Each severity or treatment/self-help variable was treated as a dependent variable, and all demographic and diagnostic variables were entered simultaneously as independent variables in each multivariate GEE analysis. When the independent variable was dichotomous and the dependent variable was continuous, the raw mean and standard deviation (M[SD]) of the dependent variable are reported for the sub-group that was negative (labeled “No” in the Table column headers) for each independent variable (i.e., not African-American, unemployed, unmarried, not male; no history of the diagnosis) versus for the sub-group that was positive (labeled “Yes” in the Table column headers) for each independent variable (i.e., African-American, male; employed, married, past history of the diagnosis). When both the independent and dependent variables were dichotomous, the raw proportions of respondents endorsing the dependent variable were compared (and are shown in the two columns at the far right of the Table) for the sub-group that was negative (“No”) for each independent variable (i.e., not African-American, not male; unemployed, unmarried, no history of the diagnosis) versus for the sub-group that was positive for each independent variable (i.e., African-American,
