Abstract
Objectives
Retrospective harmonisation of data obtained through different instruments creates measurement error, even if the underlying concepts are assumed the same. We tested a novel method for item-level data harmonisation of two widely used instruments that measure emotional and behavioural problems: the Child Behavior Checklist (CBCL) and the Strengths and Difficulties Questionnaire (SDQ).
Methods
Item content of the CBCL and SDQ was mapped onto four dimensions: emotional problems, peer relationship problems, hyperactivity/inattention and conduct problems. A diverse test sample was drawn from four prospective longitudinal birth cohort studies in Australia and Europe who used one or both instruments. The pooled sample included 5188 data points assessing children and adolescents aged 6–13 years (N = 257–704 participants per cohort). Measurement invariance was assessed using latent variable multi-group confirmatory factor analysis.
Results
Fifteen items from the CBCL and SDQ were mapped onto four dimensions allowing for measurement invariance testing as part of a stepwise process. Partial strict invariance between CBCL and SDQ assessments was established for all four dimensions.
Conclusions
The harmonised dimensions of emotional, peer relationship, hyperactivity/inattention and conduct problems are invariant across the CBCL and SDQ suggesting that these dimensions can be reliably compared with limited measurement error.