[Single indicator or index? Comparison of measures of social differentiation]

We have examined whether the effects of indicators of social differentiation differ from index measures. The comparisons were performed with analyses on health inequalities. Several types of health impairment (subjective health, diabetes, 7-day prevalence of back pain) were considered in order to draw conclusions on the generalisability of the results. The effects of education, occupational position and income were examined alone and in combined analyses. These were compared with effects obtained by Winkler's index of socioeconomic status. In addition, an index of cumulative disadvantage was introduced. This includes the upper and the lower extreme categories of the three main indicators of social differentiation.

The analyses were based on data of the Federal Health Survey 1998. Only the age groups between 30 and 65 years were considered. The available case number was n=4 692 men and women. Subjective health, diabetes and the 7-day prevalence of back pain were used as dependent variables. All analyses were performed using logistic regression. Differences between effect sizes were tested using the likelihood difference test.

Effects of the three main indicators of social differentiation were obtained in nearly all analyses. This holds both if they had been considered alone or in combination. Use of Winkler's classification produced similar or smaller effect sizes than analyses with single indicators. Marked additive effects of single indicators (cumulative disadvantage) emerged for subjective health and for diabetes, but not for back pain.

The application of index measures may conceal effects of single indicators, thus differential relationships may not be discovered. In studies on health inequalities, indices should not be used at all or only be used in combination with single indicators.