Two series of mediational models (o

Two series of mediational models (one model per dependent variable–suicide ideation and suicide attempt) were conducted in a path analytic framework using the Mplus Version 6.1 (B.O. Muthen & L.K. Muthen, 1998–2011) software. Mplus can accommodate simultaneous modeling of multiple mediators and both continuous and dichotomous outcomes and provides estimates and significance tests of indirect effects using the Sobel test, which is appropriate for both continuous and dichotomous outcomes (Mackinnon, Lockwood, Brown, Wang, & Hoffman, 2007; Mackinnon, Lockwood, Hoffman, West, & Sheets, 2002; Muthen, 2011). For models with either dichotomous or continuous outcome, the same path analytic strategy was conducted, with different estimators. Specifically, for models including suicidal ideation as an outcome, model parameters were estimated by means of maximum likelihood estimation. For models including the dichotomous suicide attempt outcome variable, the models were fit to the data by means of MLR estimation, which provides maximum likelihood parameter estimates with standard errors and a chi-square test statistic that are robust to non-normality. The MLR standard errors are computed using a sandwich estimator. Sobel tests of indirect effects were conducted for both continuous and dichotomous outcomes.