Die Konstruktion eines empirisch bestimmten Sozialschichtindexes mittels optimaler Skalierung am Beispiel von Deutschland

Numerous epidemiological studies have shown an association between social status and disease. In alcohol research, socio-economic status (SES) has been associated with drinking patterns, misuse and alcohol problems. SES is a construct which is usually built by using the indicators education, occupation or income separately or combined as a summary score. However, either approach involves theoretical assumptions about the explanatory value of a chosen indicator before the data have been analysed. In this study we have created a gender-specific metric social status indicator for Germany by using all three single SES-indicators (education, occupation, income). We used national representative data from a postal survey from 2000. The age range was restricted to 25 - 59 years. To construct the indicator we used optimal scaling (categorical principal components analysis). Therefore no theoretical assumptions were necessary about the hierarchical order of educational or occupational categories. The optimal-scaling approach allows variables to be scaled on different levels. We used education and occupation as the original categorical data and income (equivalent income) as continuous data. The result of the scaling is a two-dimensional solution. The first dimension explains that variance which could be described by status consistencies (corresponding levels for education, occupation and income). The second dimension of the indicator results from the low correlation between education and income for some individuals (status inconsistencies). The two-dimensional indicator yields differentiated results which would not be visible using a one-dimensional SES-indicator such as education or a summary score. With regard to drinking patterns there are clear differences between middle or low-educated women or men with higher incomes and other social groups. Middle or low-educated men or women with a high income are more likely to be heavy episodic drinkers than people in other social status groups.