The ‘gender gap’ in support for independence has become one of the best known features of Scotland’s political landscape in the run up to the referendum. Women have long been less likely than men to support independence. Even when the polls narrow – as in recent weeks they have – women remain less likely to say they will vote Yes. In the most recent YouGov poll that showed a substantial swing to Yes, only 42% of women said they would vote Yes (after leaving aside the undecideds who remain more numerous amongst women) compared with 52% of men, a gender gap of ten points. This is no isolated example, If we look at the most recent poll from each of the six companies that are polling regularly in the referendum, on average the gender gap still stands at nine points.
In talking and writing about the ‘gender gap’ in recent weeks, I have been asked about the evidence on gender gaps in other political contests and, more crucially, whether there is any evidence that these gaps close in the final weeks of campaigning. Two previous votes spark particular interest, for reasons that are probably fairly obvious – first, the 2011 Scottish Parliament election, and second, the 1995 Quebec independence referendum.
In 2011, the SNP were well behind in the opinion polls at the start of the Holyrood campaign – a TNS BMRB poll conducted in late February to early March put the SNP constituency vote among ‘committed voters’ at 29%, well behind Labour’s at 44%. However, following a huge swing from Labour, the SNP went on to take 69 of the Parliament’s 129 seats, forming a majority government under an electoral system designed to make such a feat nigh on impossible. Several commentators have suggested that one factor behind this swing to the SNP was a narrowing gender gap – that women were hostile to the party at the start of the campaign but were persuaded to vote for them by the time they went to the polls on May 5th.
So what did the polls actually show? The picture is far from consistent. The polls appear to disagree both on whether or not there was any gender gap in SNP support at the start of the 2011 campaign, and on whether or not any gap that did exist narrowed over the course of campaign.
Three companies – TNS BMRB, YouGov and IPSOS MORI – conducted Scottish polls near the start of the 2011 campaign and just before polling day (a fourth, Progressive Scottish Opinion, also conducted several polls but as their data are not readily available online I have not included them in this analysis). Two of these companies – IPSOS MORI and TNS BMRB – found a significant gender gap in SNP support in polls conducted just before the campaign started. Women were 8 percentage points less likely to say they would vote SNP according to TNS’s March constituency vote estimates, while in February MORI put the gap at 9 points. However, according to YouGov there was little if any gender gap – a poll in February put the gap between men and women on SNP constituency vote at an insignificant two points, while in their March poll women actually appeared more likely than men to intend to vote SNP.
At the other end of the campaign, the evidence was equally inconsistent. TNS BMRB’s final pre-election poll put the SNP constituency vote at 47% amongst men and 44% amongst women – suggesting that the 8 point gap they previously found had indeed narrowed. However, MORI’s final pre-election poll again found a 9 percentage point difference in the proportion of men and women who planned to give their constituency vote to the SNP (50% vs 41%). YouGov, meanwhile, put SNP constituency support at 44% amongst men and 40% amongst women – a 4 point gap which, while hardly striking was, if anything, slightly wider than the gap they found two months earlier. (The picture on regional vote is broadly similar, though the gender gap on MORI’s regional vote did fall slightly, from 8 points in February to 6 points in May.)
So based on this evidence, it is not clear that there was a narrowing gender gap in support for the SNP over the weeks of the 2011 Scottish Parliament election campaign. While findings from two out of three polling companies indicate that there was a significant gender gap in SNP support at the start of the campaign, only one out of three suggests that this gap had significantly shrunk by the end of it. Of course, some readers will undoubtedly point out that the 2011 elections were hardly the finest hour for polling companies, all of whom underestimated SNP support on the regional list vote. However, their constituency vote predictions were reasonably accurate – in their final pre-election polls, all three companies put SNP support at 45% once those unlikely to vote or who did not know how they would vote were excluded – very close to the actual SNP share of 45.4%.
Data on the demographic profile of voters in the 1995 Quebec referendum is less readily available. That event took place in the early days of the internet, when polling tables were not routinely archived online. However, two polls conducted by Leger and Leger – one in late September and one between 23rd and 25th October, just prior to the vote, which took place on 30th October 1995 – both found that women were less likely than men to vote ‘Oui’. Although the proportion of both men and women voting in favour of Quebec becoming independent rose between the two polls, the gender gap was still 7 percentage points in late October, very similar to the 8 point gap that existed in late September. So according to those polls at least, the gender gap was still in evidence just before polling day and may have been a crucial factor in the extremely narrow victory the No side eventually achieved (50.58% voted No, 49.42% Yes, on a 94% turnout).
Of course, neither the Quebec referendum nor the result of the 2011 Scottish Parliament elections can tell us what will happen on the 18th September. The fact that there is no clear evidence that the gender gap closed in these other cases does not tell us that the gap in men and women’s support for a Yes vote will not close in the next fortnight. That the referendum is without precedent is what makes predicting the outcome highly uncertain. But findings from Quebec in particular arguably reinforce the view that whether or not the gender gap can be closed in the final weeks of campaigning may well prove key to the final outcome.