Social Generalizations: Biases and Solutions.

The author conducted a cross national analysis of sociological research reported in leading journals at two points in time over a ten year period to determine if sociologists' ability to produce valid social generalizations had improved significantly over the recent past. The official journals of the United States (American Sociological Review), Canada (Canadian Review of Sociology and Anthropology), and Great Britain (Sociology) were selected for examination. Also included in the analysis were two other highly reputed journals--the American Journal of Sociology and the British Journal of Sociology. Only articles reporting the results of empirical research were included in the comparative analysis. All of the research reports contained in the five journals were examined to determine the research design employed, the type of sample or population investigated, and the method used to collect the data. Results include the following. Particularly in the American journals, but also in the Canadian Review, the analysis revealed changes over the past ten years in the methodological characteristics of published studies such that sociologists' ability to make valid social generalizations may be improved substantially. By taking readings at several points in time, selecting a broadly based sample, and not relying exclusively upon one or two methods of data collection, it is extremely likely that the generalizations produced are less affected by measurement artifacts and instead reflect effects of relevant factors as they impinge upon the variables sociologists seek to explain. (Author/RM)