I find the concept of feminization—how the presence or predominance of women in certain roles or occupations affect those roles and occupations, their legitimacy, compensation, etc.—to be fascinating and directly affect areas I work in (nonprofits, service, social media). Below is the abstract from a paper presented to the American Sociological Association by Paula England, Paul Allison, Yuxiao Wu, and Mary Ross entitled “Does Bad Pay Cause Occupations to Feminize, Does Feminization Reduce Pay, and How Can We Tell with Longitudinal Data?” (2004):

Predominantly female occupations pay less than “male” jobs, even after adjusting for skill demands. The devaluation perspective sees sex composition to affect wages; it says that gender bias affects employers’ decisions about the relative pay of “male” and “female” jobs. The queuing or relative-attractiveness view sees occupations’ sex composition to be affected by their reward level, with less attractive jobs going to women because employers prefer men and can get them in jobs that pay well. Past longitudinal research on how changes in occupations pay and sex composition are related has employed the cross-lagged panel (lagged-Y- regressor) model, generally finding support for the devaluation but not the queuing/relative attractiveness view. We argue that a stronger statistical approach to assessing causal dynamics is a fixed-effects model with lagged independent variables. Using CPS data from 1983 to 2001, we test these two perspectives. We find support for neither idea. That is, generally, the feminization of occupations does not lower their wages, and a fall in occupations’ relative wages does not lead to feminization. We conclude that in earlier historical processes, as occupations and organizations originate, there was a causal relationship between pay and sex composition, but that the continuing relationship is due to institutional inertia freezing in that early relationship, rather than to ongoing causal dynamics.

Emphasis mine.