I just read both this study* and another written in 2011**, and examined some of their citations. (One*** struck me as more relevant to your direction in addressing the salience of gendered perception).
This is quite interesting. The distancing from and sub-stereotyping of STEM-skilled women is incredibly robust across STEM-disinterested female [and male] subjects, regardless of age or self-perceived gender salience. Girls/women not already interested in STEM almost universally report
decreased likelihood of studying STEM after interaction with an overtly feminine STEM role model, essentially based on the representativeness heuristic. Overtly feminine STEM role models also decrease younger girls' perception of their abilities
regardless of their initial interest in STEM or personal characteristics.
As an aside, no one herein found a correlation between salience of model femininity and girls' positive feelings or perceived similarity with them. i.e., "This argues against the idea that girls connect better with feminine women. It also suggests that participants’ reactions to our role models were not driven by perceived dissimilarity" (Betz).
In short, the feminine role model works for you and I because we're
already interested in STEM (and believe we can succeed). Overtly feminine STEM models can be encouraging, but aren't eye-opening. In fact, they're actually eye-
closing.
*"My Fair Physicist? Feminine Math and Science Role Models Demotivate Young Girls"
Social Psychological and Personality Science, Betz & Sekaquaptewa (2012)
**"Do Female and Male Role Models Who Embody STEM Stereotypes Hinder Women's Anticipated Success in STEM?"
Social Psychological and Personality Science, Cheryan, et al. (2011)
***"Constraints into Preferences: Gender, Status, and Emerging Career Aspirations"
American Sociological Review, Correll (2004)
Note these particular links probably only work if you attend Penn State.