The first question that I’ll try to answer is one that I raised last summer. Do Stanford voters not like female (ASSU) Senators? Could that explain why women have been outnumbered 14:1, 3:2, and now 2:1 in the last three ASSU elections? My earlier post indicated that the answer was no and that, in fact, the low number of female Senators might be more influenced by the propensity of female candidates to drop out of the race. My latest results serve to help build on that theory, finally achieving a basic level of significance.
What do we see?
First, here’s a basic regression across the three years of available data (2009, 2010, 2011) that looks at the likelihood that a candidate will drop out after having gotten at least 100 petitions, but before the actual election. (Note: here, I’m using an OLS regression, which has heteroskedastic errors, but the results are extremely similar using a probit regression (marginal effect of being female is a negative 14.4 percent chance of running), so I’m using these results for ease of discussion.)
The binary variable “female” is now significant at the .1 significance level. This indicates that it is highly probable that female candidates are likely to behave differently from their male counterparts in ASSU races. What do they do differently? They’re about 14.4 percent more likely to drop out of the race. Is this because female candidates are less prepared? Not according to the data. Female candidates who drop out have, on average, 121.2 petitions. Male candidates who drop out have only 116.25 on average. Even the average male candidate who ran averages only 123.5 petitions! According to the model presented above, all else equal, female candidates would need to have almost 43 more petitions to be as equally likely to compete as a male candidate – that’s almost 35 percent more than the average candidate.
What happens to the women who do compete? They do just as well, if not better than, their male opponents.
(Note: the following regression does not use a Heckman selection correction, but, again, the results are extremely similar between simple OLS and the corrected model, so I will use OLS here.)
Here, the value of the “female” variable is not quite significant (this means that the results could be due to chance), but it’s worth noting that it’s approaching significance and that it has a positive value: it definitely does not seem to be the case that girls are suffering at the ballot box, even controlling for their higher petition totals.
What policy prescriptions does this provide, if we’re interested in having more female Senators? Encouraging more women to run and then keeping them in the race. I imagine that this is exactly the sort of role for which the Women’s Coalition was designed, so hopefully we’ll see more support from them for female candidates in order to help shrink this discrepancy.
What other questions can we answer? I’m going to leave it until next time, but the next question on the agenda is this: if SOCC faces a competitive counter-slate (SBS, SUN), does it do worse?