I disorders. Fourth, the PDs were assessed at different time periods and may have introduced some bias in the associations with Axis I disorders at Wave 2, although Eaton and colleagues (2011) compared Wave 2 borderline PD with Wave 1 borderline-related constructs found no evidence for bias related to differential follow-up. As a sensitivity analysis, we conducted hierarchical ESEM models separately for Wave 1 and 2 PD assessments and the regression coefficients were similar to those in the full model. Therefore, our findings are robust despite this limitation.
