Establishing agreement between parent-reported and directly-measured behaviours

THE QUALITY AND ACCURACY OF research findings relies on the use of appropriate and sensitive research methods. To date, few studies have directly compared quantitative measurement methods in the early childhood field and the extent to which parentreported and directly-measured behaviours agree is unclear. Existing studies are hampered by small sample sizes and the use of statistical techniques which quantify the magnitude of association between measures (e.g. correlations), but not agreement. Here we review the limitations of existing method comparisons and suggest how alternative statistical approaches such as the Bland-Altman Method and ordinary least products regression can be readily applied in the early childhood context. Understanding agreement (and disagreement) between measurement methods has potential to reduce research costs and improve data quality, with important implications for researchers, clinicians and policy-makers.