code as “hit / push / attack”).

Target child and peer behaviors used throughout the WBOS “ Talked prosocially.” This code covers a range of positive verbal behaviors for children. It refers to age-appropriate attempts to engage in conversation. This can include non-aversive statements, questions, providing information to others, requesting information, and making non-hostile comments about activities, events, or people. “Talked prosocially” emphasizes normal everyday conversing, without an attempt to achieve a particular behavioral response from the target. Note that “ask / tell” would be used when a child asks or tells someone to do something other than talk. Examples John makes a comment to another child about how many fish he caught. Jamal tries to explain to an adult why he was in the off-limits area. Susan asks another child if she could borrow some activity supplies. Devon says, “Can you tell me what you know about today’s activities?” (This would be coded as talked prosocially rather than ask / tell, as it is primarily conversational.) Non-examples Steve talks or mumbles to himself rather than another person. Crystal argues with another child about who will be first in line. (Code as “argued / quarreled”.) Devon says, “Can you hand me that ball?” (Code as “ask / tell”, as it emphasizes a specific action to be performed.)

“ Complied / cooperated.” Used when a child acts in accordance with expectations or appropriate requests made by peers or adults. This includes adherence to daily routine without prompting or in response to an instruction or reminder. Yielding or submissive behavior in response to peer bullying is NOT coded as comply/cooperate. Examples When a counselor asks Mika to get back in line, Mika says OK and gets back in line. Davon returns from waterfront and hangs up his towel in response to a reminder. At the end of art, Jake starts cleaning the table without being asked. Sasha hands the ball to another child when that child asks her for it. Non-examples John throws a rock at the cabin after Roger told him to do it in a threatening way. (Code John as “yielded / gave in”.) James told another kid to hang up his towel. (Code as “asked / told”.)

Children’s Self Perceptions

Page 66 of 75

“ Hit / pushed / attacked.” Used when a child harms or attempts to harm someone physically, or when a child attempts to physically control someone in a rough or hostile way. Clear attempts to inflict harm should be coded with this code, even if the attempt failed (e.g., the child throws an object at someone, but misses). Accidental physical contact, even if harm results, does not fit this code. Examples John pushes another child in line in a rough way. Susan elbows another child while she is being escorted out of the activity. Susan swings at another girl but misses. Non-examples John and another child are engaged in rough-and-tumble play, with no harm intended. Susan rushes out of the door and bumps into another child by accident. An adult restrains a child during an intervention. (Code as “physically controlled”.)

“ Teased / bossed.” Tease refers to verbal or non-verbal behavior that is intended to taunt, pester, provoke, or embarrass someone. Non-verbal teasing includes mocking or the use of provoking facial or hand gestures. Boss refers to a child commanding someone to do something in a coercive way. If a threat or implied threat occurs, code as boss even if the threat is not carried out. Direct instructions that are not coercive should be coded as “ask / tell”. Examples

John says, “You’re a stupid jerk!” Susan mimics another child’s voice in a sarcastic way. John says, “You’d better get off my bed or else!” Susan says, “Get out of my way! Non-examples In a friendly, clearly joking way, John says, “You goofball.” (Code as “talked prosocially.”) In a helpful way, John says, “That’s not how the lock works.” (Code as “talked prosocially.”) In a non-coercive way, Devon says, “Hand me the red bow next to you.” (Code as “asked / told”.)

“ Argued / quarreled.” Use when the child verbally resists instructions, requests, or statements by others in an oppositional manner. To be coded as such, the behavior must be contrary and quarrelsome. Expression of dissatisfaction with something in an age-appropriate, non-oppositional way should not be coded with this code. Examples When clean-up time is announced, John says, “I’m not gonna clean up my space.” When given a time out, Susan says, “I’m not taking any stinking time out.” When a child asks for the ball, John says, “I’m not sharing anything with you.” Non-examples John says, “If you don’t leave me alone, I’m gonna get you!” (Code as “teased / bossed”.) When told to take a time out, Susan throws something at the adult. (Code as “hit / pushed / attacked”.) When told to swim, John says, “I can’t because my ear is infected.” (Code as “talked prosocially”.)

“ Whined / complained.” This code is used when the child uses a babyish, plaintive, or high-pitched voice that is age-inappropriate or immature. The code is also used when the child fusses, gripes, or expresses dissatisfaction, discontent, or resentment in a tiresome way. Examples When swim time is announced, John says plaintively, “I don’t waaaaaaaaaaaaaana go to swim.” At art, Susan says in a babyish voice, “We never have any good art supplies.” After a minor scratch, John fusses in an exaggerated way about needing to go see the nurse. Non-examples John argues, but not in a babyish way, with an adult about an “unfair” rule. (Code as “argued / quarreled”.) Susan talks about how sad she is that she won’t have visitors on visitors’ day. (Code as “talked prosocially”.)

“ Looked sad / cried.” Looked sad occurs when a child shows non-verbal expressions of unhappiness or grief, appears downcast, somber, or depressed. Cried refers to shedding tears out of sadness, sorrow, or being hurt, either psychologically or physically. Examples Susan begins to cry when she learns that she won’t have visitors on visitors’ day. John’s shoulders slump and he looks down dejectedly when no one picks him for their athletic games. Non-examples Susan cries tears of joy when she learns that her mother will arrive on visitors day. You think a child may be feeling sad about a letter he got from home, even though there is no immediate expression of sadness.

Children’s Self-Perceptions 68 “ Withdrew / isolated.” Used when a child avoids contact with others, physically isolates self from others, or pulls back when someone attempts to interact with him/her. The pulling back from social contact is more important than the physical distance per se. A child can withdraw from interactions even when peers are playing nearby. Examples John sits on the periphery of the group during an activity and does nothing. Susan sits on her bed during free time and stares at the ceiling of the cabin. When another child asks John to play a game, he returns to reading his comic book. When praised by an adult, Susan says nothing and turns away. Non-examples John sits on his bunk during quiet time, as instructed by adults. (Code as “comply / cooperate”.) Susan works on her clay pot by herself while other children do the same.

Additional codes used for social event ratings and behavior chains “ Asked / told.” Used when a child makes a request or tells a peer to do something in a noncoercive way. It can refer to a request for specific help or specific information from a peer, or a more controlling action such as telling another child to follow the rules. A peer asking another child to do something in a coercive way would be coded as peer tease / boss (see above). (See “gave instructions” for the adult version of this code.) See also “talked prosocially” for distinctions from that code. “Asked / told” emphasizes a request for a specific action; “talked prosocially” emphasizes normal conversation, without a request for a specific action. Examples Wendy says to a peer, “Can you get off my bunk?” Gabriella says to another child, “You better hurry up or we will be late for Think City.” A child asks, “How do I get to Ames Hall?” (A specific action or type of information is requested.) Non-examples Maria demands, “You’d better get off my bed or else!” (Code as peer “teased/bossed”.) Bobby shouts, “Get out of my way!” (Code as peer “teased/bossed”.) Wanda asks another peer, “Do you like my art project?” (Code as peer “talked prosocially”.)

“ Gave instructions.” Use when an adult tells an individual child to do something, either immediately or in the very near future. This code does not refer to instructions about how to do something (e.g., how to bait a hook) or general suggestions with no specific action (e.g., telling a child to work hard on having a good summer). Instructions are distinct from warnings (see below). Imposing a time out is handled by a separate code (see disciplined/punished). Examples An adult says to a specific child, “Please go sit on your bunk.” An adult says, “Joey, let Nathan have a turn now.” An adult requests, “Sasha, tell the group the best thing that you did today in activities.” Non-examples “If you don’t stop talking, you’ll get a time out.” (Code as “warned/reprimanded”.) “Go take a time out on the rock.” (Code as “disciplined/punished”.) “Think about your goals, Danny, so you can make good choices today.” (Code as “talked prosocially”.)

“ [Adult] Talked Prosocially.” This code covers a range of positive verbal behavior directed at the target child by adults. Prosocial talk refers to positive, constructive, non-aversive statements or questions,

Children’s Self-Perceptions 69 including: providing information to others, requesting information from others, making comments about activities, events, or people, or otherwise attempting to engage in conversation with children. Examples An adult tells a child he did a good job in cleaning up his bunk. An adult asks a child if he has ever used a kayak before. An adult describes the daily schedule to a child. Non-examples An adult instructs a child to clean up his bunk. (Code as “gave instructions”.) An adult tells the group what a good job they did in group therapy today. (Not directed to an individual child.)

“ Warned / reprimanded.” Warn is used when an adult states to a particular child that some negative consequence will follow if the child does something. Warn may be used if the contingency is implicit, provided it is still clear to the child. Reprimand refers to an adult scolding, admonishing, or telling the child that some behavior is inappropriate. Examples “If you don’t do your chore, you will miss first activity.” “You’d better be out of the water and on the bench by 10.” (A consequence is implied, but clear at swimtime.) “It’s not OK to tease, Joey.” Non-examples “Gerald, you need to clean up now and get ready for the next activity.” (Code as “gave instructions”.) “The Walkers are clearly showing me that they aren’t ready to go to the WIT social tonight.” (Not directed to an individual child.)

“ Disciplined / punished.” This code refers to adults giving a consequence to a particular child for some behavior, including loss of privileges, a time out, or time away from the group. The consequence must be clear and specific and declared at the time. It does not include physical restraint (see “physically controlled”). However, if an adult imposes a consequence during or after a restraint, that would be coded as “disciplined/punished.” Examples An adult says, “Janiya, you need to sit on the time out rock for 1 minute.” An adult tells the child, “Because you could not stop arguing, you must give up your turn.” An adult says to Marcus, “Because you were 10 minutes late, you’ll have to sit out for 10 minutes of our next activity.” Non-examples An adult restrains a child who is physically attacking another child. (Code as “physically controlled”) An adult announces to the group, “Because the group was so rowdy, we will all have a minute of quiet time.” (Not directed to an individual, not coded)

Codes used exclusively in behavior chains “ Ignored / no response.” Used to indicate that an interactant (peer or adult) intentionally disregarded, did not acknowledge, or did not respond to a behavior of the target child. Note that ignore does not refer to situations in which it is unclear whether the individual heard or saw the child do the behavior that is being “ignored”. Examples An adult purposefully did not acknowledge Jerry when he argued about cleaning up.

Children’s Self-Perceptions 70 A peer did not say or do anything when Denisha asked to have the crayons. Non-examples An adult says, “I can’t hear you when you talk to me in that whiny voice.” (Code as “warn / reprimand”.) A peer smiles but says nothing when Vivian helps her tie her shoe. (Code peer as “showing positive emotion”.)

“ Showed positive emotion.” Use to indicate a nonverbal affective response showing warmth, affection, connection, enjoyment, pleasure, etc. The positive emotion must be directed at the target child being coded. Examples A peer responds to Greg’s helpful behavior by smiling. An adult responds to Iliana’s cooperative behavior during clean-up by giving her a high-five. A peer responds to Kyle’s cooperation by doing a celebration dance. Non-examples A peer laughs when Henry gives in to his teasing and starts to cry. (Code peer as “teased/bossed”.) An adult says to the Falcons, “I am so happy to report that everyone has earned the incentive.” (Not directed to an individual child.)

“ Yielded/gave in” This code should be used to indicate a submissive response by a peer to a controlling behavior by a child. It should also be used to indicate that an adult or peer has abandoned a recent request or intention. This abandonment is usually the result of noncooperative behavior by the target child. This could include abandoning attempts to redirect a child or not following through on stated consequences (not recommended practice). Examples Sammy gives the ball to a peer who was threatening to fight him for it. Tiana stops trying to defend herself to a peer falsely accusing her of stealing and gives her the candy. An adult who warned Ursula to get down notices her still on the rock but doesn’t do anything about it. Exasperated, an adult gives up on her attempts to redirect Vinny’s rude behavior at lunch. Non-examples An adult ignores Jerry when the child complains about clean up time. (Code adult as “ignored/no response”.) Denisha lets a peer go first at archery because she didn’t get a turn last time. (Code peer “comply / cooperate”.)

“ Physically controlled” This code should be used to indicate adult use of physical intervention or restraint. This includes any physical intervention with a child designed to de-escalate, avoid a potential crisis, or keep a child safe. It can include an adult physically guiding a child away from an escalating argument with a peer, blocking a child from someone with his/her body, basket-holding, or the use of other holding techniques. The physical control must be direct and be targeted at a specific child. Examples An adult takes Jamiel by the arm and walks him away from a heated argument with a peer. An adult uses a basket-hold on Kevin after the child hits someone to help him calm down and control his body. Non-examples After two children start fighting, an adult moves the rest of the group along so others don’t get involved.

Children’s Self-Perceptions 71 An adult says, “If you don’t put the rock down I will have to hold your hands.” (Code as “warned/disciplined”.)

Children’s Self-Perceptions 72

Appendix D. Justification for Aggregation over Peer and Adult Situations; Correlation Between Peer and Adult Perceptions Data Behavior

Time 1

Time 2

Time 1 & 2

Aggression

.64**

.52**

.66**

Withdrawal

.70**

.73**

.73**

Prosocial

.68**

.55**

.79**

Mean Aggregate

0.67

0.60

0.73

Note. ** p < .01

Children’s Self-Perceptions 73

Appendix E. Temporal Stability of Self-perceptions and Behavioral Observations, by Age and Gender Perceptions

Context-specific Observations

Global Observations

Young

Young

Old

Old

Young

Young

Old

Old

Young

Young

Old

Old

Girls

Boys

Girls

Boys

Girls

Boys

Girls

Boys

Girls

Boys

Girls

Boys

Aggression

.90**

.40*

.81**

.72**

.73**

.34*

.72**

.23

.74**

.60**

.70**

.59**

Withdrawal

.70**

.60**

.61**

.61**

.28

.51**

.44*

.40**

-.044

.70**

.73**

.64**

Prosocial

-.08

.55**

.63**

.71**

.57

-.02

.10

.00

.31

.71**

.48*

.67**

0.51

0.52

0.68

0.68

0.53

0.28

0.42

0.21

0.34

0.67

0.64

0.63

Behavior

Mean Aggregate

Note. * p < .05, ** p < .01

Children’s Self-Perceptions 74

Appendix F. Concurrent Validity of Children’s Self-perceptions and Context-specific Observations of Behavior, by Age and Gender Measure

Time 1

Time 2

Young Young Old

Old

Time 1&2

Young Young Old

Old

Young Young Old

Old

Girls

Boys

Girls Boys Girls

Boys

Girls Boys Girls

Boys

Girls Boys

Aggression

.26

.34*

.83** .29* .34

.32

.47*

.38*

.69** .37**

Withdrawal

.20

.27

.52*

.07

.52**

.55** .28* .26

.46**

.66** .29*

Prosocial

.14

.21

.17

-.03 -.14

-.127

-.07

.28* -.16

.11

.09

.08

0.20

0.27

0.51

0.11 0.18

0.24

0.32

0.27 0.16

0.32

0.48

0.25

.35

.26

.37

Mean Aggregate

Note. * p < .05, ** p < .01.

Children’s Self-Perceptions 75

Appendix G. Concurrent Validity of Children’s Self-perceptions and Global Observations of Behavior, by Age and Gender Measure

Time 1

Time 2

Young Young Old

Old

Time 1&2

Young Young Old

Old

Young Young Old

Old

Girls

Boys

Girls Boys Girls

Boys

Girls Boys Girls

Boys

Girls Boys

Aggression

.39

.55**

.74** .50** .43

.12

.55** .33* .46

.39*

.74** .44**

Withdrawal

.09

-.01

-.21

-.08

.25

.30

.18

-.16

.10

.21

.03

-.15

Prosocial

.40

.25

.15

.15

-.14

.16

-.20

.02

.09

.29

.03

.10

0.29

0.26

0.23

0.19

0.18

0.19

0.18

0.06 0.22

0.30

0.27

0.13

Mean Aggregate

Note. * p < .05, ** p < .01.