Profiles of anger control in second-grade children: Examination of self-report, observational, and physiological components

Abstract

The current study used latent profile analysis (LPA) to examine anger control in 257 second-grade children (∼8 years of age). Anger was induced through losing a game and prize to a confederate who cheated. Three components of anger control were assessed: self-report of awareness of anger, observed intensity of angry facial expressions, and skin conductance reactivity. These components served as indicators in an LPA conducted to determine whether distinct groups of children who differed in anger control profiles would emerge. Five groups were found: (a) Physiology-and-Expression Controllers (high self-report, low expression, low physiological arousal), (b) Expression-Only Controllers (high self-report, low expression, high physiological arousal), (c) Non-controllers (high self-report, high expression, medium physiological arousal), (d) Non-reactive (low self-report, low expression, low physiological arousal), and (e) Non-reporters (low self-report, medium expression, medium physiological arousal). These findings are discussed in terms of implications for the assessment of children’s anger control skills and intervention programs for children’s anger management.

Introduction

Controlling anger is an important developmental task of middle childhood, and children who struggle to regulate anger are at risk for externalizing disorders (e.g., Cole, Teti, & Zahn-Waxler, 2003). As early as infancy, babies learn that expressions of anger are not well received by others (Malatesta & Haviland, 1982). As they develop, children’s understanding that the expression of anger is frowned up on becomes increasingly sophisticated (Scheff, 1984), particularly in the peer context (Shipman, Zeman, Nesin, & Fitzgerald, 2003). Children in middle childhood report that, among peers, they are much less likely to express anger than any other emotion, because they believe angry expressions will produce adverse interpersonal consequences (e.g., Underwood, Hurley, Johanson, & Mosley, 1999). In fact, children who express anger toward peers are likely to be rejected (e.g., Eisenberg, Fabes, Guthrie, & Reiser, 2000). Children in middle childhood may be especially motivated to control angry feelings when with peers in an effort to appear cool and avoid embarrassment (e.g., Lemerise & Dodge, 2008). Thus, in middle childhood, a premium is placed on maintaining emotional composure and controlling anger.

The current study sampled second-grade children, who are in the early stages of this middle childhood period. Previous observational studies of anger expression in middle childhood have shown that second graders express more anger and control their anger less expertly than fourth and sixth graders (Underwood et al., 1999). Thus, we chose to study second graders to learn more about the emergence of anger control in the early years of middle childhood.

Furthermore, the current study used a procedure designed both to induce angry feelings in children and to motivate them to control their anger. Anger was induced by having children lose a competitive board game and a desired prize to a confederate peer who cheated. The procedure aimed to promote anger control by placing children with a same-age peer. This peer was an unfamiliar child, and children may be even more likely to conceal anger when with unfamiliar peers than with close friends (Underwood, Coie, & Herbsman, 1992). Thus, this procedure was well-suited to both evoke anger and motivate children to control anger.

In the past, researchers have typically assessed anger control in middle childhood using paper-and-pencil measures completed by children, parents, or teachers (e.g., the Children’s Anger Management Scale by Zeman and colleagues; Zeman et al., 2001, Zeman et al., 2002). Questionnaire measures of children’s anger control allow for efficient assessment, and they are clearly important for this reason. However, for the current project, our goal was to develop a laboratory-based assessment of anger control suitable for the middle childhood period. Researchers have not used laboratory-based assessments in the past, and we hoped to fill this gap in a number of ways. In particular, we aimed to design a procedure that was ecologically valid, standardized anger-inducing stimuli across children, and allowed for the assessment of multiple components of children’s anger control (e.g., observational, physiological).

Anger control is one part of the broader and more complex construct of emotion regulation. One of the most commonly-cited definitions in the child development literature states that emotion regulation includes “the extrinsic and intrinsic processes responsible for monitoring, evaluating, and modifying emotional reactions … to accomplish one’s goals” (Thompson, 1994, pp. 27–28). This definition implies that regulation may involve either increasing or decreasing the experience or expression of an emotion. In the current study, however, we focused more narrowly on the control, or reduction, of the particular emotion of anger.

Theorists conceptualize three components of emotion and its regulation: an experiential component, an expressive component, and a physiological component (e.g., Cole et al., 2004, Thompson et al., 2008). The experiential component is most often assessed through self-report, the expressive component through observation, and the physiological component through measures such as heart rate and skin conductance. In the current study, we included self-report, observational, and physiological measures of children’s anger in the context of the anger-inducing game with the peer confederate.

Traditionally, theorists have assumed that these components cohere and vary together as a unitary construct (Ekman, 1992, Izard, 1977). However, empirical support for this coherence is mixed and weak (see Mauss, Levenson, McCarter, Wilhelm, & Gross, 2005, for a succinct review and for their own findings). Moreover, in four separate studies of middle-childhood participants playing anger-inducing games with peer confederates (Casey, 1993, Hessler and Katz, 2007, Hubbard et al., 2004, Underwood and Bjornstad, 2001), weak correlations emerged across the self-report, observational, and physiological components of anger.

Thus, although some children may show high or low levels of all three components of anger, other children may display high levels of some components but low levels of others, explaining the weak relations among components. This idea suggests that each of these components is useful in understanding how children control anger. However, the components may be better conceptualized as anger control profiles, in which children could display high or low levels of each component independent of the others.

The goal of the current study was to examine whether distinct profiles of the three components emerged when second-grade children were placed in an anger-inducing situation. Many different profiles are possible if children potentially could display low, medium, or high levels of self-reported anger, observed anger, and physiological arousal. Because this study is the first to examine anger control profiles in children, we considered our work to be exploratory. Thus, with the exception of a group predicted to struggle with all aspects of anger control, we did not hypothesize that particular profile types would emerge. Instead, we empirically investigated whether distinct groups would emerge. Latent profile analysis (LPA) was used to investigate whether children clustered into groups based on the levels of self-reported, observed, and physiological anger that they displayed. These analyses represented a reexamination of data from a previous study by Hubbard et al., 2002, Hubbard et al., 2004.

The examination of profiles across the three components may have implications for the assessment of children’s anger control. If children fall into distinct groups with different anger control profiles, this finding will suggest that assessment of all three components is essential to understanding each child’s approach to anger control. These profiles will contain important information not available from any single measure of anger, and they will indicate that we should avoid assessing children’s anger control using only one of the components.

Our primary goal was to examine whether distinct profiles of anger control emerged when children were placed in an anger-eliciting situation. A secondary goal was to investigate whether the groups that emerged differed in classroom-based social preference and aggression. Studies suggest that children who struggle with anger control are more aggressive and rejected than other children (e.g., Denham et al., 1990, Fabes and Eisenberg, 1992). For this reason, we hypothesized that a group of children who particularly struggle with anger control would emerge and that this group would be significantly lower on social preference and higher on aggression than the other groups. We predicted that this group would display the highest levels of self-reported anger, observed anger, and physiological arousal of all groups.

Thus, in contrast to our general hypothesis about a lack of coherence among the anger control components across all children, we predicted that this most dysregulated group of children would display coherence, in the form of elevated responses across all three components. This prediction is counter to two previous studies suggesting that children with externalizing problems show even weaker coherence across emotion components than their peers (Hastings et al., 2009, Marsh et al., 2008). However, in these studies, other emotions besides anger were induced. We hypothesized that children who are peer-rejected and aggressive would show greater coherence than their peers when the emotion evoked was anger in particular. Furthermore, coherence across emotion components may be greater as emotional intensity increases (Davidson, 1992, Tassinary and Cacioppo, 1992). We expected that the externalizing group of children would experience more intense anger than their peers in response to losing the game to the confederate peer who cheated, resulting in their high levels of all three components.

Our final goal was to explore whether the groups that emerged differed on sex or race/ethnicity. In previous work, boys have been shown to struggle more than girls with both aggression and emotion regulation (e.g., Moffitt and Caspi, 2001, Saarni, 1984). Thus, we predicted that the group of children who had the most difficulties with anger control would contain more boys than girls. In all other respects, however, our analyses regarding sex and racial/ethnic differences across the groups were exploratory.