Except where noted (footnote to Table II supp.), the formal test of equal regression slopes for each threshold on the ordinal scales could not be rejected and confirmed the validity of the ordinal models. We report the expβ or odds ratio (the odds of increasing one unit in the outcome variable given a one unit increase in the fitted term when all other fitted terms are held constant) and significance (two-tailed) according to the Wald statistic. The interdependence between twin pairs was ignored in the ordinal regressions, yielding anti-conservative tests of main effects and interactions.
