Controlling for What? Movements, Measures, and Meanings in the US Gender Wage Gap Debate

Abstract

Studies of the political power of economic knowledge have tended to foreground the role of causal claims in the form of grand theories or more narrow findings produced by experimental methods. In contrast, scholars have paid relatively little attention to the role of economic experts' descriptions. This article highlights one category of influential, quantitative descriptive claim: stylized facts. Stylized facts are simple empirical regularities in need of explanation. Focusing on the example of the gender wage gap in the United States, this article showcases how stylized facts travel into political debates, and how the choices made in characterizing an aspect of economic life (such as controlling for full-time work, but little else) interact with social movement activism, and folk understandings of economic life. The gender wage gap was first calculated in the 1950s, but did not take on special importance until the 1960s–1970s, when feminists rallied around the statistic as a useful aggregate measure of women's economic disempowerment. Academics soon followed, and sociologists and economists began to publish studies documenting trends in the gap and trying to account for its sources. The comparable worth movement of the 1980s explicitly argued that the wage gap resulted from occupational segregation and the devaluation of women's work. As that movement faltered in the late 1980s, the gender wage gap became increasingly understood through the lens of women's choices and trade-offs between work and family, and occupational segregation dropped out of the narrative. Throughout this period, the gap was frequently misunderstood or misrepresented as reflecting the narrow sort of same-job, different-pay discrimination made illegal by the 1963 Equal Pay Act, adding confusion to the public debate over women's economic position. These dynamics showcase how technical choices made in the identification of stylized facts, such as statistical controls, are simultaneously deeply political choices.