Abstract
Using data from a subsample of 913 study children and their friends who participated in the NICHD Study of Early Child Care and Youth Development, the interactive contributions of child-reported attribution biases and teacher-reported child emotional intensity (EI) at Grade 4 (M = 9.9 years) to observed child–friend interaction at Grade 6 (M = 11.9 years) were examined. Study children's hostile attribution bias, combined with high EI, predicted more negative child–friend interaction. In contrast, benign attribution bias, combined with high EI, predicted more positive child–friend interaction. The findings are discussed in light of the “fuel” interpretation of EI, in which high-intensity emotions may motivate children to act on their cognitive biases for better or for worse.
| Original language | English (US) |
|---|---|
| Pages (from-to) | e114-e131 |
| Journal | Child development |
| Volume | 90 |
| Issue number | 1 |
| DOIs | |
| State | Published - Jan 1 2019 |
ASJC Scopus subject areas
- Pediatrics, Perinatology, and Child Health
- Education
- Developmental and Educational Psychology
Fingerprint
Dive into the research topics of 'Interactive Contributions of Attribution Biases and Emotional Intensity to Child–Friend Interaction Quality During Preadolescence'. Together they form a unique fingerprint.Cite this
- APA
- Standard
- Harvard
- Vancouver
- Author
- BIBTEX
- RIS
In: Child development, Vol. 90, No. 1, 01.01.2019, p. e114-e131.
Research output: Contribution to journal › Article › peer-review
}
TY - JOUR
T1 - Interactive Contributions of Attribution Biases and Emotional Intensity to Child–Friend Interaction Quality During Preadolescence
AU - Chen, Xi
AU - McElwain, Nancy L.
AU - Lansford, Jennifer E.
N1 - The NICHD Study of Early Child Care was directed by a Steering Committee and supported by NICHD through a cooperative agreement (U10) that calls for a scientific collaboration between the grantees and the NICHD staff. We wish to express our appreciation to the principal investigators, site coordinators, and participants of the NICHD Study of Early Child Care and Youth Development. We are also grateful to Dr. Kristen Bub for her advice on the statistical analyses. Xi Chen was supported by a University of Illinois Graduate College Distinguished Fellowship while working on this manuscript. During a Grade 4 laboratory visit, the study child and friend independently completed a questionnaire designed to assess attribution biases (Crick,). Children were presented with provocation situations, in which the intent of the provocateur was ambiguous, and were asked to imagine the provocation had happened to them. Three situations depict instrumental provocation focusing on acts of potential physical aggression (e.g., your radio is broken by a peer), and two situations depict relational provocation focusing on potential rejection (e.g., when some students pass by, they look at you, chat with each other, and laugh). Children answered two forced-choice questions for each situation. First, children chose one of four reasons for the provocation. Two reasons reflected hostile intent (e.g., the kid was mad at you), and two reasons reflected benign intent (e.g., it was an accident). Second, children were asked whether the provocateur was trying to be mean (hostile intent) or not (benign intent). For all questions, responses indicating hostile intent were coded as 1, and responses indicating benign intent were coded as 0. We note that, for the purposes of the NICHD SECCYD, this measure was truncated from its original version of 10 situations (five instrumental and five relational; see Crick,) to five situations (three instrumental and two relational), and we computed an overall attribution bias score across responses to the five situations (10 items; ??=.78 for study children; ??=.73 for friends; see Laible et?al., for the same approach). Possible scores ranged from 0 to 1, with higher scores indicating a greater tendency to attribute hostile intent and lower scores indicating a greater tendency to attribute benign intent. In previous studies using this measure, a more hostile attribution bias was associated with aggression (Crick et?al.,), and a more benign bias predicted prosocial behavior (Laible et?al.,) among third- to sixth-grade children. During a Grade 4 laboratory visit, the study child and friend independently completed a questionnaire designed to assess attribution biases (Crick,). Children were presented with provocation situations, in which the intent of the provocateur was ambiguous, and were asked to imagine the provocation had happened to them. Three situations depict instrumental provocation focusing on acts of potential physical aggression (e.g., your radio is broken by a peer), and two situations depict relational provocation focusing on potential rejection (e.g., when some students pass by, they look at you, chat with each other, and laugh). Children answered two forced-choice questions for each situation. First, children chose one of four reasons for the provocation. Two reasons reflected hostile intent (e.g., the kid was mad at you), and two reasons reflected benign intent (e.g., it was an accident). Second, children were asked whether the provocateur was trying to be mean (hostile intent) or not (benign intent). For all questions, responses indicating hostile intent were coded as 1, and responses indicating benign intent were coded as 0. We note that, for the purposes of the NICHD SECCYD, this measure was truncated from its original version of 10 situations (five instrumental and five relational; see Crick,) to five situations (three instrumental and two relational), and we computed an overall attribution bias score across responses to the five situations (10 items; ??=.78 for study children; ??=.73 for friends; see Laible et?al., for the same approach). Possible scores ranged from 0 to 1, with higher scores indicating a greater tendency to attribute hostile intent and lower scores indicating a greater tendency to attribute benign intent. In previous studies using this measure, a more hostile attribution bias was associated with aggression (Crick et?al.,), and a more benign bias predicted prosocial behavior (Laible et?al.,) among third- to sixth-grade children. At Grade 4, teachers of the study child and the friend, respectively, completed a questionnaire that was developed from similar measures of EI used in previous studies (Eisenberg et?al.,; Larsen & Diener,). Teachers rated the child's frequency of experiencing and expressing strong emotions as described by 10 items (e.g., ?When this child feels an emotion, either positive or negative, he or she feels it strongly?) on a 5-point scale, ranging from 1 (never) to 5 (always). EI scores were computed as the sum of teacher ratings on the 10 items, with five items reverse scored so that higher values indicated higher EI (??=.83 for study children, ??=.78 for friends). Scores had a possible range from 10 to 50. In previous studies using similar measures, higher EI predicted lower social competence in middle childhood (e.g., Eisenberg et?al.,). At each time point, the study child and friend were observed in a series of interactive sessions in a laboratory setting. At Grade 4, child?friend dyads participated in four sessions in the following order: (a) free play (10?min); (b) plan a party task (10?min), in which the dyad planned a birthday party for both of them; (c) snack (7?min), in which the dyad had snacks together (the amount and type of snacks were designed to elicit discussion or negotiation); and (d) pick-up sticks game (8?min), in which the winner received a prize. At Grade 6, the child?friend dyad participated in seven sessions in the following order: (a) snack (7?min; same as Grade 4 snack); (b) Jenga game (8?min), in which the winner received a prize; (c) pictionary game (7?min), in which the dyad played as a team to win points; (d) plan a kids-only vacation task (7?min), in which the dyad planned a vacation together; (e) who wants to be a millionaire game (8?min), in which the dyad played as a team to win points; (f) hand game (5?min), in which the dyad played the rock, paper, scissors game, and the child who won the most times received a prize; and (g) choosing prizes (5?min), in which the dyad used the points that they had jointly earned in exchange for prizes. The interactions were videotaped through a one-way mirror and coded based on previous coding schemes (e.g., Allhusen, Flyr, Parke, & Clarke-Stewart,) and research assessing friendship quality via observational methods (e.g., Dishion, Patterson, & Griesler,). The coders rated the individual behaviors of the study child and the interaction quality of the child?friend dyad across the play sessions at each time point on 5-point scales (1?=?not at all characteristic to 5?=?highly characteristic). Ratings on four behaviors were used in the current report: (a) the study child's negative behavior (e.g., intrusiveness, hostility, verbal, and physical aggression), (b) the dyad's negative interaction (e.g., conflict, disagreement, aversive interchanges), (c) the study child's positive behavior (e.g., positive engagement with friend, accommodating the friend's wishes), and (d) the dyad's overall friendship quality (e.g., reciprocity, enjoyment, agreement). Other ratings of child?friend interaction were not used in this report because they were not available at both time points. Interobserver reliability was assessed for 11% of the dyads (n?=?100 and 103, Grades 4 and 6, respectively). Although double coding 20% of cases is the typical ?rule of thumb,? reliability samples based on smaller proportions of cases are appropriate when samples are large because the absolute number (vs. proportion) of cases is more relevant to the representativeness of the reliability subsample (see Yoder & Symons,). Reliability coefficients were computed via intraclass correlations and were moderate to high for the study child's negative (.94 and.91, Grades 4 and 6, respectively) and positive (.87 and.92) behavior, as well as the dyad's negative interaction (.89 and.91) and overall friendship quality (.94 and.91). Participants were a subsample of 913 study children drawn from the full sample of 1,364 children participating in the NICHD SECCYD. Study children were recruited from hospitals at 10 sites across the United States based on several criteria (e.g., children who were hospitalized for more than 7?days following birth or had obvious disabilities were excluded; mothers who had a known or acknowledged substance abuse problem were excluded; see NICHD Early Child Care Research Network, for further details) and followed from 1?month of age to 15?years of age. In this report, we examined data collected when study children were in Grades 4 and 6 because observational assessments of child?friend interaction during middle childhood and early adolescence were available only at these two time points. Although study children's attribution biases and EI were also measured at Grades 3 and 5, we only used the Grade 4 measures to rigorously test attribution biases and EI as predictors of later friendship quality controlling for earlier friendship quality measured at the same time point (i.e., Grade 4). The 913 study children (49.6% boys) who had complete data on observed interaction with a friend at Grade 6 were included in our subsample. The study children averaged 9.9 (SD?=?0.30) and 11.9 (SD?=?0.34) years of age at the Grade 4 and Grade 6 time points, respectively. Among the study children, 77.8% were non-Hispanic White, 11.7% were Black, 5.8% were Hispanic, and 4.7% were another ethnicity or more than one ethnicity. Mothers averaged 14.46 (SD?=?2.40) years of education, as reported by mothers when study children were 1?month of age. The average family income-to-needs ratio (averaged across data collected at 6, 15, 24, 36, and 54?months) was 3.67 (SD?=?2.71). At each time point, a friend was identified by the study child's mother using several criteria (e.g., the friend was not the study child's sibling; the friend was within 2?years of age of the study child; and preferably, the friend was the same gender as the study child). Once these criteria were met, preference was given to the study child's closest friend. According to study children's report, 77.3% and 74.5% were ?best friends? with the identified friend (Grade 4 and Grade 6, respectively), 18.9% and 23.7% were ?close friends,? and 3.5% and 1.4% were ?ordinary friends.? In most cases, the friend was the same gender as the study child (94% at Grade 4, 92.2% at Grade 6), and we note that when we recomputed our main model tests with data from only same-gender friends, results were consistent with the results reported below and in Table?2. For 320 study children, the same friend participated at both visits. Compared with study children excluded from the current report due to missing data on Grade 6 outcomes (n?=?451), study children included were more likely to be girls, ?2(1, N?=?1,364)?=?4.74, p?=.030, and had mothers with higher levels of education, t(1,361)?4.79, p?<.001. No differences, however, emerged on study child ethnicity, family income, or the Grade 4 measures (i.e., attribution biases, EI, and child?friend interaction). Participants were a subsample of 913 study children drawn from the full sample of 1,364 children participating in the NICHD SECCYD. Study children were recruited from hospitals at 10 sites across the United States based on several criteria (e.g., children who were hospitalized for more than 7?days following birth or had obvious disabilities were excluded; mothers who had a known or acknowledged substance abuse problem were excluded; see NICHD Early Child Care Research Network, for further details) and followed from 1?month of age to 15?years of age. In this report, we examined data collected when study children were in Grades 4 and 6 because observational assessments of child?friend interaction during middle childhood and early adolescence were available only at these two time points. Although study children's attribution biases and EI were also measured at Grades 3 and 5, we only used the Grade 4 measures to rigorously test attribution biases and EI as predictors of later friendship quality controlling for earlier friendship quality measured at the same time point (i.e., Grade 4). The 913 study children (49.6% boys) who had complete data on observed interaction with a friend at Grade 6 were included in our subsample. The study children averaged 9.9 (SD?=?0.30) and 11.9 (SD?=?0.34) years of age at the Grade 4 and Grade 6 time points, respectively. Among the study children, 77.8% were non-Hispanic White, 11.7% were Black, 5.8% were Hispanic, and 4.7% were another ethnicity or more than one ethnicity. Mothers averaged 14.46 (SD?=?2.40) years of education, as reported by mothers when study children were 1?month of age. The average family income-to-needs ratio (averaged across data collected at 6, 15, 24, 36, and 54?months) was 3.67 (SD?=?2.71). At each time point, a friend was identified by the study child's mother using several criteria (e.g., the friend was not the study child's sibling; the friend was within 2?years of age of the study child; and preferably, the friend was the same gender as the study child). Once these criteria were met, preference was given to the study child's closest friend. According to study children's report, 77.3% and 74.5% were ?best friends? with the identified friend (Grade 4 and Grade 6, respectively), 18.9% and 23.7% were ?close friends,? and 3.5% and 1.4% were ?ordinary friends.? In most cases, the friend was the same gender as the study child (94% at Grade 4, 92.2% at Grade 6), and we note that when we recomputed our main model tests with data from only same-gender friends, results were consistent with the results reported below and in Table?2. For 320 study children, the same friend participated at both visits. Compared with study children excluded from the current report due to missing data on Grade 6 outcomes (n?=?451), study children included were more likely to be girls, ?2(1, N?=?1,364)?=?4.74, p?=.030, and had mothers with higher levels of education, t(1,361)?4.79, p?<.001. No differences, however, emerged on study child ethnicity, family income, or the Grade 4 measures (i.e., attribution biases, EI, and child?friend interaction). At Grade 4, study children and friends independently completed questionnaires to assess attribution biases, and teachers completed questionnaires to assess children's EI. At each time point, child?friend interaction during a series of interactive tasks in a laboratory setting was videotaped. All videotapes were shipped to a central site for coding, and coders were blind to participant characteristics and other study results. During a Grade 4 laboratory visit, the study child and friend independently completed a questionnaire designed to assess attribution biases (Crick,). Children were presented with provocation situations, in which the intent of the provocateur was ambiguous, and were asked to imagine the provocation had happened to them. Three situations depict instrumental provocation focusing on acts of potential physical aggression (e.g., your radio is broken by a peer), and two situations depict relational provocation focusing on potential rejection (e.g., when some students pass by, they look at you, chat with each other, and laugh). Children answered two forced-choice questions for each situation. First, children chose one of four reasons for the provocation. Two reasons reflected hostile intent (e.g., the kid was mad at you), and two reasons reflected benign intent (e.g., it was an accident). Second, children were asked whether the provocateur was trying to be mean (hostile intent) or not (benign intent). For all questions, responses indicating hostile intent were coded as 1, and responses indicating benign intent were coded as 0. We note that, for the purposes of the NICHD SECCYD, this measure was truncated from its original version of 10 situations (five instrumental and five relational; see Crick,) to five situations (three instrumental and two relational), and we computed an overall attribution bias score across responses to the five situations (10 items; ??=.78 for study children; ??=.73 for friends; see Laible et?al., for the same approach). Possible scores ranged from 0 to 1, with higher scores indicating a greater tendency to attribute hostile intent and lower scores indicating a greater tendency to attribute benign intent. In previous studies using this measure, a more hostile attribution bias was associated with aggression (Crick et?al.,), and a more benign bias predicted prosocial behavior (Laible et?al.,) among third- to sixth-grade children. At Grade 4, teachers of the study child and the friend, respectively, completed a questionnaire that was developed from similar measures of EI used in previous studies (Eisenberg et?al.,; Larsen & Diener,). Teachers rated the child's frequency of experiencing and expressing strong emotions as described by 10 items (e.g., ?When this child feels an emotion, either positive or negative, he or she feels it strongly?) on a 5-point scale, ranging from 1 (never) to 5 (always). EI scores were computed as the sum of teacher ratings on the 10 items, with five items reverse scored so that higher values indicated higher EI (??=.83 for study children, ??=.78 for friends). Scores had a possible range from 10 to 50. In previous studies using similar measures, higher EI predicted lower social competence in middle childhood (e.g., Eisenberg et?al.,). At each time point, the study child and friend were observed in a series of interactive sessions in a laboratory setting. At Grade 4, child?friend dyads participated in four sessions in the following order: (a) free play (10?min); (b) plan a party task (10?min), in which the dyad planned a birthday party for both of them; (c) snack (7?min), in which the dyad had snacks together (the amount and type of snacks were designed to elicit discussion or negotiation); and (d) pick-up sticks game (8?min), in which the winner received a prize. At Grade 6, the child?friend dyad participated in seven sessions in the following order: (a) snack (7?min; same as Grade 4 snack); (b) Jenga game (8?min), in which the winner received a prize; (c) pictionary game (7?min), in which the dyad played as a team to win points; (d) plan a kids-only vacation task (7?min), in which the dyad planned a vacation together; (e) who wants to be a millionaire game (8?min), in which the dyad played as a team to win points; (f) hand game (5?min), in which the dyad played the rock, paper, scissors game, and the child who won the most times received a prize; and (g) choosing prizes (5?min), in which the dyad used the points that they had jointly earned in exchange for prizes. The interactions were videotaped through a one-way mirror and coded based on previous coding schemes (e.g., Allhusen, Flyr, Parke, & Clarke-Stewart,) and research assessing friendship quality via observational methods (e.g., Dishion, Patterson, & Griesler,). The coders rated the individual behaviors of the study child and the interaction quality of the child?friend dyad across the play sessions at each time point on 5-point scales (1?=?not at all characteristic to 5?=?highly characteristic). Ratings on four behaviors were used in the current report: (a) the study child's negative behavior (e.g., intrusiveness, hostility, verbal, and physical aggression), (b) the dyad's negative interaction (e.g., conflict, disagreement, aversive interchanges), (c) the study child's positive behavior (e.g., positive engagement with friend, accommodating the friend's wishes), and (d) the dyad's overall friendship quality (e.g., reciprocity, enjoyment, agreement). Other ratings of child?friend interaction were not used in this report because they were not available at both time points. Interobserver reliability was assessed for 11% of the dyads (n?=?100 and 103, Grades 4 and 6, respectively). Although double coding 20% of cases is the typical ?rule of thumb,? reliability samples based on smaller proportions of cases are appropriate when samples are large because the absolute number (vs. proportion) of cases is more relevant to the representativeness of the reliability subsample (see Yoder & Symons,). Reliability coefficients were computed via intraclass correlations and were moderate to high for the study child's negative (.94 and.91, Grades 4 and 6, respectively) and positive (.87 and.92) behavior, as well as the dyad's negative interaction (.89 and.91) and overall friendship quality (.94 and.91). Using Mplus 7.3 (Muth?n & Muth?n, 1998?2012), we fitted a structural equation?model to assess whether study children's attribution biases, EI, and the Attribution Bias???EI interaction at Grade 4 predicted child?friend interaction at Grade 6, above and beyond child?friend interaction at Grade 4. At both time points, ratings of study children's negative behavior and dyads? negative interaction were indicators of a latent variable of negative child?friend interaction, and ratings of study children's positive behavior and dyads? overall friendship quality were indicators of a latent variable of positive child?friend interaction. We examined both study children's behavior and dyadic interaction ratings as indicators of the latent factors because (a) study children's behavior toward the friend and dyadic interaction quality, which both underlie friendship quality, were closely related (rs ranged from.52 to.77, see Table), and (b) a latent factor based on the two related scores is likely to be a more robust and reliable measure of child?friend interaction quality than using a single score. The covariances between the error terms of the latent negative and positive interaction variables at Grade 6 were estimated, as were covariances among the predictors at Grade 4. Bivariate correlations are reported above the diagonal, and partial correlations controlling for study children's gender and ethnicity and maternal years of education are reported below the diagonal. SC?=?study child; DY?=?dyad; G4?=?Grade 4; G6?=?Grade 6. *p?<.05. **p?<.01. To reduce collinearity between the main effects and the interaction term, we used centered scores (raw score minus the mean) of attribution bias and EI in all model tests. Because we used centered scores, the main effect of attribution bias reflected its relation with the outcome when EI was at the mean (i.e., 0), and the main effect of EI reflected its relation with the outcome when attribution bias was at the mean (i.e., 0), even when the Attribution Bias???EI interaction was included in the model. Testing a ?main effects only? model (i.e., without the Attribution Bias???EI term) yielded main effect parameters identical to those reported below and in Table. G4?=?Grade 4; G6?=?Grade 6. EI?=?emotional intensity. *p?<.05. **p?<.01. ***p?<.001. In probing significant Attribution Bias???EI interactions, we plotted the association between study children's attribution bias and negative (or positive) child?friend interaction at high (1?SD above M) and low (1?SD below M) levels of EI, and we tested the simple slopes. Furthermore, to differentiate between the ?disruptor? and ?fuel? interpretations, we conducted complementary simple slope analyses for the associations between EI and child?friend interactions when attribution bias was more hostile versus benign. The ?disruptor? interpretation would be supported if the simple slope analyses together indicated that (a) a more hostile attribution bias predicted more negative interaction (and less positive interaction) only when EI was high, and (b) higher EI predicted more negative interaction and less positive interaction only when children held a more hostile attribution bias. In contrast, the ?fuel? interpretation would be supported if the complementary simple slope analysis (i.e., ?b? above) indicated that higher EI predicted more negative interaction when children held a more hostile attribution bias, and more positive interaction when children held a more benign attribution bias. Among the 913 study children included in the current report, some data were missing on the predictors at Grade 4 (i.e., 3.4% on attribution biases, 14.9% on EI, and 10.8% on child?friend interaction; also see Table). We used full information maximum likelihood (FIML)?which offers less biased estimates compared with more traditional methods such as listwise deletion (Schafer & Graham,)?to handle these missing data. We did not include study children who had missing values on Grade 6 outcomes because including those cases does not provide additional information for model estimation (Acock,). Given that the chi-square test of model fit is almost always significant when sample sizes are large, we used the comparative fit index (CFI) and root mean square error of approximation (RMSEA) to assess model fit. CFI values of 0.90 and above indicate an adequate fit, and values of 0.95 and above indicate a good fit (Hu & Bentler,). RMSEA values <?0.08 indicate an adequate fit, and values <?0.05 indicate a good fit (Browne & Cudeck,). We also conducted four sets of follow-up analyses, including multigroup analyses to assess whether the hypothesized associations differed as a function of study child gender or friend status (i.e., whether same vs. different friends participated in Grade 4 and Grade 6 visits). Furthermore, for dyads in which the same friend participated at both time points (n?=?320), we controlled for the friend's Grade 4 attribution bias and EI in our main model. Finally, we created separate subscales of positive, negative, and general EI from the EI questionnaire completed by teachers and tested the hypothesized model with these subscale scores as predictors. To tease apart valence-specific EI, we controlled for positive EI when testing negative EI as a predictor and vice versa (see Supporting Information for more details). Using data from 913 focal children (referred to as ?study children?) who participated in the NICHD Study of Early Child Care and Youth Development (SECCYD), we addressed two objectives. First, we examined the extent to which study children's self-reported attribution biases at Grade 4 predicted the quality of observed interaction between the study children and a designated friend at Grade 6, above and beyond child?friend interaction quality observed at Grade 4. Second, we assessed study children's EI as reported by teachers at Grade 4 as a moderator of the associations between study children's attribution biases and child?friend interaction quality. Based on prior theoretical (e.g., Frijda,; Lemerise & Arsenio,; Mascolo & Fischer,) and empirical (e.g., Burgess et?al.,; Laible et?al.,; Orobio de Castro et?al.,; Runions & Keating,) work, we hypothesized that (a) study children's more hostile versus benign attribution bias would be associated with higher levels of negative child?friend interaction and lower levels of positive child?friend interaction, and (b) EI would moderate these associations, such that they would be stronger among study children who had higher levels of EI. In addition, we sought to test the ?disruptor? (e.g., Lemerise & Arsenio,) versus ?fuel? (e.g., Frijda,) interpretations to better characterize interactions between attribution biases and EI on child?friend interaction quality. We depict hypothetical results that would support the ?disruptor? and ?fuel? interpretations in Figure. As seen in Figures?a and c, the interaction patterns are the same for negative child?friend interaction: A more hostile attribution bias combined with high EI predicts more negative child?friend interaction. The interaction patterns diverge, however, when predicting positive child?friend interaction. In Figures?b and d, a more hostile attribution bias is associated with lower levels of positive interaction only when EI is high, yet the simple slopes at high versus low EI show different patterns in relation to each other. Evidence for the ?disruptor? interpretation (Figure?b) indicates that a more hostile attribution bias combined with high EI predicts less positive interaction, which suggests that high EI may disrupt positive interaction when children tend to attribute hostile intentions. In contrast, evidence for the ?fuel? interpretation (Figure?d) indicates that a more benign attribution bias combined with high EI predicts more positive interaction, which suggests that high EI may promote positive interaction when children tend to attribute benign intentions. We also conducted follow-up analyses to assess the robustness of our main results, including tests of study child gender and friendship status (i.e., whether the study child was observed interacting with the same friend vs. different friends at Grade 4 and Grade 6) as additional moderators. Furthermore, for dyads in which the same friend participated at both time points, we controlled for the friend's attribution bias and EI at Grade 4. Finally, although individuals who experience positive emotions intensely also tend to experience negative emotions intensely (Diener et?al.,; Rydell et?al.,), it is not well understood whether positive and negative EI play similar roles in children's social functioning (Eggum et?al.,; Murphy & Eisenberg,). We explored whether findings from our main model, which examined general EI, were consistent when positive and negative EI were tested as distinct predictors. The importance of friendship for adjustment is well established (see Furman & Rose,; Hartup,). In particular, the transition period from childhood to adolescence, approximately between 9 and 13?years of age, is marked by the need for interpersonal intimacy with a ?chum? or close friend (Sullivan,). During this period, children spend increasing time with peers (Larson & Richards,) and regard friends as increasingly important providers of support (Furman & Buhrmester,). Furthermore, friendships characterized by high levels of intimacy and support predict more optimal psychological and behavioral adjustment (e.g., Waldrip, Malcolm, & Jensen-Campbell,) and buffer against the negative effects of stressors, such as victimization (e.g., Hodges, Boivin, Vitaro, & Bukowski,), loneliness (Nangle, Erdley, Newman, Mason, & Carpenter,), and negative parenting (Lansford, Criss, Pettit, Dodge, & Bates,). Friendships, however, can also have negative features such as conflict, dominance, and rivalry, which are not merely the inverse of positive features (Berndt,), and such negative features predict preadolescents? poorer behavioral adjustment, lower school grades, and greater school disruption (e.g., Berndt & Keefe,; Burk & Laursen,). Therefore, understanding factors that contribute to both positive and negative features of friendships during the transition from?childhood to adolescence is of theoretical and practical significance. To this end, we investigated two intrapersonal factors?attribution biases and emotional intensity (EI)?that are important for children's peer relationships generally yet are less well understood as predictors of friendship quality specifically. Furthermore, despite long-standing interest in understanding the increasing coordination of affective and cognitive processes in children's socioemotional and moral development (see Calkins & Bell,; Malti,), and theoretical work that integrates emotion-related processes within a social information processing framework (Lemerise & Arsenio,), few studies have examined how attribution biases and emotional processes interact to predict children's social behavior toward peers (for exceptions, see Mathieson et?al.,; Runions & Keating,), and no existing study to our knowledge has examined such cognition?emotion interactions in predicting friendship quality. In addressing these gaps, we investigated the extent to which children's attribution biases predicted positive and negative child?friend interaction quality over a 2-year period from Grades 4 to 6 and assessed children's EI as a moderator of these associations. Because children show more affection in their relationships with friends versus nonfriends (Newcomb & Bagwell,) and because adolescents show marked increases in emotional reactivity to peer-related stimuli (see Somerville,), EI may play an especially important role in moderating the link between attribution biases and child?friend interaction quality during the transition to adolescence. According to Crick and Dodge (), children's social behaviors are guided by their social information processing steps?including encoding and interpreting social cues, clarifying goals, and identifying, selecting, and enacting behavioral responses?that occur rapidly and often outside of conscious awareness. Interpretation of social cues, in particular, has been examined in relation to children's general peer functioning. It is well established, for instance, that children and adolescents who interpret a peer's ambiguous provocation as intentionally harmful (e.g., ?He meant to break it.?) are more aggressive and more likely to be rejected by their peers (e.g., Crick, Grotpeter, & Bigbee,; Dodge & Coie,; Orobio de Castro, Veerman, Koops, Bosch, & Monshouwer,; Runions & Keating,). Paralleling this link between a hostile attribution bias and aggressive behavior, preadolescent children who interpret ambiguous peer provocations as benign (e.g., ?It was an accident.?) exhibit higher levels of prosocial behavior toward peers (Laible, McGinley, Carlo, Augustine, & Murphy,). Less is known about the extent to which preadolescents? attribution biases predict the quality of their interactions with friends specifically. Not surprisingly, children make fewer hostile attributions (Peets, Hodges, Kikas, & Salmivalli,) and more prosocial attributions (Burgess, Wojslawowicz, Rubin, Rose-Krasnor, & Booth-LaForce,) toward friends than toward neutral or unspecified peers. Nonetheless, because children's attribution biases toward a specific peer are related to their attribution biases toward peers in general (Hubbard, Dodge, Cillessen, Coie, & Schwartz,), it is likely that children's general tendency toward hostile (or benign) attributions will forecast the child's interaction quality with specific peer partners, such as friends. When confronted with a friend's provocation, children who hold a more hostile attribution bias may retaliate in outwardly aggressive ways as they do toward general peers. Yet, given that friendships are voluntary relationships based on mutual enjoyment, cooperation, and self-disclosure, hostile attribution bias may also operate in more subtle ways. Children may retaliate with more nuanced contention, such as rejecting the friend's offers, being sarcastic, or withdrawing their support from the friend. At the same time, children who hold a more benign attribution bias may bring these positive expectancies to their friendships and show more support and prosocial behavior toward friends. Furthermore, because friendships are dynamic, dyadic relationships, a child's heightened negativity or suppressed positive behavior?rooted, in part, in the child's tendency toward a hostile attribution bias?may elicit more negative behavior from his or her friend and, therefore, generate more negative and less positive child?friend exchanges. Of the few studies to investigate children's attribution biases as a predictor of friendship outcomes, findings have been consistent with the above propositions. Young adolescents who tended to attribute negative intent toward unspecified peers, as measured via hypothetical vignettes, reported more conflict with best friends (Bowker, Spencer, & Salvy,; Spencer, Bowker, Rubin, Booth-LaForce, & Laursen,). Similarly, preadolescents? more hostile attributions toward best friends concurrently predicted more self-reported conflict in those friendships (Spencer et?al.,) and lower levels of self-reported positive friendship characteristics (but for girls only; Dwyer et?al.,). These past studies, however, were limited by cross-sectional designs and measurement of both attribution biases and friendship quality using self-report instruments. We addressed these limitations in the?current study by employing a longitudinal design and observational measures of child?friend interaction. EI reflects the intensive characteristic of temperamental reactivity to emotion-provoking events (Rothbart & Derryberry,; Thomas, Chess, & Birch,), as well as the dimension of ?arousal? in Russell's () framework of core affect, which is orthogonal to the dimension of ?valence.? Notably, EI generalizes across specific valence or content, such that individuals who experience positive emotions intensely also tend to experience negative emotions intensely (Diener, Larsen, Levine, & Emmons,; Rydell, Berlin, & Bohlin,). General EI as a relatively stable trait is associated with feelings of arousal and daily mood variability, and has been linked to distinct cognitions when processing emotion-provoking stimuli among adults (see Larsen & Diener,). In middle childhood, greater general EI reported by parents and teachers was associated with children's lower scores on socially competent behaviors at school (Eisenberg et?al.,). Although prior research indicates that EI has direct implications for social functioning, EI may also moderate associations between attribution biases and peer outcomes. Theoretical frameworks suggest two alternate views. First, Lemerise and Arsenio () proposed that intense emotions may disrupt the online processing of social information at each step, from failing to assess the situation from multiple perspectives to enacting responses that are inappropriate to the situational cues. According to this ?disruptor? view of emotions, children who hold a more hostile attribution bias may be especially prone to engage in negative interactions with friends when those children are also characterized by high EI. A second, alternate view is based on functionalist theories of emotion that emphasize emotions as critical to motivating goal-directed behavior (Magai & McFadden,; Saarni, Mumme, & Campos,). Consistent with this viewpoint, dynamic coactions of emotion and cognition generate and organize individuals? behaviors during social interactions (Mascolo & Fischer,). Specifically, individuals continually appraise how well their concerns are met and attribute why concern-promoting or violating events occur. Individuals? appraisals and attributions shape their emotional experiences (Scherer, Schorr, & Johnstone,), and emotional arousal induces action readiness and motivates responses to promote individuals? concerns (Frijda,). Therefore, intense emotions may not be disruptive per se but may function as ?fuel? that facilitates individuals to behave in ways consistent with their appraisals and attributions. High EI may motivate children who hold a more hostile attribution bias to engage in more negative exchanges but may motivate children who hold a more benign attribution bias to engage in more positive exchanges with friends. Several studies indicate that the combination of intense negative emotions and hostile attribution bias may play a role in children's and adolescents? generation of aggressive behaviors. Compared with nonaggressive children, aggressive children do not only exhibit more hostile attributions but also report higher anger/distress to hypothetical provocations by peers (e.g., Burgess et?al.,; Crick,). In addition, highly aggressive boys typically explain their aggressive responses to provocation vignettes as driven by rage (Orobio de Castro, Verhulp, & Runions,). Notably, hostile attributions and other aggressogenic thoughts were positively associated with aggressive behaviors but only for children and young adolescents who exhibited high dispositional anger (Roos, Hodges, Peets, & Salmivalli,; Runions & Keating,). Likewise, girls? hostile attribution bias was associated with greater relational aggression, but only when self-reported emotional distress to relational provocations was also high (Mathieson et?al.,). These findings can be explained by both the ?disruptor? and ?fuel? interpretations. Experiencing intense anger or distress may strengthen the link between hostile attributions and aggression by ?disrupting? social information processing or by ?motivating? children and adolescents to act on their aggression-related cognitions. To tease apart these alternate interpretations, it is necessary to (a) assess EI at a more global level so that it is not confounded with negative valence and (b) examine whether high EI combined with a more benign attribution bias predicts positive peer outcomes. Importantly, no prior study to our knowledge has examined how EI interacts with attribution biases to predict the quality of children's interactions with friends specifically. Given that children have more positive and intimate exchanges with friends than with general peers (see Newcomb & Bagwell,), the positive moderating effects of EI, as suggested by the ?fuel? interpretation, may have more opportunities to manifest in friendships. Using data from 913 focal children (referred to as ?study children?) who participated in the NICHD Study of Early Child Care and Youth Development (SECCYD), we addressed two objectives. First, we examined the extent to which study children's self-reported attribution biases at Grade 4 predicted the quality of observed interaction between the study children and a designated friend at Grade 6, above and beyond child?friend interaction quality observed at Grade 4. Second, we assessed study children's EI as reported by teachers at Grade 4 as a moderator of the associations between study children's attribution biases and child?friend interaction quality. Based on prior theoretical (e.g., Frijda,; Lemerise & Arsenio,; Mascolo & Fischer,) and empirical (e.g., Burgess et?al.,; Laible et?al.,; Orobio de Castro et?al.,; Runions & Keating,) work, we hypothesized that (a) study children's more hostile versus benign attribution bias would be associated with higher levels of negative child?friend interaction and lower levels of positive child?friend interaction, and (b) EI would moderate these associations, such that they would be stronger among study children who had higher levels of EI. In addition, we sought to test the ?disruptor? (e.g., Lemerise & Arsenio,) versus ?fuel? (e.g., Frijda,) interpretations to better characterize interactions between attribution biases and EI on child?friend interaction quality. We depict hypothetical results that would support the ?disruptor? and ?fuel? interpretations in Figure. As seen in Figures?a and c, the interaction patterns are the same for negative child?friend interaction: A more hostile attribution bias combined with high EI predicts more negative child?friend interaction. The interaction patterns diverge, however, when predicting positive child?friend interaction. In Figures?b and d, a more hostile attribution bias is associated with lower levels of positive interaction only when EI is high, yet the simple slopes at high versus low EI show different patterns in relation to each other. Evidence for the ?disruptor? interpretation (Figure?b) indicates that a more hostile attribution bias combined with high EI predicts less positive interaction, which suggests that high EI may disrupt positive interaction when children tend to attribute hostile intentions. In contrast, evidence for the ?fuel? interpretation (Figure?d) indicates that a more benign attribution bias combined with high EI predicts more positive interaction, which suggests that high EI may promote positive interaction when children tend to attribute benign intentions. We also conducted follow-up analyses to assess the robustness of our main results, including tests of study child gender and friendship status (i.e., whether the study child was observed interacting with the same friend vs. different friends at Grade 4 and Grade 6) as additional moderators. Furthermore, for dyads in which the same friend participated at both time points, we controlled for the friend's attribution bias and EI at Grade 4. Finally, although individuals who experience positive emotions intensely also tend to experience negative emotions intensely (Diener et?al.,; Rydell et?al.,), it is not well understood whether positive and negative EI play similar roles in children's social functioning (Eggum et?al.,; Murphy & Eisenberg,). We explored whether findings from our main model, which examined general EI, were consistent when positive and negative EI were tested as distinct predictors.
PY - 2019/1/1
Y1 - 2019/1/1
N2 - Using data from a subsample of 913 study children and their friends who participated in the NICHD Study of Early Child Care and Youth Development, the interactive contributions of child-reported attribution biases and teacher-reported child emotional intensity (EI) at Grade 4 (M = 9.9 years) to observed child–friend interaction at Grade 6 (M = 11.9 years) were examined. Study children's hostile attribution bias, combined with high EI, predicted more negative child–friend interaction. In contrast, benign attribution bias, combined with high EI, predicted more positive child–friend interaction. The findings are discussed in light of the “fuel” interpretation of EI, in which high-intensity emotions may motivate children to act on their cognitive biases for better or for worse.
AB - Using data from a subsample of 913 study children and their friends who participated in the NICHD Study of Early Child Care and Youth Development, the interactive contributions of child-reported attribution biases and teacher-reported child emotional intensity (EI) at Grade 4 (M = 9.9 years) to observed child–friend interaction at Grade 6 (M = 11.9 years) were examined. Study children's hostile attribution bias, combined with high EI, predicted more negative child–friend interaction. In contrast, benign attribution bias, combined with high EI, predicted more positive child–friend interaction. The findings are discussed in light of the “fuel” interpretation of EI, in which high-intensity emotions may motivate children to act on their cognitive biases for better or for worse.
UR - http://www.scopus.com/inward/record.url?scp=85038435133&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85038435133&partnerID=8YFLogxK
U2 - 10.1111/cdev.13012
DO - 10.1111/cdev.13012
M3 - Article
C2 - 29266173
AN - SCOPUS:85038435133
SN - 0009-3920
VL - 90
SP - e114-e131
JO - Child development
JF - Child development
IS - 1
ER -