The History of Women and the History of Statistics

Abstract

Like other fields of social history, women's history needs statistics—both in the sense of statistical data and in the sense of a methodology to arrange and analyze that data. Women's history as a field needs because can answer crucial questions about the character and scope of changes in women's Uves over time and across place. Statistics provide information about aggregates, distributions, rates, and correlates that are absolutely essential to the corpus of women's history knowledge and, from my reading, that can be discovered no other way. Even historians who write case studies, traditional historical narrative, and biography impUcitly raise questions of context and representativeness that can only be answered quantitatively. Statistical knowledge, in other words, makes up a necessary though not a sufficient form of women's history knowledge. And despite a somewhat weightier tradition of quaUtative analysis within women's history, quantitative, social scientific, or statistical treatments of the situation of women in the past have prolif erated as the field has grown.1 Yet the marriage, so to speak, between quantitative or social science history and women's history—my apologies to Heidi Hartmann here—has not always been a particularly happy one. Women's historians sensed that statistics as a field posed problems for women's history, and they were quick to discover the biases, omissions, and misrepresentations in histori- cal data and in classification schemes and constructs that form the canon of statistical method.2 Several themes recur in these critiques. Existing historical statistical series tend to contain less information on women than that available for men. Sometimes, fewer data were coUected on women in the first place; other times, they were simply not reported. Second, data on women can suffer from undercounts. In surveys, interviewers sometimes did a worse job of collecting data from women. Third, classification schemes for record- ing and reporting information on women contain biases. Agency statisti- cians processing data sometimes excluded women from certain reporting categories—for example occupational Ustings—even if women Usted the information on the data collection form. OveraU, age cohort or occupa- tional data might be reported in less detaU than were sirrular data for men. FinaUy, conceptions of appropriate household organization and respect- able family Ufe Umited the possible range of household and family situa- tions that would be recorded and reported about women. Much of this