Reading between the lines: Implicit assessment of the association of parental attributions and empathy with abuse risk

Abstract

Objective

Researchers in the child maltreatment field have traditionally relied on explicit self-reports to study factors that may exacerbate physical child abuse risk. The current investigation evaluated an implicit analog task utilizing eye tracking technology to assess both parental attributions of child misbehavior and empathy.

Method

Based on the observation that readers experience comprehension difficulty when encountering passages inconsistent with their beliefs, an eye tracker gauged the extent of difficulty parents experienced reading vignettes that inappropriately characterized a child as culpable for misbehavior and that presented a non-empathic child interaction.

Results

Results suggest self-reports of attributions and empathy are related to both child abuse potential and discipline intentions; however, the eye tracking analog for empathy correlated with abuse potential but not punishment decisions whereas the analog for attributions correlated with punishment decisions but not abuse potential.

Conclusions

Such contrasts between self-report and analog assessment underscore the need for continued research studying theorized abuse risk constructs using alternative approaches to better identify the important risk markers associated with elevated child abuse risk and to minimize methodological overlap.

Introduction

Researchers have invested considerable energy attempting to predict the likelihood a parent will engage in abuse, a concept typically referred to as child abuse potential (Milner, 1994). Child abuse potential has been linked with harsh parenting style and more frequent use of parent-child aggression, including abusive tactics (Rodriguez, 2010a). Not surprisingly, abusive parents typically administer excessive, harsh discipline (Veltkamp & Miller, 1994). Greater confidence regarding which factors truly elevate child abuse risk is thereby critical for advancing prevention and intervention efforts. The current study presents a preliminary investigation into a novel approach to studying factors previously theorized to predict physical child abuse potential.

One model of cognitive processes that theoretically lead a parent to engage in physical abuse is Social Information Processing (SIP) theory, delineated by Milner (2000). SIP theory postulates that pre-existing cognitive schema (e.g., a parent's child- and discipline-related beliefs) exist prior to processing information from a new interaction. Then, when faced with a discipline situation, 4 stages begin. The parent must first accurately attend to and perceive the situation in Stage 1, then in Stage 2, the parent develops expectations and interpretations of the situation. Problems with accurate perceptions followed by biases in interpretation may be compounded by the parent's difficulty integrating mitigating information that could account for the child's behavior or difficulty considering alternative responses in Stage 3. In Stage 4, the parent who selects physical discipline encounters difficulty monitoring their implementation of discipline that could culminate in abuse. Considerable research supports individual elements of stages (e.g., Milner, 2000, Rodriguez, 2010b) as well as multiple components from different stages (e.g., De Paúl et al., 2006, Rodriguez and Richardson, 2007).

The present study examined 2 factors previously proposed in the SIP model, parental attributions (a Stage 2 process) and empathy (a Stage 3 process). Attributions that a child is misbehaving intentionally can affect a mother's decision to implement physical discipline (Ateah & Durrant, 2005). Abuse-risk mothers make negative dispositional attributions for children's negative behaviors yet attribute positive behaviors to external causes (Dadds, Mullins, McAllister, & Atkinson, 2003). Indeed, physically abusive mothers interpreted child misbehavior as intentional annoyance significantly more often than a matched sample of non-abusive mothers (Haskett, Scott, Willoughby, Ahern, & Nears, 2006). Collectively, research generally supports that parental attributions of children's behavior represent a potentially important component in increased abuse risk, although others have suggested that higher abuse risk parents do not demonstrate more negative attitudes toward children (Risser, Skowronski, & Crouch, 2011) and questions have arisen regarding the optimal measurement of attributions (Milner, 2003).

Negative attributions can be potentially attenuated by integrating mitigating information in the third SIP stage, such as a parent's empathic ability that could inform of a situation-specific context that lessens a child's perceived responsibility (Milner, 2000). An inability to experience empathy has been previously linked to physically abusive parenting (Feshbach, 1989, Milner and Dopke, 1997). Low empathic perspective-taking ability has been associated with greater child abuse potential in community (Rodriguez & Richardson, 2007) and at-risk samples (McElroy and Rodriguez, 2008). Empathy and attribution of child intent were specifically targeted for the current investigation because both suffer from one of the obstacles in confidently predicting their true role in elevated physical child abuse risk: our methodological reliance on measuring these concepts via self-report.

Many researchers in this field often conclude their studies citing as a limitation their dependence on eliciting information directly from participants via questionnaires, often not even anonymously. This limitation is exacerbated by designs investigating both the risk factors and the dependent variables using self-report from the same source, which raises concerns about source bias as well as questions about item or conceptual overlap between questionnaires for the predictors and the proposed outcome(s). Even in those instances when participants are assured of anonymity, such designs rely on the participants’ candor and accuracy. This strategy of obtaining self-report is considered an explicit assessment technique. Unfortunately, with more controversial constructs, explicit self-reports are more likely to lead to responses that are manipulated by participants (Fazio & Olson, 2003). Clearly this self-report design flaw is problematic for researchers in a number of fields, but this approach is particularly challenging for researchers studying child abuse risk and its related constructs (DeGarmo, Reid, & Knutson, 2006). In fact, honesty in reporting could lead to negative consequences which may be realistically appraised as threatening for parents (Bennett, Sullivan, & Lewis, 2006), inducing participants to inaccurately respond to self-report questions. Those already substantiated for abuse seem especially likely to distort their responses. Thus, respondents to questionnaires may intentionally misrepresent their true attitudes but some respondents may even unconsciously present a socially desirable image because they want to believe this positive self-perception (Fazio & Olson, 2003). Yet this field continues to rely heavily on self-report to assess risk because alternative assessment approaches are limited.

To better control biases in responding, analog tasks strive to assess a concept using implicit means (Fazio & Olson, 2003). Analog tasks are designed to assess the concept without the participant being explicitly aware of what the task is attempting to measure and/or how it is being scored. If a respondent is not overtly aware of what is being measured, they are less able to misrepresent their response. Analog tasks differ in how much automatic, unconscious processing is engaged in the task, wherein more conscious processing enables a participant to manipulate their responses (Fazio & Olson, 2003). Designing analog tasks thus demands creativity to minimize participant awareness. Another challenge for the application of analog tasks lies in the fact that some may be labor-intensive or costly. However, such assessment approaches are essential when evaluating models for abuse risk (DeGarmo et al., 2006); questions continue to arise about whether a given factor is indeed associated with abuse risk because of doubts about whether the assessment tools simply overlap in content or are manipulated by the respondent. For example, if an analog task can demonstrate an association with self-report measures of a given concept, then those utilizing the self-report measure can be more confident in the latter's use as a proxy.

Relatively few analog approaches have been considered in the child maltreatment field. Some early research investigated parents’ punitive responses under stress (Passman & Mulhern, 1977) or elicited discipline responses in real time while parents watched a video of child misbehavior (Fagot, 1992), moderately implicit approaches attempting to approximate parent discipline behavior. More recently, subliminal priming with photos has been used to study implicit parental attitudes of hostility, where abuse risk parents rated ambiguous child pictures as more hostile relative to low-risk parents (Farc, Crouch, Skowronski, & Milner, 2008), although higher abuse risk parents did not evidence more negative attitudes toward children (Risser et al., 2011). Higher abuse risk parents have greater difficulty processing either positive or negative stimuli when it is incongruent with a subliminal prime rather than simply evidencing a bias toward negative stimuli only (Crouch et al., 2010). This recent line of research highlights the need to clarify some of the implicit cognitive processing occurring in parents.

The current study evaluated an analog approach based on eye tracking technology that requires no participant response and thus would be relatively free from conscious processing. Eye trackers monitor and analyze the movement of the eye during reading (Rayner, 1998). While reading, eyes display short ballistic movements, known as saccades, or pause on sections of text, known as fixations. English is read left to right but readers occasionally evidence right-to-left eye movements, known as regressions, to review previously read material (Rayner, 1998). Reading speed is affected by the frequency and duration of fixations and regressions as well as one's overall comprehension of the material (Rayner, Chace, Slattery, & Ashby, 2006). Reading researchers have found that readers spend more time reading and rereading text inconsistent with the reader's expectations or general world knowledge (Cook and Guéraud, 2005, Cook and Myers, 2004, Rayner et al., 2006). Based on these findings, we measured parents’ eye movements as they read two types of vignettes: stories either appropriately or inappropriately represented children as culpable for an incident; stories that portrayed either an empathic or non-empathic scenario. Consider the first example in Table 1. In the culpable condition, the child is instructed not to touch the cupcakes, but the child eats them anyway. The subsequent target sentence, “The child knew their behavior was wrong….” should be easy to comprehend, and reading times should be relatively fast. However, in the non-culpable condition, where the child is instructed to get their father and does so, the subsequent target sentence should be much more difficult to comprehend. Thus, reading times for the target sentence in the non-culpable condition should be much slower than in the culpable condition. The second example in Table 1 contains a target sentence (“Fred saw Mary and laughed at the look on her face”) that either represents an empathic or non-empathic response. The target sentence should be more difficult to understand, and thus should be read more slowly, in the non-empathic condition than in the empathic condition. Based on these principles, we utilized eye tracking technology to assess reading speed, evaluating whether individuals would evidence greater difficulties reading material that inappropriately presented children as culpable or presented a non-empathic scenario.

The present study investigated whether an analog task based on the reading inconsistency paradigm could assess both parental attribution of child intentionality and empathy as it may relate to physical abuse risk and punishment intentions. Parents self-reported on their child abuse potential and punishment intentions as well as on their perception of their empathic ability and attributions of child intent. Analog scores from the eye tracking task were compared to scores on explicit measures of self-reported empathy and attributions and abuse potential and punishment intentions. Because the analog approach is entirely implicit (no participant response is required), eye tracking scores were expected to show modest correlations with explicit measures.

Specifically, parents who demonstrated greater difficulty reading non-empathic vignettes were predicted to self-report greater empathy, obtain lower child abuse potential scores, and indicate less inclination to punish children. Similarly, parents who demonstrated greater difficulty reading inappropriate implications of child culpability were expected to self-report less inclination to believe children misbehave intentionally to be annoying and would demonstrate lower abuse potential and less intention to punish misbehavior. Such a design will provide a novel approach to considering how these two theorized SIP processes, attributions and empathy, may indeed relate to elevated child abuse risk.