Table 3.

Comparison of the PR of the models with robust variance and the logistic regression model, for the 5 outcomes, adjusting by numerical age

Model		Cocaine consumption	Marijuana consumption	Cigarettes consumption	Alcohol consumption	Risk of consumption of psychoactive substances
Robust Negative Log-Binomial regression	PR IC 95% See	2.789 (0.873,17.003) 0,7152	3.381 (1.890,6.836) 0.3150	2.168 (1.468,3.342) 0.1915	1.245 (1.050,1.478) 0.0361	1.087 (0.931,1.270) 0.0186
Robust Log Poisson – Cox	PR IC 95% See	2.788 (0.879,16.949) 0,7152	3.371 (1.906,6.767) 0.3156	2.186 (1.512,3.313) 0.1900	1.244 (1.094,1.422) 0.0361	1.087 (0.972,1.219) 0.0186
Robust log-binomial regression	PR IC 95% See	2.787 (0.885,16.894) 0.7152	3.361 (1.921,6.694) 0.3164	2.208 (1.563,3.287) 0.1879	1.245 (1.159,1.337) 0.0452	No converge
Unconditional binomial logistic regression	OR IC 95% See	2.831 (0.690,11.538) 0.719	3.624 (1.917,6.848) 0.325	2.513 (1.664,3.794) 0.210	2.830 (2.212,3.620) 0.126	3.495 (2.430,5.027) 0.185
Robust Unconditional binomial logistic regression	OR IC 95% See	2.965 (0.723,12.155) 0.720	3.702 (1.956,7.007) 0.326	2.536 (1.672,3.848) 0.213	2.810 (2.199,3.589) 0.125	3.462 (2.421,4.951) 0.183

See Standard error of estimate