Two studies, one recently released, shed light on Duke’s ever-contested campus climate: the Greek Culture Initiative’s Report on Gender and Greek Experience and the Report on Gender and the Undergraduate Experience. Today, we focus on the former, which contains some useful insights but many statistical missteps as well.
Out of the flurry of facts and charts laid out by the Report on Gender and Greek Experience, several pieces of information caught our interest. The study includes findings on sense of belonging, perceived respect from others and confidence. First, greek students feel they belong at Duke more than their non-greek counterparts. Although greek women report a higher sense of belonging than independent women and women in selective living groups, the report finds that, overall, there is a drop in a sense of belonging with women of different years, with the senior class reporting a lower sense of belonging than the freshmen class. The study also finds upperclass women feel less respected from their male peers than underclass women.
If these findings are valid, they can substantiate concerns about the female Duke experience. Belonging within the greek system is good, but the report suggests that women’s sense of belonging and perceived respect from others significantly decline over the four years.
However, we have some concerns with the statistical validity of the report itself.
First, it is unclear whether the gendered phenomena mirror a larger societal trend or if they are unique to Duke. To that end, the report should have compared Duke-specific findings to those of our peers, but such comparative statistics were not made available in the final report.
At one point, the report compares Duke’s rates of unwanted sexual contact to national sexual assault rates. However, “unwanted sexual contact” and “sexual assault” are highly different terms that could yield different responses.
We saw no mention of indexes or scales being pilot-tested or borrowed from standard survey-based studies, which usually establish baselines useful for comparative purposes. It is hard to gauge the exceptionality of a Duke-specific statistic if we have no sense of its occurrence in the general population.
Most damningly, some major findings—including those on female belonging and perceived respect—are not based on longitudinal data. Instead of tracking a single class of women throughout four years, the study takes a snapshot in time, polling all four classes currently enrolled in the undergraduate student body. Because longitudinal studies track the same people, the differences observed in the respondents is less likely to be caused by generational differences or sample bias. This report, which is not longitudinal, is weaker as a result.
Ideally, as data is collected in the future by the Greek Culture Initiative—and Duke’s other major surveying bodies—scales, question wording, comparative data and longitudinal data would all be meticulously compiled. Without sound statistical practice, the results—even if true—are nevertheless suspect.
Although the report’s results are believable to some extent, we cannot give them serious weight with the flaws in the statistical analysis. Duke should adopt better practices for future initiatives to produce sound information that can be eventually used to benefit campus culture.