Elsevier

Child Abuse & Neglect

Volume 27, Issue 3, March 2003, Pages 285-302
Child Abuse & Neglect

Interpretations of child compliance in individuals at high- and low-risk for child physical abuse

https://doi.org/10.1016/S0145-2134(03)00007-3Get rights and content

Abstract

Objective: Our studies compared individuals at high- and low-risk for child physical abuse on measures of social information processing.

Method: Two studies were conducted using similar methods. Twenty-eight childless women in Study 1 and 36 mothers in Study 2 read vignettes of parent-child interactions in which the child’s level of compliance was difficult to interpret. Participants were asked a series of questions about the child’s behavior and their own reactions.

Results: Accuracy and bias in identifying compliant behavior were assessed using a signal detection paradigm. In both samples, high- and low-risk participants did not differ in their overall accuracy in identifying children’s behaviors. However, they used different evaluation standards such that high-risk participants were biased toward seeing more noncompliance and low-risk participants were biased toward seeing more compliance. High- and low-risk participants also made different types of errors in interpreting children’s behavior. Low-risk participants were more likely to misinterpret noncompliant behavior as compliant, and there was a trend for high-risk participants to not perceive compliant behavior when it occurred. There were no differences in reported disciplinary responses in either study and the results for affective reactions were mixed.

Conclusions: Specific differences in social information processing between high- and low-risk individuals replicated across samples, suggesting a reliable association between evaluation standards and risk of child physical abuse. However, the absence of differences in reported discipline and inconsistent findings on affective reactions indicate the need to identify the mechanism through which cognition influences parenting behavior.

Résumé

Objectif: Nos études ont comparé comment des personnes qui présentent des risques élevés ou peu élevés par rapport à la maltraitance traitent de renseignements à caractère social.

Méthode: Deux études, utilisant des méthodes semblables, ont été menées. À 28 femmes sans enfants (étude # 1) et 36 mères (étude # 2), on a fait lire des historiettes portant sur des interactions entre parent et enfant où le degré d’obéissance de l’enfant aux demandes des parents était difficile à discerner. Les participantes ont répondu à des questions concernant le comportement de l’enfant et leurs propres réactions.

Résultats: On a évalué l’exactitude et le parti-pris des participantes lorsqu’elles identifiaient le comportement obéissant. On a remarqué peu de différences entre les deux échantillons au niveau de leur capacité à identifier justement les comportements des enfants. Toutefois, les deux groupes ont eu recours à des normes différentes pour évaluer ces comportements. Les participantes à risque élevé pour la maltraitance démontrent des partis-pris en ce qu’elles perçoivent plus de nonobéissance qu’il n’existe en réalité, tandis que les participantes à risque inférieur sont portées à percevoir un plus haut niveau de subordination. Les deux groupes ont aussi fait des erreurs diverses dans leur interprétation des comportements des enfants: les participantes à risque non élevé étaient plus aptes à considérer les comportements insubordonnés comme étant subordonnés; on note une tendance chez les participantes à risque élevé de ne pas reconnaı̂tre les comportements subordonnés lorsqu’ils sont présents. Les deux études n’ont révélé aucune différence au niveau de la discipline que les participantes privilégient face à ces comportements. On a noté toute une gamme de réactions affectives.

Conclusions: Dans les deux échantillons, les différences spécifiques au niveau du traitement de renseignements à caractère social sont semblables, ce qui porte à croire qu’il existe une relation fiable entre, d’une part, les normes à utiliser pour évaluer les comportements et, d’autre part, les risques de maltraiter les enfants. Toutefois, l’absence des différences au niveau de la discipline à exercer, et le manque de cohérence au niveau des réactions affectives suggèrent qu’il faudra étudier davantage comment le cognitif influence le comportement parental.

Resumen

Objetivo: Nuestros estudios compararon sujetos alto y bajo riesgo para el maltrato fı́sico infantil en una serie de medidas de procesamiento de información social.

Método: Se llevaron a cabo dos estudios utilizando métodos similares. En el primer estudio, 28 madres sin hijos y en el segundo estudio 36 madres leyeron viñetas sobre interacciones padres-niños en las cuales el nivel de obediencia del niño era difı́cil de interpretar. Se les preguntó a los participantes sobre una serie de cuestiones acerca de las conductas del niño y de sus propias reacciones.

Resultados: La exactitud y el sesgo en la identificación de la conducta de obediencia fue evaluada utilizando el paradigma de la detección de señales. En ambas muestras, los participantes alto y bajo riesgo no mostraron diferencias en la exactitud global para identificar las conductas de los niños. Sin embargo, se observa que los participantes utilizaron diferentes normas de evaluación. Los sujetos alto-riesgo presentaban un sesgo que les hacı́a ver más conductas de no obediencia, mientras que los sujetos bajo riesgo percibı́an más conductas de obediencia. Los sujetos alto y bajo riesgo hicieron diferentes tipos de errores en la interpretación de las conductas de los niños. Los sujetos bajo riesgo tenı́an más tendencia a equivocarse considerando la conducta de desobediencia como conducta de obediencia. Los sujetos alto riesgo tendı́an a no percibir la conducta de obediencia cuando ésta habı́a ocurrido. No se observaron diferencias en las conductas de disciplina notificadas en ambos estudios y los resultados sobre las reacciones afectivas fueron diversos.

Conclusiones: En ambas muestras se replican una serie de diferencias especı́ficas entre sujetos alto y bajo riesgo en el procesamiento de la información social, lo que sugiere que existe una asociación fiable entre las normas de evaluación y el riesgo para el maltrato fı́sico infantil. Sin embargo, la ausencia de diferencias en las conductas notificadas de disciplina y los hallazgos inconsistentes en relación con las reacciones afectivas indican la necesidad de identificar los mecanismos a través de los que las cogniciones influyen en la conducta parental.

Introduction

Theory predicts multiple pathways to the perpetration of child physical abuse and research has supported the importance of individual, family, community, and cultural factors in understanding abusive behavior (Belsky, 1993, Gelles, 1998, Milner & Dopke, 1997). The study of individual factors associated with child physical abuse generally examines the social interactive nature of abuse. That is, the focus has been on both parent and child characteristics associated with abuse. Many researchers (e.g., Bugental et al., 1990, Dietrich et al., 1990) posit that abusive parental behavior often occurs in reaction to difficult child behavior, but parental cognitions and affective reactions to the child are believed to be more important than the child behavior itself (Bugental et al., 1990). In an effort to delineate the cognitive and affective processes that may contribute to harsh discipline and child physical abuse, Milner, 1993, Milner, 2000 proposed a social information processing model that includes four stages: perception of behavior; expectations, interpretations, and evaluations that assign meaning to the behavior; information integration and response selection; and response implementation and monitoring. In Milner’s model, individuals who have committed child physical abuse and individuals at high-risk for child physical abuse are expected to process child-related information differently from individuals who have not committed child physical abuse and are at low-risk to do so. The hypothesized differences in information processing set the stage for escalation of conflict in parent-child interactions, more severe parental responses to child behavior, and the occurrence of abuse.

Evidence is accumulating to support the association between child physical abuse and many specific information processing activities (see Milner, 1993, Milner & Dopke, 1997). For example, data suggest that abusive and nonabusive parents differ in their expectations and evaluations of child behavior as well as the ways in which they integrate information, select responses, and implement responses in parent-child encounters. However, relatively little is known about how abusive and nonabusive parents differ in their interpretations of child behavior. There is some evidence that abusive and high-risk parents believe their children display more behavior problems (e.g., Bradley & Peters, 1991; Kolko, Kazdin, Thomas, & Day, 1993). However, it is not clear whether these beliefs reflect accurate assessments of the children’s behavioral and adjustment difficulties or biased interpretations of normative child behavior. Evaluating the accuracy of parental reports is made especially difficult because of the well-documented deleterious effects of abuse on children’s affect regulation, adjustment, attachment, and behavior (e.g., Herrenkohl et al., 1984, Kavanagh et al., 1988).

In the only two known studies directly comparing parent and observer reports of child behavior problems among children who have been abused (Mash et al., 1983, Reid et al., 1987), parents who had abused were found to report more behavior problems in their children than parents who had not abused whereas the reports of observers did not differ for abused and nonabused children. These data provide preliminary evidence that parents who have abused and those who have not abused differ in their interpretations of child behavior. However, parent reports in both studies were based on global ratings of child behavior over extended time periods whereas observer reports reflected discrete behaviors during a short time period. The lack of situational and behavior concordance between parent and observer reports in these studies makes direct comparison unfeasible (Richters, 1992) because abused children may exhibit behavior problems in the home environment that are not evident during observational sessions. To address the lack of situational and behavioral concordance that has hampered interpretation of earlier results, child behavior should be standardized or parent and observer reports should be based on the same samples of behavior.

We conducted two studies examining the interpretations of child behavior made by individuals at high- and low-risk to commit child physical abuse. We standardized the child behavior to be interpreted by providing participants with vignettes depicting parent-child interactions in which the child either complied or failed to comply with parental requests. We chose to focus on child compliance and noncompliance because child noncompliance is often cited as a precipitant to abusive incidents by parents (Dietrich et al., 1990; Herrenkohl, Herrenkohl, & Egolf, 1983; Kadushin & Martin, 1981). To strengthen the test of differences between individuals at high- and low-risk to commit child physical abuse in social information processing, we used a process-based methodology in addition to an experience-based methodology (McFall, Treat, & Viken, 1998). Experience-based methodologies typically measure cognitions through questionnaire formats and assume individuals are both willing and able to convey accurately introspective knowledge. In contrast, process-based methodologies do not make such assumptions (see Bargh & Chartrand, 1999, McFall et al., 1998) and, therefore, rely on observable information processing patterns to understand how information is processed. That is, experience-based studies tend to rely on questionnaires to obtain information, whereas process-based studies gather data in a controlled fashion and use established mathematical models to convey the nature of cognitive processing. We incorporated both experience- and process-based measures in our study using a signal detection paradigm to assess interpretations of child compliance and questionnaires to assess adults’ affective and behavioral reactions to child behavior. We chose to use the signal detection paradigm to evaluate difference in adult interpretations because standards for evaluating child compliance/noncompliance may be outside the participants’ awareness and therefore not easily articulated on experience-based questionnaires.

Signal detection theory (SDT) addresses the ways in which patterns are recognized under conditions of ambiguity and allows for the measurement of both accuracy and bias (Snodgrass & Corwin, 1988). The basic paradigm used in SDT research involves the presentation of a signal in a field of noise. In our studies participants read a series of vignettes in which half of the vignettes depicted a child complying with a parents’ request and half depicted a child not complying with the request. Compliant child behavior was the signal and noncompliant behavior was the noise. In the SDT paradigm “accuracy” refers to the proportion of correct decisions and includes “hits” (i.e., correctly identifying the signal) and “correct rejections” (i.e., not identifying the noise as signal). In our studies, participants who correctly identified many compliant behaviors as compliance and did not misidentify noncompliant behaviors as compliance received high accuracy scores. “Bias” in SDT indicates the type of errors people make in differentiating between signal and noise. Some people use a fairly liberal standard in identifying the signal and therefore make “false-alarm” errors in which they tend to identify noise as signal. In our studies, participants who used a liberal criterion would make false-alarm errors by identifying noncompliant behaviors as compliance. Other people use a conservative or strict standard and will make “miss” errors in which they fail to identify the signal. In our studies, participants who made miss errors would fail to identify instances of compliance. Based on Milner’s model and previous empirical findings (Mash et al., 1983, Reid et al., 1987), we predicted that individuals at high-risk to commit child physical abuse would be less accurate in identifying child compliance and use a more conservative criterion in deciding if a child complied to a parent’s request than individuals at low-risk to commit child physical abuse.

We tested our predictions in two samples using similar procedures. In Study 1, a sample of undergraduate women identified child compliance in a set of 34 written vignettes. In Study 2, a community sample of mothers of young children identified compliance in a set of 28 vignettes. In both samples, participants were designated as being at high-risk or low-risk to commit child physical abuse based on their scores on the Child Abuse Potential Inventory (CAP) (Milner, 1986). Although our primary goal was to examine interpretations of child behavior, we also assessed other reactions that have been associated with child physical abuse (Dopke & Milner, 2000, Milner, 1993, Milner, 2000). Specifically, measures of affect and behavioral reactions were included to examine whether high-risk participants reported more negative emotions and more punitive discipline in response to child behavior than low-risk participants.

Section snippets

Participants

Sixty-three female college students enrolled in undergraduate psychology courses at a medium sized public university participated in the study. Preliminary selection of participants involved eliminating protocols containing an invalid CAP (Milner, 1986; see the “Child Abuse Potential Inventory” section below for a description of the CAP), the criterion upon which study groups were formed. Of these, 23 were eliminated because their responses to CAP items produced an invalid profile. The percent

Results

Means, standard deviations, t tests, and effect sizes for the signal detection indices for high- and low-risk individuals are presented in Table 1. Contrary to expectations, high-risk individuals were not significantly different from low-risk individuals on the accuracy index. That is, the groups did not differ in the proportion of correct decisions made. The predicted difference on the bias index was significant, however, with high-risk, compared to low-risk, individuals using a more

Participants

To determine if the findings from a college sample would replicate with a parent sample, 80 mothers with a child between 2 and 5 years of age participated in a second study. Of these, 17 (21.25%) had invalid CAP protocols and were excluded from further analyses. In this sample, we used Milner’s (1986) recommended CAP cutoff score of 166 to assign mothers to the high-risk group. We attempted to use the recommended CAP score of 91 as the upper limit for a low-risk group; however, because of the

Results

As can be seen in Table 2, high-risk, compared to, low-risk mothers endorsed more stress in the parenting role, reported higher levels of negative affect, and reported greater use of hostile/coercive parenting. Moreover, mothers in the high-risk group reported a higher frequency of common child behavior problems in their own children and perceived those behaviors as more problematic than mothers in the low-risk group. High-risk mothers also reported lower levels of supportive/engaged parenting

General discussion

The present studies were designed to examine the role that interpretations of child behavior might play in harsh discipline and child physical abuse. Guided by Milner’s social information processing model (1993, 2000) and using a SDT paradigm, individuals at high-risk and low-risk to commit child physical abuse were compared on accuracy and bias in identifying child compliance and noncompliance. Risk groups were also compared on levels of negative affect and punitive and inductive discipline

References (47)

  • W.D. Bauer et al.

    Abusing, neglectful, and comparison mothers’ responses to child-related and nonchild-related stressors

    Journal of Consulting and Clinical Psychology

    (1985)
  • J. Belsky

    Etiology of child maltreatment: A developmental-ecological analysis

    Psychological Bulletin

    (1993)
  • S.R. Boggs et al.

    Concurrent validity of the Eyberg Child Behavior Inventory

    Journal of Clinical Child Psychology

    (1990)
  • E.J. Bradley et al.

    Physically abusive and nonabusive mothers’ perceptions of parenting and child behavior

    American Journal of Orthopsychiatry

    (1991)
  • D. Bugental et al.

    Caregiver beliefs and dysphoric affect directed to difficult children

    Developmental Psychology

    (1990)
  • L.A. Clark et al.

    Mood and the mundane: Relations between daily life events and self-reported mood

    Journal of Personality and Social Psychology

    (1988)
  • Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). New York: Academic...
  • Dunsterville, E. (1995). A test of accuracy in the discrimination of child compliance and noncompliance. Unpublished...
  • S.M. Eyberg et al.

    Assessment of child behavior problems: The validation of a new inventory

    Journal of Clinical Child Psychology

    (1978)
  • Forehand, R. L., & McMahon, R. J. (1981). Helping the noncompliant child: A clinician’s guide to parent training. New...
  • Gelles, R. J. (1998). The youngest victims: Violence towards children. In R. K. Bergan (Ed.), Issues in intimate...
  • R.C. Herrenkohl et al.

    Circumstances surrounding the occurrence of child maltreatment

    Journal of Consulting and Clinical Psychology

    (1983)
  • G.W. Holden et al.

    Why 3-year-old children get spanked: Parent and child determinants as reported by college-educated mothers

    Merrill-Palmer Quarterly

    (1995)
  • Cited by (33)

    • The social information processing model in child physical abuse and neglect: A meta-analytic review

      2020, Child Abuse and Neglect
      Citation Excerpt :

      Research conducted with high-risk of abuse and abusive parents has been showing that these parents are more likely to hold more inaccurate and biased preexisting cognitive schemata. For example, this research has shown that these parents value physical punishment as a disciplinary technique (e.g., Ateah & Durrant, 2005; Rodriguez, 2018; Wang, Wang, & Xing, 2018), hold unrealistic expectations about child development (e.g., Azar & Rohrbeck, 1986; Haskett, Scott, Willoughby, Ahern, & Nears, 2006; McElroy & Rodriguez, 2008), have negative implicit attitudes towards children (e.g., Risser, Skowronski, & Crouch, 2011), present higher accessibility of negative schemata (e.g., Crouch et al., 2012, Crouch, Risser et al., 2010; Hiraoka et al., 2014; Milner et al., 2011), show an external locus of control (e.g., Rodriguez, 2010; Rodriguez & Richardson, 2007), are less empathic (e.g., Francis & Wolfe, 2008; Pérez-Albéniz & De Paúl, 2003; Pérez-Albéniz & De Paúl, 2004), and present more negative affect (e.g., Bradley & Peters, 1991; Dadds, Mullins, McAllister, & Atkinson, 2003; Dopke, Lundahl, Dunsterville, & Lovejoy, 2003; Pérez-Albéniz & De Paúl, 2006). Surprisingly, much less attention has been given to neglectful parents.

    • Child physical abuse risk moderates spontaneously inferred traits from ambiguous child behaviors

      2013, Child Abuse and Neglect
      Citation Excerpt :

      Such differences were not observed for high CPA risk parents: The implicit influence of negative trait inferences was similar for vaguely implied negative traits and strongly implied negative traits. For low-risk parents, when negative traits are only vaguely implied, their implicit influence may be insufficient to override the positive interpretive tendencies of low CPA risk parents (Crouch et al., 2010; Dopke, Lundahl, Dunsterville, & Lovejoy, 2003). In contrast, strongly implied negative traits may lead low CPA risk parents to override their positive interpretive tendencies, resulting in attribution of a negative trait to a child.

    View all citing articles on Scopus

    Authorship order was determined randomly and does not reflect relative contributions of the authors.

    1

    Study 2 is based on a thesis completed by Brad W. Lundahl under the supervision of M. Christine Lovejoy in partial fulfillment of the requirements for the Master of Arts degree at Northern Illinois University.

    View full text