Driven by data; ridden with liberty.
Members of the University of Bath Students’ Union voted for their full-time officer team for the next academic year, 2015-16. To raucous applause in the main bar, the officer-elects were: Jordan ‘JK’ Kenny (President), Lucy Woodcock (Education), Matt Humberstone (Community), Wadzi Pasipamire (Activities) and Holly Clemens (Sport). Congratulations to all candidates for their competent and respectful campaigns.
In light of this result, I have updated the analysis on the gender balance of University of Bath Students’ Union officers. This analysis was induced by the election of an all-male officer team in the previous round of contests. (Please note: this analysis contained a miscalculation which is corrected in both the visualisation and the below discussion. This did not perturb the conclusion, and the full report will highlight this mistake. I sincerely apologise for my error.)
Using Tableau Public, the key statistics have been visualised in a storyboard:
The question arose if the number of women elected to these positions is significantly beneath the expectations of sortition. The null hypothesis is that there is no gender preference among the members of the University of Bath Students’ Union. The alternate hypothesis is that the voters have a preference for male candidates. The hypothesis test will take place at 90% confidence.
In the elections from 2009-10 to 2014-15, there were 13 electoral contests with candidates of both genders. In these elections, there were 16 women and 25 men. From these 13 elections, four were won by women. As each election must yield a winner, it forms a Bernoulli trial, with differing probabilities based on the number of men and women running. The sum of distinct Bernoulli trials with differing probabilities forms a Poisson Binomial distribution, which may be calculated through a recursive formula. The p-value is the probability that we would observe data that is at least as extreme as what actually occurred, given the null hypothesis is assumed to be true. In this case, it is the probability we would see four or fewer women triumph in these 13 elections. That p-value is 0.2773, which means there is insufficient evidence to reject the null hypothesis at 90% confidence (or even at 80% confidence).
In the five elections for the five SU Officer positions this year, there were candidates of both genders. As three women won these elections, the probability that captures our interest is the probability that seven or fewer women would win through sortition. Now, the p-value is 0.4071. Again, this means we fail to reject the null hypothesis at 90% confidence. It should be noted the null hypothesis would not be rejected at 60% confidence. The consequence is that the alternate hypothesis does little to explain the observed data.
Even in student politics, data beats speculation.