In Defence of Liberty

Driven by data; ridden with liberty.

Statistics and Lampposts VIII: Let’s Get Statistical

The University of Bath Students’ Union Officer team for 2014-15 is Jordan Kenny, Paul Goodstadt, Tommy Parker, Freddy Clapson and Ben Jessup. It was noted by a bathimpact editorial the new SU Officer was all-male. A consequent question is whether students display any substantial bias towards a candidate based upon their sex, and if quotas are desirable.

(Video: Bath CTV)

To misquote Olivia Newton-John, let’s get statistical. In the last election, there were four women running for three positions. Even if Officers were chosen randomly from the nominees, an all-male team would occur with a probability of one in nine. It would be a mistake to draw such dramatic conclusions from a single dataset.

Over the past six years, there have been 77 candidates for the five SU Officer positions, from 2009-10 to 2014-15. I am excluding the defunct VP Communications position, which ended in 2010-11. In the 30 roles, students elected 23 male Officers and 7 female Officers. This disparity is largely explained by a discrepancy in nominations: there were 77 candidates; with 55 men and 22 women. 16 of these 30 contests have only had candidates of a single sex: 13 all-male and 3 all-female. The phenomenon of electoral bunching can be seen with the female nominations: 41% of all female candidates run for the Community Officer position, and the only all-female races have been this role. For the other four positions, women only represent 20.3% of all candidates. Four Officers have been elected unopposed.

There have been 77 candidates for the five positions over the past six years. (Visualisation: Datawrapper)

There have been 77 candidates for the five positions over the past six years. (Visualisation: Datawrapper)

The Null Hypothesis

In the 14 contests with candidates from both sexes, ten men and four women elected. There was not an even number of men and women running for every race. Just to scare people who read this newspaper to get away from statistics, we should do a hypothesis test. The null hypothesis is that students show no sex-based bias towards candidates in these independent elections, so the probability of a female candidate being elected is precisely the proportion of female candidates to the total number of nominees. The alternate hypothesis should be accepted if the p-value – the probability that we would see four or fewer female Officers from these fourteen elections if the null hypothesis is true – is less than 0.05. Plugging the numbers into a recursive formula, the p-value turns out to be 0.196. There is insufficient evidence to reject the null hypothesis. Like a true statistician, I require more information.

Quotas have been suggested to ensure greater numbers of female Officers. An output quota would be unfeasible in our current system, where we have five separate roles with their own specialisms and remits. Whilst I maintain that the SU missed out on Kat Agg’s Presidency, how would you choose who to exclude from being elected? Another method might be to prohibit male candidates from running for the Community Officer position, but this would mean missing out on David Howells in 2010-11 and Tommy Parker. This seems like an unsatisfactory solution. There is also a distinction between proper stratification of a sample, and the act of representation: it is the essential nature of representation that people will convey problems they themselves have not experienced.

The prominent concern for the SU is how to get more students – particularly women – to stand. Whilst sex appears to have no significant effect on being elected, there may be many reasons as to why women choose not to stand, including a perception of gender bias amongst voting students. These reasons might actually be positive, as possible candidates may wish to not delay their career. SU President Ellie Hynes wrote on social media: “The problem could in fact be that we need to do a better job of promoting being an SU Officer as a year of work that will enhance your career in the years after.” This is a difficult problem: let’s hope the new team is up to the challenge.

Note: This article was written for and published in bathimpact.


One comment on “Statistics and Lampposts VIII: Let’s Get Statistical

  1. Pingback: Updating Some Analysis | In Defence of Liberty

Comments are closed.


This entry was posted on March 25, 2014 by in Statistics and tagged , .
%d bloggers like this: