Academic journal article Merrill-Palmer Quarterly

A Multitrait–Multimethod Approach to Assessing Childhood Aggression and Related Constructs

Academic journal article Merrill-Palmer Quarterly

A Multitrait–Multimethod Approach to Assessing Childhood Aggression and Related Constructs

Article excerpt

Aggression has become the most common primary presenting concern for children referred for therapy and related services, because this behavior causes significant impairment across both home and school contexts (Armbruster, Sukhodolsky, & Michalsen, 2004; Nelson & Finch, 2000). Moreover, aggressive behavior has been linked to a host of negative outcomes, including delinquency, risky sexual behavior, school dropout, and depression (e.g., Huesmann, Eron, Lefkowitz, & Walder, 1984; Loeber et al., 1993; Moffitt, 1993). As a result, researchers have attempted to understand the origins, developmental trajectory, and types of aggression in an effort to better prevent and intervene with children's aggressive behavior.

These efforts, however, have been hindered by two complexities related to the assessment of childhood aggression. First, aggressive behavior has evolved into a multidimensional construct with theories and assessments targeting a variety of different subconstructs, including proactive and reactive aggression, callous-unemotional (CU) traits, and anger dysregulation, making it difficult to determine the unique contribution of each subconstruct to our understanding of children's overall aggressive behavior. Second, a variety of sources are often used by researchers to measure children's aggression, including parents, teachers, and self-reports from children. The use of different informants across studies presents challenges when attempting to untangle the extent to which results translate across investigations and sources, as well as the degree to which findings are dependent on the source of information used to assess aggressive behavior. The purpose of the current study was to use a multitrait-multimethod (MT-MM) design and a correlated trait-correlated method (CT-CM) analytic approach to address each of these issues.

Data from MT-MM designs, in which multiple traits are measured through multiple informants, have historically been examined through correlation matrices to disentangle method from trait variance and to evaluate convergent and discriminate validity among the measured traits (Campbell & Fiske, 1959). However, more contemporary analytic approaches resulting from an MT-MM design operate within a latent variable modeling framework (Bauer et al., 2013; Eid, Lischetzke, Nussbeck, & Trierweiler, 2003; Eid et al., 2008; Lance, Noble, & Scullen, 2002) and are termed CT-CM analyses. These analytic approaches provide greater clarity with respect to the role of both methods and traits on ratings of children's aggression.

In the current study, ratings were collected from parents, teachers, and children on the four constructs of proactive aggression, reactive aggression, CU traits, and anger dysregulation. Then, a model was constructed in which the observed ratings were used to create two sets of latent variables. First, latent variables representing each of the informants were created by using data from each of the constructs. Second, latent variables indexing each of the constructs were formed by using data from each of the informants. This analytic model provided robust measures of the four constructs that were purged of error variance in general and method variance attributable to informant in particular. The approach also afforded a more robust means of gauging convergent and discriminant validity than did simple zero-order correlations.

Childhood Aggression as a Multidimensional Construct

The first issue to complicate the assessment of children's aggressive behavior is that a variety of different constructs (or traits, as they are termed in the MT-MM approach) have emerged in this research area. However, these constructs have been examined in largely independent literatures by mostly separate groups of researchers. Thus, it is difficult to determine the extent of overlap of these constructs or the degree to which each construct contributes uniquely to our understanding of childhood aggression. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.