Validity indices for interpreting informant discrepancies in ADHD assessment

In ADHD assessment, discrepancies between informants’ test scores complicate decision-making with respect to treatment and educational adaptations. These informant discrepancies may be due to meaningful differences such as variations in assessed behavior across settings (i.e., school vs. home) but may also be due to response biases or unsystematic error. We propose using response-pattern-based validity indices for studying which is the most plausible of the two explanations. These indices detect invalid test scores through identifying extreme, repetitive, or inconsistent response patterns. To illustrate the use of validity indices for interpreting informant discrepancies, we used a subset of data (N = 431) from the self-report and parent-report version of Conners ADHD rating scales collected in the NeuroIMAGE study. Pairs of self-report and parent-report scores were classified as either discrepant or non-discrepant, and validity indices were applied to classify self- and parent-report scores as suspect (i.e., potentially invalid due to rater effects) or not. Finally, we demonstrate how information from validity indices can be taken into account in diagnostic decision-making.