Importantly, through a systematic comparison of CFA models, we found that adjustments for method effects were necessary to achieve adequate model fit. This suggests that scores based on summed PDD items are confounded with method effects/measurement error. Future work should use a CFA approach with method effects adjustments to model PDD. Basic SEM texts describe approaches to adjust for measurement error such as correlating the uniqueness of items with similar wording (i.e. the “correlated uniqueness” approach) or using latent method factors (Brown 2006).
