The initial specification of the models predicting marriage converged, and no adjustments were necessary. The initial model specifications of the models predicting divorce did not converge. Troubleshooting revealed that the subject-level random intercept had very low variance, suggesting that additive genetic variance did not make a substantive contribution to the likelihood of divorce in our sample. Convergence was achieved by removing the subject-level random intercept. Nakagawa’s pseudo-R2 are presented for random effects from the generalized linear mixed effects model predicting divorce (Nakagawa et al., 2017).
