TABLE 2: (Cont) Logistic regression of poor physical health on personal habits and social connections by total sample, gender and average inequality in urban area, Moscow, Russia, Sept. 1991.

only significant odds ratios are reported (p<.001; ^^p<.01; ^p<.05; *p<.09);

Odds Ratios were calculated with Huber White Sandwich robust estimator of variance, adjusted for clustering and all variables in model;

OR indicates the amount of change in physical health for one unit increase in continuous predictor variable or for the level shown if predictor is categorical.