David Jeffery is a lecturer in politics at the University of Liverpool. This piece is based on an article co-authored with Tim Heppell, an Associate Professor of British Politics at the University of Leeds; Richard Hayton, an Associate Professor of Politics at the University of Leeds, and Andrew Crines, a lecturer in British Politics at the University of Liverpool.

Excitement isn’t usualy the first emotion which springs to mind when the Conservative Party is mentioned, but its leadership election last year was anything but ordinary. Forced by the resignation of David Cameron just 13 months after he had won the first Conservative majority since 1992, the contest featured sexting scandals (courtesy of Stephen Crabb), betrayal (with Michael Gove’s spectacular knifing of Boris Johnson, who had for so long been the prince over the water in London’s City Hall), and a very un-Conservative (and not to mention embarrassing) march on parliament by supporters of Andrea Leadsom.

An element of this election which was not exciting, however, was how MPs would vote. There was a general expectation that this would be a ‘Brexit election’, with those MPs who backed Remain broadly supporting the eventual winner, Theresa May (and this was indeed the last election she did win), whilst Leadsom or Gove would win over MPs who supported Leave.

We wanted to test this conventional wisdom and examine whether there were other effects, apart from Brexit, which influenced how MPs voted in the leadership election – specifically in the second round of the contest, between May, Leadsom, and Gove. To do this, we used a statistical technique called multinomial logistic regression, which allows us to examine the influence of a variable on a selected outcome: for example, whether an MP’s position on the EU referendum had a statistically significant effect on whether he or she voted for a certain candidate or not, controlling for the influence of other variables, such as gender.

We included a range of social, political, and ideological factors in our model which we believed may have influenced the voting behaviour of Tory MPs in the leadership election.

In terms of social variables, we included gender on the basis that May’s work with the Women2Win group might have earned her some goodwill amongst female MPs, and also education. As Philip Cowley and John Garry found in their study of the Conservatives’ 1990 leadership election, Oxbridge-educated MPs backed Michael Heseltine (Oxford) or Douglas Hurd (Cambridge) in the 1990 leadership election, whilst those who went to non-Oxbridge institutions, or didn’t attend university at all, were more likely to support Major.

Previous studies have also shown the importance of Commons experience, with longer-serving MPs having been less likely to vote for Major or Heseltine in 1990. Constituency marginality is regularly hypothesised to be important, with MPs in more marginal constituencies more likely to support the “safer pair of hands”. Thirdly, holding ministerial office was shown to be important in the 1975 leadership election, with ministers more likely to vote for Heath and backbenchers more likely to support Thatcher (according to Cowley and Matthew Bailey).

We also added a new political variable: UKIP’s vote share in the 2015 general election. We hypothesised that those MPs from constituencies with higher UKIP vote shares would feel Nigel Farage breathing down their necks (figuratively, one would hope) and, fearing being outflanked on their right, would therefore be more likely to support a Leave-backing candidate.

Finally, we considered ideological variables. We used same-sex marriage as a proxy for social conservatism, and posited that social conservatives would be more likely to vote for Leadsom, who, unlike Gove and May, had not supported the idea of same-sex unions. Our final variable was MPs’ positions on Brexit. Clearly, we would expect this to reflect the received wisdom of the time, with Remain supporters voting for May and Leave voters being more likely to support Leadsom or Gove.

Our results did indeed confirm the prevailing assumption: MPs who voted to remain in the EU, or were undeclared, were more likely to vote for May than not (although that is not to say May did not also receive support from some Leave-voting MPs – for example, from her campaign manager Chris Grayling). Interestingly, we found that none of the other variables had a statistically significant influence on the likelihood of voting for May.

Contrastingly, we did see a cleavage emerge between support for Leadsom and for Gove. Although both candidates did enjoy an increased likelihood of support from Leave MPs, relative to those who voted Remain, there was a divergence vis-à-vis social conservatism: those who voted against same-sex marriage were more likely to support Leadsom than those who voted for the legislation, whilst the opposite was true for Gove. Leadsom also received greater support from backbenchers than from ministers: most likely because those with socially conservative views were unsuited to Cameron’s modernising agenda, and thus found it harder to move into ministerial office.

The only other statistically significant relationship to emerge from our model was that the newer MPs – those elected after 2015 – were less likely to support Leadsom compared to those elected before 1997. This could be because those who became MPs under Cameron’s leadership were inspired by his modernisation ideas, andthe social conservatism of Leadsom would thus be unappealing to these parliamentarians.

So, we see three key camps emerge in terms of voting behaviour in the 2016 Conservative leadership election: Remain-supporting MPs plumped for May, socially-liberal Leavers voted for Gove and socially-conservative Leavers voted for Leadsom. This was beneficial for May, since not only were there more Remain-backing MPs than Leave-supporting MPs in the Conservative Party, but her rivals could not draw on a unified Leave base of support.

Our article ‘The Conservative Party Leadership Election of 2016: An Analysis of the Voting Motivations of Conservative Parliamentarians’ is free to read for the next three months.