American Trade Unions and Data Limitations: A New Agenda for Labor Studies

Research on the historical level of union density in the United States is based on data or estimates that represent the sum of union members from different organizations. This results in aggregation bias, where the time-trend in union density is consistent with multiple, divergent trends among organizations. Some unions have experienced membership gains in specific industries or regions with distinct strategies that the analysis of aggregate data misses. No longitudinal data set, based on a random sample of unions, exists. We identify sources for the development of such a data set. Case studies suggest that organizational strategy, financial resources, internal politics, worker attitudes, and competition affect membership; further research on geographic and industry conditions is needed. Purposive sampling, poor understanding of aggregation, and models that do not account for the clustering of unions within larger federations or industries have retarded progress in labor studies.