Abstract
Recent major surveys of the Scottish electorate and of Scottish National Party (SNP) members have revealed a distinct gender gap in support for the party. Men are markedly more likely than women to vote for the SNP and they comprise more than two-thirds of its membership. In this article, we use data from those surveys to test various possible explanations for the disproportionately male support for the SNP. While popular accounts have focused on the gendered appeal of recent leaders and on the party’s fluctuating efforts at achieving gender equality in its parliamentary representation, we find much stronger support for a different explanation. Women are less inclined to support and to join the SNP because they are markedly less supportive of its central objective of independence for Scotland. Since men and women barely differ in their reported national identities, the origins of this gender gap in support for independence presents a puzzle for further research.
Introduction
Researchers exploring the bases of support for political parties have typically looked to social and economic factors, pre-eminently class and religion. However, this approach has enjoyed only limited success when applied to ethno-regionalist parties in general (De Winter and Cachafeiro, 2002) and to the Scottish National Party (SNP) in particular. The nationalist basis of SNP support is unmistakable: those who feel Scottish rather than British and those who share the party’s commitment to independence are disproportionately likely to support the SNP. Yet Scottish identity and support for independence cut across the traditional cleavages, and so analyses of the SNP vote have tended to emphasize heterogeneity, with researchers remarking on the party’s ability to win support across social groups (McCrone, 1992: 164–6; Miller, 1981; Paterson, 2006). Furthermore, the few patterns that did distinguish SNP electoral support, such as relative strength among younger voters and relative weakness among Catholics (Bennie et al., 1997: Ch. 8; Kendrick, 1983), are gradually being eroded. In the 2007 Scottish Parliament election, when the SNP became the largest party in Scotland for the first time, its support was even more than usually drawn from across the social and economic board (Johns et al., 2010).
However, one socio-demographic variable is strongly associated with SNP voting and membership. That variable is gender. 1 There is a marked gender gap in SNP voting. In 2007, the SNP won 35 percent of men’s but only 27 percent of women’s regional votes (Johns et al., 2010: Ch. 2), 2 and similar differences have been observed at most previous elections. Meanwhile, the SNP’s membership is even more disproportionately male than its electoral support. According to a recent survey, only 32 percent of SNP members are women. Both of these gender imbalances are pronounced relative to those found for other parties, both in the UK and elsewhere. While unusually broad in traditional social and economic terms, the SNP’s support is unusually concentrated in terms of gender. The ‘maleness’ of the SNP is especially notable given that the party has historically had a reasonable record in terms of the representation of women – at least compared with other parties in Scotland – among its senior office-holders and in publicly elected posts.
The gender gap in SNP support constitutes something of a puzzle, then, and our purpose here is to work towards solving that puzzle. The empirical bases of the article are two major pieces of survey research: the 2007 Scottish Election Study (SES) and, later that same year, a survey of the entire SNP membership (detailed information about both surveys, and about the variables and measures used for this article, is provided in Appendix A). In the two main sections of the article, we use these data to explore possible reasons for the gender gaps in voting and membership, respectively. The striking consistency of findings across the two surveys allows us, in the concluding section, to offer a straightforward explanation for the party’s gendered appeal.
The gender gap in SNP voting
In 2007, the SNP became the largest party in Scotland in terms of electoral support for the first time. Between the 2003 and 2007 Scottish Parliament elections the SNP gained 10 percentage points on the regional list vote and 9 points in the constituency votes. Beyond the shift in overall vote-shares, another striking difference between those two elections concerns the gender basis of SNP support. The party drew support fairly evenly from both sexes in 2003, but, in stark contrast, there was an 8-point gender gap in 2007: men were markedly more likely to vote for the SNP. In other words, the gains in 2007 were made disproportionately among men. However, when those two elections are seen in broader historical perspective, it is 2003 rather than 2007 that represents the exception: the tendency for the SNP to appeal more to men than women has emerged persistently in studies of the Scottish electorate (Miller, 1981: 147–8; Paterson, 2006). This point is illustrated in Table 1 , which shows the proportion of male and female voters choosing the SNP at various elections – both to the Westminster and Scottish parliaments – since October 1974. 3
Percentage SNP voting by sex, October 1974 to May 2007
Sources: British Election Studies 1974--2005; Scottish Social Attitudes 1999 and 2003; Scottish Election Study 2007.
This gender gap is wide in comparative terms. Although there is considerable variation across countries and across time, gender differences in voting behaviour are typically modest, and in some countries more or less non-existent (Inglehart and Norris, 2000; Jelen et al., 1994; Mayer and Smith, 1985; Studlar et al., 1998). Among exceptions, probably the most prominent is the USA. In the 2008 presidential election, for example, 56 percent of women but only 49 percent of men voted for Barack Obama (Center for American Women and Politics, 2009), a 7-point gap fairly typical of the tendency for males to vote disproportionately Republican and females to vote disproportionately Democratic (Seltzer et al., 1997). The SNP gender gap is therefore comparable in size with the paradigm case in the literature.
Explaining the electoral gender gap
In the absence of previous studies directly addressing this question, we can look to two sources for guidance in explaining the SNP’s electoral gender gap. The first is the recent history of the party. According to leading figures involved in SNP campaigning since devolution, the party made strenuous efforts to enhance its appeal to women voters after survey evidence from the 1999 elections and subsequent opinion polls showed SNP support to be relatively weak in this half of the electorate. There were various aspects to this strategy. One was an emphasis on policy concerns thought to be women’s priorities, such as education and health. In addition, senior women including Nicola Sturgeon and Fiona Hyslop were given more prominent roles. John Swinney, SNP leader in 2003, was less combative than his predecessor Alex Salmond and was keen to adopt a more women-friendly approach to campaigning. The efforts to win female votes eased rather, however, when Salmond resumed the leadership in 2004. Salmond’s priorities differed from those of his predecessor. His placing of Bashir Ahmad second on the SNP party list for the Glasgow region in 2007 all but ensured that the first Asian Scot would enter the Scottish Parliament, but Salmond set no objective for women’s representation. This represents a shift in priorities compared to the run-up to the 1999 election, when Salmond had given strong backing to ‘zipping’ for the party’s regional lists, a policy which would have ensured near gender parity in SNP parliamentary representation. This proposal was rejected at a special party conference in 1998. The result was that the party slipped further away from gender parity in parliamentary representation: women comprised 43 percent of the first cohort of SNP Members of the Scottish Parliament (MSPs), but this had fallen to 25 percent by 2007. The decline reflects not only the absence of mechanisms to ensure equality but also the fact that gender equality in representation slipped down the party’s agenda. On this reading, leadership, party image, perceived issue priorities and the parliamentary representation of women are possible factors determining the gender balance of SNP support. However, this account does not explain the enduring gender gap – why it is that, as leaders and campaigning styles change, the ‘default’ position seems to be a disproportionately male support for the party. Furthermore, this account tends to treat women’s voting behaviour as the phenomenon to be explained (see Miller et al., 1991). Yet, as Table 1 shows, it is the male electorate whose SNP support has fluctuated more.
The second source of guidance is the research literature on gender gaps elsewhere. This has often been couched in terms of the left–right (or liberal–conservative) spectrum, with differences in voting behaviour attributed to gender differences in attitudes or ideology (e.g. Campbell, 2004; Inglehart and Norris, 2000). While the links between sex and such variables as national identity and constitutional preferences have not been explored in the gender gap literature, these debates provide several useful pointers, in particular concerning issue priorities and value differences, which can be applied in the SNP’s case.
Given the quantitative approach taken here, these explanations are operationalized via variables measured in the SES surveys (see Appendix A for details). In order for a given variable to explain the gender gap, it needs to be related both to gender and to vote choice and in the same ‘direction’. For example, although support for Alex Salmond is strongly related to support for the SNP, it will only explain (at least part of) the gender gap if male voters are disproportionately favourable towards the party leader. It cannot explain the gap if the sexes share more or less the same view of Salmond, or if it turns out that women are actually more favourably disposed to him than men. In the latter case, the leadership variable would actually be narrowing or suppressing the gender gap: females would be still less likely to vote SNP were it not for the party’s leader. In the analyses below, we examine whether potential explanatory variables fulfil those statistical conditions and how controlling for them affects the gender gap.
Two different analytic approaches are used. First, the explanatory variables are tested one at a time, in each case measuring the gender gap net of their effect. Second, the variables are added cumulatively, in batches, and in rough causal order based on the ‘funnel of causality’ posited by the Michigan scholars (Campbell et al., 1960) and applied to SNP voting by Miller (1981). The model-building approach gives an indication of whether the gender gap remains when controlling for multiple variables. Since sex is causally prior to all of the explanatory factors investigated here, the task here is to identify those intervening variables through which that relationship operates. In short, what is it about men that makes them more likely (and about women that makes them less likely) to vote for the SNP?
Regardless of analytic approach, the statistical method is the same: logistic regression predicting an SNP regional list vote in 2007. A baseline model was estimated with gender as the single predictor. This produces the odds ratio associated with the basic gender-vote cross-tabulation and hence with the ‘gross’ gender gap of 8.3 points. Re-estimating that model with one or more additional predictors is likely to have at least some effect on the odds ratio for gender. Using the marginals from the original cross-tabulation, we can reproduce the data that would be required to generate this new odds ratio. Calculating percentages from those new data results in the ‘net’ gender gap; that is, the gap controlling for the variable (or variables) added to the model. The basis of the next section of the article is to gauge changes in the gender gap as different control variables are tested.
Testing explanatory variables individually
We begin by testing a series of ‘background’ (i.e. demographic and socio-economic) variables (see Table 2 ). The first of these is age. This is unlikely to have a large impact on the gender gap because the age profile of male and female voters is not very different. Longer female life expectancy does, however, mean that women are disproportionately represented among older voters. In previous years this might have helped to explain the gender gap, because the party’s surge in the 1970s owed much to support from newer voters and the average SNP voter was much younger than the average voter (Kendrick, 1983). However, the age gap in SNP voting had more or less closed by the 1990s (Paterson, 2006) and in 2007 actually reversed (Johns et al., 2010: Ch. 2) such that, like women, SNP supporters are slightly older than the average voter. 4 The effect of age is thus to narrow the gender gap and hence, when that effect is controlled, the gap does indeed widen slightly to 9.0 points. As expected, however, this impact is small. Controlling for the other two background variables, social class and religion, has still less effect on the gender gap. 5 Although both show noticeable variations across the sexes, with men substantially more likely to be in non-manual work and to disclaim any religious affiliation, neither variable proved more than a weak predictor of SNP support in 2007 (Johns et al., 2010).
Effect of controlling individual variables on the gender gap in SNP voting
Next we deal with three variables gathered under the broad heading of political engagement. Neither education nor newspaper readership has more than a negligible effect on the gender gap because, differences across the sexes notwithstanding, neither was more than weakly related to SNP voting in 2007. However, the results with political interest are quite different. The politically interested were substantially more likely to be men and also to vote SNP. This variable appears able to account for around one-third of the gender gap, which narrows noticeably to 5.8 points when political interest is controlled. The gender gap in political interest is a persistent finding and has a variety of causes (as well as being in part an artefact of the standard measures) (Burns et al., 2001: Ch. 4; Campbell and Winters, 2008). Less easily explained is the positive association between interest and SNP voting. However, a comparison reveals it to be a phenomenon specific to 2007: in neither 2003 nor 2005 were SNP voters disproportionately interested in politics. 6 The most likely explanation is that interest is driven in quite large part by success. On this reading, being causally posterior to party choice, interest cannot account for gender differences in SNP voting.
Turning now to the specifically Scottish dimension of SNP support, we first consider national identity. Not only has this long been recognized as a key mobilizer in support for the SNP (Brand et al., 1994; Paterson, 2006), but it might also be thought to have prima facie links with gender given that national identity is often constructed and expressed in typically masculine contexts such as international sport or military history (Cronin and Mayall, 1998; Mayer, 2000). Some scholars (e.g. Stychin, 1998) dismiss such arguments as over-simplistic, however, and the SES evidence supports their scepticism. Women were actually more likely than men (61 percent compared to 56 percent) to claim Scottishness as their primary identity (when asked to place themselves on a scale running from ‘Scottish not British’ to ‘British not Scottish’). The SNP’s particular appeal to men must have some other basis, then, because an appeal based on national identity should have particular attraction for female voters. Indeed, when national identity is controlled, the gender gap widens noticeably to 11.8 points.
Studies have found that national identity has limited direct impact on SNP support (Johns et al., 2010; Paterson et al., 2001: Ch. 5). Rather, strong Scottish identifiers support the party either because such identity generates support for the SNP’s constitutional policies, or even because Scottish identity is a manifestation rather than a cause of support for constitutional change. Before we consider the flagship policy of independence, it is worth noting that the most popular constitutional option in 2007 was further devolution (often referred to as ‘more powers’) – that is, the transfer of additional policy responsibilities, especially in the field of economic and fiscal policy, from the UK Parliament to the Scottish Parliament. Many voters favouring that option voted for the SNP (Johns et al., 2010: Ch. 5). Of course, this could only account for the gender gap if men were appreciably keener than women to see additional powers transferred to the Scottish Parliament. That was the case in 1974, prior to the 1979 devolution referendum when women were more likely to endorse the constitutional status quo. In contrast, in 2007, the ‘more powers’ option was a little more popular among women. As with national identity, the gender gap would have been even wider (more than 10 points) had it not been for women’s rather greater appetite for ‘more powers’. Meanwhile, when asked more specifically about the transfer of fiscal powers men and women shared more or less identical views. Overall, then, attitudes to devolution cannot account for the gender gap in SNP voting.
With attitudes to independence, however, matters are very different. When asked how they would vote in a straight yes/no referendum on independence, 37 percent of males but only 26 percent of females answered ‘yes’. 7 Since these attitudes were the most powerful predictor of SNP voting (Johns et al., 2010: Ch. 10), it is not surprising that they go a long way to explaining the gender gap in support for the party. Indeed, once attitudes to independence are controlled, this difference narrows almost completely (to 0.5 points, neither substantively nor statistically significant). In short, women are less likely to vote for the SNP because they are less likely to support its policy of independence. It is noteworthy that female antipathy to independence did not translate into opposition to the SNP’s promised referendum on independence. Women were slightly more likely to support that proposal. This may indicate that such a referendum was seen more as a safeguard against than as a route to independence. Controlling for support for that referendum, the gender gap was slightly wider (9.9 points), suggesting that the SNP’s showing among women was indeed improved slightly by their pledge to consult the people on independence.
The next set of variables concern the SNP’s issue positions and priorities. First, we analyse respondents’ self-placements on a scale running from substantial cuts to substantial increases in taxes and spending on public services. Among the British electorate as a whole, as Campbell (2006) shows, the average male voter falls closer to the right-wing end of this scale, and the SES results show a similar pattern. However, this ideological gender gap is narrow. Moreover, both men and women voters placed the SNP almost at the exact centre-point of this scale, belying any suggestion that women were turned off the SNP because of their more left-wing stance on fiscal policy. The gender gap of 8.3 points is therefore unaffected by controlling for that variable. On the other hand, the gender gap does narrow somewhat when we turn from issue positions to issue priorities. Respondents were asked an open-ended question about the most important issue to them when deciding how to vote in 2007. Again in line with Campbell’s (2006) findings, women were considerably more likely to mention the core public service issues of education and health. Yet these did not tend to be the top priority issues for SNP voters. Hence, when we control for issue priorities (via a straightforward ‘public services or not’ dichotomous variable), we see the gender gap close to 6.6 points. In other words, male voters responded more positively to the perceived priorities of the SNP.
Beyond the long-standing commitment to independence, the SNP’s most prominent policy position in 2007 was a proposal to replace the council tax with a local income tax. However, since this stance was endorsed by the overwhelming majority of both men and women, controlling for views on local taxation has virtually no impact on the gender gap. The SNP was also strongly critical of Labour at Westminster’s decisions to go into Iraq and to replace the Trident nuclear system, and voters opposed to Iraq and Trident were indeed substantially more likely to vote SNP. In line with numerous studies showing that women are less likely (at least in Western democracies) to support aggressive foreign policies (Eichenberg, 2007; Wilcox et al., 1996), it was female voters who were more likely to share the SNP’s standpoint on these issues. With attitudes to Iraq and Trident controlled, the gender gap widens perceptibly to 11.1 points, implying again that the SNP’s support would have been still more disproportionately male had they taken a different stance on those issues.
The final variable to be considered is evaluations of Alex Salmond, the SNP leader. Here, the SES evidence comes from a series of leadership ratings on an 11-point like–dislike scale. The mean male rating of Salmond was around half a point higher than the mean female rating, a difference which appears more substantial in the light of the general tendency for women to report more positive evaluations. Salmond was the only politician included on the survey to elicit significantly higher ratings from male respondents. The upshot is that, when leader evaluations are controlled, the net gender gap narrows by around one-third. Two points are worth noting about this. First, while females were less positive than males about the SNP leader, they nonetheless rated him more highly in absolute terms than any of the other politicians. The implication is that, insofar as leadership can account for the gender gap in SNP voting, this is because Salmond won support from men rather than losing it among women. Second, leadership evaluations are likely to be causally posterior to some of the factors already considered. For example, it could be that males preferred Alex Salmond because he led a party to which they were already particularly favourably disposed, perhaps because they share the SNP’s preference for independence. In that case, differences in leadership evaluations are a by-product and not a cause of the gender gap under study here. That possibility reinforces the need for a composite model with multivariate controls, as analysed in the next section.
Building a composite model
Our composite model is based on the same variables, analysed in the same order, as in the previous section. The only exception is political interest, which is excluded given the compelling evidence that it is endogenous to party support. The basic causal structure, beginning with fixed demographic characteristics and culminating in short-term factors like leadership evaluations, is well established and plausible. Things are less clear-cut in the middle of the funnel of causality: reciprocal effects cannot be ruled out and different orders could be defended. That said, the key decision – to introduce the ‘Scottish’ variables (national identity and constitutional preferences) before the more general issue and ideological variables – seems justified given the nature of the SNP and its support. Since the focus of interest here is again the effect of gender on vote, we do not report the coefficients or odds ratios for the various control variables, 8 but instead again track the changes in the odds ratio for gender and the implied gender gap. The results are reported in Table 3 , which shows the opening and closing of the gender gap as successive batches of variables are added to the model.
Effect of adding cumulative controls on the gender gap in SNP voting
The broad pattern of results is that which would be expected given the analysis in the previous section. However, the model-building approach here allows us to draw certain conclusions with more confidence. First, demographic and socio-economic variables are actually serving to narrow the gender gap slightly and so, when they are controlled, that gap opens out slightly. It is not because males have different backgrounds and life experiences that they are particularly prone to vote SNP. Nor is it because they feel more Scottish: again, controlling for national identity serves to widen the gender gap perceptibly. The big shift in the graph is triggered by controlling for constitutional preferences. Once again, then, the biggest single reason why men are more likely to vote for the SNP is because they are more likely to favour independence. However, whereas at the bivariate level the effect of controlling for that variable was to eliminate the gender gap almost entirely, in this analysis a 5-point gap remains. There are other factors at play in the gender–vote relationship. The other most notable such factor is attitudes towards the SNP’s leader. Controlling for those evaluations had the potential to reduce the gender gap quite substantially, even with an extensive series of prior variables held constant. Alex Salmond’s particular popularity among men is not simply a consequence but also to some extent a cause of males’ positive attitudes towards his party. The final point to note is that there remains a gender gap (albeit not a statistically significant difference) of around three percentage points even controlling for all of these other factors. Given that this is around two-fifths of the size of the original gap, we can claim only partial success in our attempt to explain what it is about male voters that makes them more likely to support the SNP.
The gender gap in SNP membership
The SNP’s difficulties in attracting support from women are magnified when we turn from voters to party members. Only 31.8 percent of SNP membership survey respondents were women. 9 To some extent this reflects a tendency pervasive across time and across countries for party membership to be less common among women (Lovenduski and Hills, 1981; Widfeldt, 1995). However, as Table 4 shows, the SNP has a very small proportion of women by most yardsticks. This does not appear to be a particularly Scottish phenomenon. There is little comparable data on other parties in Scotland but, where sample sizes permit a meaningful comparison between a party’s membership in Scotland and in Britain as a whole, the proportion of women is not consistently smaller in the former case. It is the SNP figure that is the outlier, being some way further from gender parity than any of the others – Scottish or British – in the table. The purpose of this section is to seek possible explanations for that disparity, and specifically to assess whether it seems to be driven by the same factors that proved important when we examined the electoral gender gap. First, though, we need to say a word about our empirical approach in this section.
Sex breakdown of membership of different parties
Sources: Various party membership studies.
Explaining the membership gender gap
The empirical approach in this section is necessarily different. The election study data included voters for all parties and thus allowed us to identify the factors determining whether or not voters chose the SNP. By contrast, the membership survey does not include non-members (or members of other parties). In considering the relationship between gender and SNP membership, then, we are confined to comparing male and female members of the party. Any inferences drawn from such comparisons depend on the assumption of parallel gender differences among non-members. The previous section is therefore useful not only in highlighting the kinds of factors most likely to explain the gender gap in membership, but also because it provided some baseline information about the relationship between gender and key variables – such as national identity and constitutional preferences – in the electorate as a whole. Considered alone, the survey of SNP members is severely limited in what it can tell us about the determinants of membership. In conjunction with a (roughly contemporaneous) survey of the entire electorate, however, a detailed profile of male and female SNP members becomes more informative. Nonetheless, given that our information about non-members is drawn from other data collections, our evidence about why women may be less likely to join the SNP is also necessarily indirect.
We begin by looking at age among SNP members. At 59 years, the average age of SNP members is quite high, making this – as in the previous section – an unpromising candidate for explaining the gender gap. Demographic differences mean that, other things remaining the same, a more mature support is likely to be a disproportionately female support, and indeed the small gap between the mean female age (60.4) and mean male age (57.9) is similar in size to the corresponding gap in the overall population due to longer female life expectancy. Hence the gender differences in the left-hand panel of Table 5 are expected: the proportions of women are larger in the older age groups. However, if we look only at those who joined since 2005, then a more nuanced pattern emerges that cannot be ascribed solely to demographic patterns. The first point to note is that, among recent joiners, the proportion of females falls to just 28.2 percent. While the SNP has enjoyed increased recruitment in recent years, it has been relatively unsuccessful in attracting women, particularly younger women. Females are especially underrepresented in the younger categories of new members, and overrepresented not in the oldest but in the older middle-aged categories. 10 There is no obvious explanation for this pattern; what is clear is that it does not bode well for any imminent narrowing of the gender gap in membership.
Sex of membership by age group
Another of the background variables considered earlier, religion, looks a more promising potential explanation for the gender gap in SNP membership. In the population as a whole, as noted earlier, men are substantially less likely (by 10 percentage points) to report a religious affiliation. And the proportion of SNP members claiming a religious affiliation (57 percent) was noticeably smaller than the corresponding proportion in the 2001 census in Scotland (67 percent). Evidence from a religion by gender analysis of the membership is inconclusive, however. The gender difference in religious affiliations within the membership is exactly the same 10-point gap as in the population. This leaves open the possibility that greater religiosity among women (partly) explains their under-representation in the SNP. But of course the converse could be true: it may be that the SNP membership is less religious because it contains fewer women. Unfortunately, our analyses cannot clarify causal direction so as to remove that ambiguity.
For obvious reasons, the membership survey respondents were asked not about general interest in politics but about a range of activities and involvements within the party. The gender gaps in these variables were typically modest. Overall, 34 percent of men described themselves as ‘very’ or ‘fairly’ active within the party compared to 31 percent of women. Men were slightly more likely to attend local party meetings regularly and markedly more likely to have stood for elected office within the party. On the other hand, they were less likely to have canvassed voters by telephone or to have attended the party’s national conference. Once again, there are limits to the inferences that can be drawn from these comparisons. Nevertheless, they give the impression that the SNP can call upon active members of both sexes. Gender differences in involvement in the population can probably explain why most parties struggle to attract as many female as male joiners (see Table 4), but they seem unlikely to account for the SNP’s particularly wide gender gap in membership. This conclusion is bolstered by the age profile of the party, which suggests that family commitments – one reason why women might be less likely to join – are likely to have eased for many members. Only 7 percent of SNP members have children below the age of five under their care.
Another possible reason for the membership gender gap is sex differences in levels of political efficacy. This has two broad aspects: internal or self-efficacy, i.e. a feeling of being personally capable of participating, and external or collective efficacy, i.e. the belief that institutions and conditions allow that participation to make a difference. Unsurprisingly, given the upbeat post-election mood in the party, external efficacy scores were high among both men and women. This was true not only on general questions about the amount of influence wielded by members, but also on a specific question about gender parity: hardly anyone of either sex (7 percent of women and 5 percent of men) agreed that ‘the party does not do enough to ensure that equal numbers of men and women are selected as SNP candidates’. Again, then, it seems unlikely that women are put off from joining the party because they see no opportunity to make a contribution or to have a say. However, there are signs – echoing previous findings from surveys of broader electorates (e.g. Vaus and McAllister, 1989) – that women members have a lower sense of internal political efficacy. Women were rather less likely to agree that they ‘could have a real influence’ and markedly less likely to agree that they ‘could do a good job as an SNP councillor’. The fact that these differences persist even among party members tends to arouse suspicions that they may also partly explain why women were less likely to join in the first place. Again, however, we cannot make this argument with great conviction in the absence of precisely comparable data among non-members. In any case, it cannot explain the particularly wide gender gap in the SNP’s membership.
We therefore turn to the ‘Scottish’ variables that are the specific underpinnings of SNP support. As we have seen, there is little difference in terms of national identity between women and men in the electorate. Table 6 shows that this is also true among members, less surprisingly in this case given the strong skew towards the Scottish end of the spectrum. Given that men and women both inside and outside of the party are equally likely to feel Scottish, it seems highly unlikely that national identity can explain the gender gap in membership. There is a potentially important caveat to this. This standard measure of identity is relative rather than absolute: we know that men are no more likely to feel Scottish rather than British, but these data leave open the possibility that they feel their Scottish identity more strongly. We discuss this further in concluding the article.
National identity by sex among SNP members
As among the electorate, the results are very different when we turn from national identity to the constitutional question. Just as female voters are markedly less radical in their constitutional preferences, so too are women members of the SNP, more than one in six of whom cited ‘further devolution short of independence’ as their preferred option. Women were correspondingly less likely to endorse the SNP’s official policy of independence in the EU (though, interestingly, no less likely to favour independence outside the EU). The 9-point gap in support for party policy is more or less identical in size to the corresponding gap within the electorate (whether we compare with the same five-option question as in Table 7 or with a simpler referendum vote intention question). It is also worth emphasizing that, among both members and voters, those women who oppose independence do not instead tend to favour the status quo. The ‘more powers’ option is the one that attracts more female support. Differing appetites for independence thus look like an explanation for the gender gap in SNP membership, just as they did for that same gap in electoral support.
Constitutional preferences by sex among SNP members
The discovery that women, both within the SNP and in the wider electorate, are less keen on independence does not in itself constitute an explanation of the gender gap in the party’s membership. The missing premise is that constitutional preference is a key driver of joining the party. And that point, while plausible, cannot be conclusively demonstrated in the absence of data from those who choose not to join the party. However, there is evidence from an open-ended question in the SNP membership survey asking members to describe the main reasons they joined the party. Independence is easily the most commonly cited reason overall, but women are less likely to be attracted to the party by that policy: 47 percent of men and 39 percent of women cite independence as the reason for joining. Aside from the independence gap that we have observed throughout, another noteworthy gender difference is that women were more likely to report social network reasons for joining, in particular family links with existing members (in several cases their partners). This perhaps indicates that women require a stronger situational impetus to join political parties. However, this tends to explain the gender gap in membership of parties across the board. When it comes to the particularly male profile of SNP membership, attitudes to independence again look the most plausible explanation.
Given the prominence of left–right ideology in previous explanations of partisan gender gaps, it is worth looking briefly at how male and female SNP members placed themselves and the party on a left–right scale. These placements hardly vary. As with the average voter, the average member places the party slightly to the left of the centre of the scale, and there is no difference at all between men and women in this regard. In terms of self-placements, women were marginally only marginally further left (by less than one-tenth of a standard deviation), not enough to account for a large gender gap even if there were compelling evidence that left–right ideology is a prominent motivation for membership.
The final variable considered here is leadership evaluations. Earlier, we found evidence that Alex Salmond’s particular appeal to male voters contributed to explaining the electoral gender gap. It is more difficult to test for such a leadership effect on membership since members of both sexes are likely to evaluate their leader very positively, particularly in the aftermath of the most successful election in the party’s history. In this context, evidence that female SNP members were nonetheless more lukewarm in their endorsement of their leader would constitute persuasive (if still circumstantial) evidence that the leadership was failing to attract women to join the party. We can explore this via a series of questions asking members to rate a number of senior politicians – from both the SNP and other parties – on an 11-point like–dislike scale. Table 8 first reports the mean ratings given by both male and female SNP members to Alex Salmond and Nicola Sturgeon, the leader and deputy leader of the party. There is no evidence that Salmond appealed particularly to men in the party, as his mean rating was slightly higher among females. The gender gap is slightly narrower than with Nicola Sturgeon, but ratings of both politicians conform to the general tendency, indicated by respondents’ overall ratings of male and female SNP politicians, for women members to give slightly higher ratings than men to politicians of both sexes. That tendency – itself hardly marked – is more noticeable than any inclination among females to be particularly generous to female politicians. Alex Salmond may struggle to win electoral support from women and, lacking data from ‘potential’ SNP members, we cannot be sure that he is not partly responsible for the disproportionately male membership of the party. However, the evidence here suggests that he exerts no deterrent effect on those already favourably enough disposed to the party to consider membership.
Mean ratings of politicians by sex among SNP members
Conclusion
Despite having had prominent senior female politicians, the SNP has generally polled better among men and has a disproportionately small proportion of women members. Popular speculation about the reasons for this has focused on two factors: first, fluctuations in the party’s commitment to and record on gender equality in representation; second, on the party’s leadership. We found some evidence that the gender gap in SNP electoral support in 2007 was partly due to the relative popularity of Alex Salmond among male voters. However, our analyses reveal a much more powerful explanation for the gender gaps in both electoral support and membership. The party’s flagship policy is Scottish independence. Support for that option is a strong motivation both for its voters and its members and such support is markedly stronger among men than among women. This goes a long way to solving our puzzle.
Of course, this simply leads to another gender gap puzzle: why are there differences in support for independence? Addressing this question is a further research task in itself. Meanwhile, we will briefly outline three possible answers. The first refers to a point previously noted about the measures of national identity used in this article. Prior studies have shown that feeling Scottish and not British is far from a sufficient condition for favouring independence (Brown et al., 1999; Denver et al., 2000; Paterson et al., 2001). It seems quite likely that support for independence is driven as much by the intensity as by the direction of national identity, and men may feel their ‘Scottishness’ more intensely. A second possibility lies in the gender differences in issue priorities also mentioned above. Economic, foreign and defence policies are the most significant areas for which responsibility would be transferred to Scotland in the event of independence, and these are both more often the concern of men. Since responsibilities for female priority issues, health and education, have already been devolved, women have in a sense less to gain from independence. The third explanation is an application of the recurring finding in social psychology and (a little less consistently) micro-economics that men tend to be less risk-averse than women (Arch, 1993; Byrnes et al., 1999; Eckel and Grossman, 2003). The extent of risk involved in Scottish independence is of course strongly disputed but, insofar as its impact is uncertain, there may be reason to expect women to be warier. Adjudicating between these and other explanations will require detailed and tailored measures of the kind that were not available in the surveys used here. Future research should take a comparative as well as a methodological direction. In similar substantive contexts, the Scottish case has often been compared to several others, most often Quebec and Catalonia. That and other cases offer survey data with which the relationships between gender, national identity and constitutional preferences can be further scrutinized. These relationships await detailed empirical exploration in those contexts, and there is no existing literature on gendered tendencies in nationalist or ethno-regionalist parties into which this article slots neatly. However, it could prove an early contribution to such work.
Our final point concerns framing. Just as gender gaps in voting vary according to the way gender and related issues are presented and discussed at different elections, it seems probable that the gender gap in support for independence will depend on the framing of that policy in party and media discourse. In other words, a greater appeal to men is not intrinsic to the policy of independence but is the result of the way that policy has been defined and discussed. Further research into how and why independence proves less popular among women is therefore of more than academic interest. It could guide the SNP – and the parties opposed to independence – about how the issue should be framed to their maximum advantage.