A psychometric study of the Emotional Availability Scales: Construct validity and measurement invariance between depressed and nondepressed mother–infant dyads.

The Emotional Availability Scales (EAS) are the most widely reported observational assessment measure of parent–child relationships and has been of particular interest in understanding differences between samples of depressed and nondepressed mothers and their offspring. Despite its widespread use, psychometric validation of the factor structure in normative samples and the measurement of invariance within clinical samples has not been published. We evaluated the internal structure (dimensionality, reliability, convergent, and discriminant validity) of the EAS fourth edition using a nondepressed sample of 157 Australian women and their infants aged 6 months, including testing the measurement invariance of the EAS between the same nondepressed sample (n = 157), and a depressed group (n = 185) of mother–infant dyads, using MPlus. Participants were recruited from tertiary hospitals, and depression status was established using a diagnostic measure. Higher-order confirmatory factor analyses on the EAS’ six dimensions supported a unidimensional factor solution in our data. Full measurement invariance was not demonstrated due to metric noninvariance of the maternal nonintrusiveness and child responsiveness dimensions. Full scalar invariance supported mean comparisons, and a medium effect of .78SD lower mean emotional availability for the depressed group was found; Cohen’s d = .63, 95% CI [.41, .85]. While arguments exist for the clinical utility of differentiating between multiple dimensions of emotional availability, the current findings do not support a multidimensional factor structure or full multigroup measurement invariance of the EAS. Similar psychometric investigations of the EAS in clinical and nonclinical samples are needed.