Linking the Child Behavior Checklist (CBCL) with the Multidimensional Assessment Profile of Disruptive Behavior (MAP-DB): Advancing a dimensional spectrum approach to disruptive behavior

Abstract

Disruptive behavior in childhood is common. It spans from normative child misbehaviors to clinically-significant and impairing problems. While there are many rating scales evaluating such behaviors, historically, measurement has emphasized counting the number of symptoms present rather than assessing the normal-abnormal spectrum of behavioral expression. This study uses data from 644 early school age children aggregated from two data sources to statistically link a commonly used symptom count measure, the Child Behavior Checklist (CBCL), to a more developmentally-sensitive measure, the Multidimensional Assessment Profile of Disruptive Behavior (MAP-DB). Two links between conceptually similar scales on each measure were developed: CBCL Conduct Problems and MAP-DB Aggression; and CBCL Oppositional Defiant Problems and MAP-DB Temper Loss. We compared two innovative methods-Item Response Theory (IRT) and Deming regression-to determine the optimal linking relationship. Results suggest IRT methods were superior in reducing linking error compared to Deming regression. While Deming regression accurately modeled the mean scores (thus minimizing linking bias), this method could not adequately address the floor effect for scores on the CBCL. For practical purposes, this study provides a crosswalk of score conversions between the CBCL and MAP-DB, such that data aggregation and group comparisons can be made across the two measures; this enables longitudinal analyses with historically-collected CBCL data to transition to the more innovative dimensional scales of the MAP-DB without undo loss of extant data. This study furthers efforts to shift from historical symptom counts to more developmentally-sensitive measurement across the disruptive behaviors spectrum.

Keywords: Assessment; CBCL; Deming Regression; Developmental psychopathology; Disruptive Behavior; Item Response Theory; Linking; MAP-DB.

Figure 1:. Deming Regression Score Conversions

Non-IRT score conversions, using Deming regression with theoretical and sample-observed variance ratios. As is apparent, there are significant floor effects on the CBCL, while the MAP-DB makes finer distinctions among those within the average severity ranges.

Figure 2:. Conduct Problems and Aggression Bland-Altman Plots

Bland-Altman plot representing the agreement between scores linked from the CBCL and obtained on the MAP-DB Early School Age Form. The mean of the linked and actual Aggression score is on the x-axis, whereas the difference (actual minus linked) is on the y-axis. A pronounced floor effect from the CBCL is apparent, regardless of method used for linking.

Figure 3:. Oppositional Defiant Problems and Temper Loss Bland-Altman Plots

Bland-Altman plot representing the agreement between scores linked from the CBCL and obtained on the MAP-DB Early School Age Form. The mean of the linked and actual Aggression score is on the x-axis, whereas the difference (actual minus linked) is on the y-axis. A pronounced floor effect from the CBCL is apparent, regardless of method used for linking.