"SAT" Scores and College Major Gender Composition

Original Post: Randal Olson, Average IQ of Students by College Major and Gender Ratio

Criticism:  The original post compares "IQ" with 'women as a percent of total students' and reaches the conclusion "the more female-dominated a college major is, the lower the average IQ of the students studying in the major".  However, it is likely that the original "IQ/SAT" source is in fact GRE data. Remember, the SAT is taken by college applicants in high school, while the GRE is taken in preparation for graduate school. This casts suspicion on the original conclusion and subsequent discussion of intelligence in the author's comment section/ Reddit submission. 

I hypothesize that the observed negative correlation between female-heavy disciplines and lower IQ (or in this case, GRE scores) is explainable through an undefined preceding event (e.g. self-selection, societal pressure) prior to adoption of a major. Following this, these individuals do not go on to receive advanced mathematical training and tend to score lower on the quantitative portion of the GRE. It's not unreasonable to think that a humanities graduate might not have performed advanced calculus since high school! 

We plot the percentage of women in a given major (W%T Ratio) against the total, verbal, and quantitative GRE scores.


Average GRE Score vs. % Women in a Major
We are able to recreate the same general trend the original author stated: there is a negative correlation between GRE score and the percentage of women in a given major.  We supplement each node with the 'number of individuals in a given major' -- this is reflected in the size of a given node. We note that the top-left quadrant focuses on mathematics, while the bottom-right quadrant leans toward liberal arts.

Educational Testing Services, http://www.statisticbrain.com/iq-estimates-by-intended-college-major/
National Center for Education Statistics, http://nces.ed.gov/programs/digest/d13/tables/dt13_318.30.asp
R-Squared = .61, P Value < .0001


Average Verbal GRE Score vs. % Women in a Major
We investigate the trending seen in the total GRE score by examing the verbal and quantitative components separately.

We note that the distribution in the verbal score appears random. We posit that the differences in GRE score are not explainable by the verbal section (supported by the R-Squared and P Values). 

Educational Testing Services, http://www.statisticbrain.com/iq-estimates-by-intended-college-major/
National Center for Education Statistics, http://nces.ed.gov/programs/digest/d13/tables/dt13_318.30.asp
R-Squared = .039, P Value = .145

Note: Major labels are only visible where possible, all data nodes are present in all three graphs.


Average Quantitative GRE Score vs. % Women in a Major
By digging into the quantitative scores we almost identically reproduce the trend in our first graph: there is a negative correlation between GRE score and the '%women in a major.'

Educational Testing Services, http://www.statisticbrain.com/iq-estimates-by-intended-college-major/
National Center for Education Statistics, http://nces.ed.gov/programs/digest/d13/tables/dt13_318.30.asp
R-Squared = .752, P Value < .0001


Conclusions:

My analysis differs from the original author as I believe that the GRE is undertaken predominately by individuals who have either completed college or have at least finished a substantial portion. As such, you would expect that majors with a heavy focus on mathematics (e.g. Engineering, Sciences) require completion of mathematical courses and subsequently outperform those majors with little-no emphasis on math (e.g. Arts, Education).

There is no statistically significant trend in the verbal scores and the negative correlation observed in 'total GRE' is driven by the quantitative scores. The original author's conclusion "the more female-dominated a college major is, the lower the average IQ of the students studying in the major", is simply not true or supported by the data.  I believe the difference in quantitative scoring is due to the mathematical training individuals enrolled in those majors receive. 

What's most interesting here is that women, assuming a uniform distribution of major choice, are underrepresented in STEM and as such receive less exposure to advanced mathematics during their time at college.


Charting & Data Source: 
Produced in Tableau 8.1 by Justin Flechsig
ETS + Statistic Brain, IQ/"SAT" Score
NCIS, Degree/Sex of Student

Data was joined via Excel's fuzzy logic plugin + manual cleaning.