Diverging electoral fortunes in Scotland and Wales: national identities, national interests, and voting behavior

ABSTRACT Despite an extraordinary degree of political turbulence in the UK, the 2021 Scottish and Welsh election results were remarkably similar to those recorded at the previous elections in 2016. While this period spanned the 2016 EU referendum, Britain’s exit from the EU, and the coronavirus pandemic, these upheavals appear to have had little impact on devolved election results. From a comparative perspective, however, such continuity only underlines the extent to which these nations’ political trajectories have diverged since the establishment of devolution in 1999. Using individual-level survey data from twelve election studies over two decades, we show how changing patterns in the relationship between national identity and party support have driven these differing trajectories. In doing so, we provide the first comparative analysis of voting over time in both countries. Additionally, we show how national identity helps to sustain single-party dominance in Scotland and Wales.


Introduction
The story of the sixth devolved elections in Scotland and Wales was one of continuity. In Scotland, the Scottish National Party (SNP) was returned as the largest party for the fourth consecutive election since 2007, with vote and seat shares nearly identical to those gained at the previous election in 2016 (Johns 2021). In Wales, Labour added to their unbroken record of success at devolved elections, having been returned as the largest party at every election since the establishment of the country's devolved legislature in 1999 ). Yet while there may have been striking continuity between the 2016 and 2021, viewed from a longer-term perspective, voting behavior in Scotland and Wales has evolved substantially in the near quartercentury since voters were first asked to choose representatives for the Scottish Parliament and then National Assembly for Walessince 2020, the Senedd or Welsh Parliament.
In 1999, the Labour partyat that point the dominant party in both Scotland and Wales, and still basking in the glow of Tony Blair's sweeping victory at the 1997 UK general electionemerged victorious after facing strong challenges from two (sister) nationalist parties, the Scottish National Party and Plaid Cymru (with the latter, in fact, outperforming the former on the regional list element of the respective countries' voting system). 1 From this very similar starting point, electoral fates have since diverged considerably. By the third set of devolved elections in 2007, Labour in Scotland was being out-polled by the SNP. In 2011 that party consolidated its position by winning a majority of seats, and has since emerged victorious from every local, Scottish, UK and European election contested in Scotland, due in part to a realignment precipitated by the 2014 independence referendum (Henderson et al. 2022, 107).
By contrast, while the SNP was supplanting Labour as Scotland's dominant party-political forcecannibalizing much of their competitor's voter base along the waythe Plaid Cymru tide was receding. After what was termed the 'quiet earthquake' of 1999, Labour regrouped and succeeded in reinforcing its long-standing hegemony over Welsh politics (Trystan, Scully, and Wyn Jones 2003). 2 All this despite consistent support among the electorate for further devolution of powers to Wales and, more latterly, even increased interest in and support for independence (Griffiths 2021) that is, for the apparent raison d'être of Plaid Cymru.
To explain these diverging trajectories, we conduct the first comparative, over-time analysis of voting in both territories. After examining the latest Scottish and Welsh election results and changes since 1999 utilizing data from the Scottish Social Attitudes survey, Scottish Election Study and Welsh Election Study data (SSA, SES and WES respectively), we theorize the electoral divergence between them with reference to the role of national identity in shaping political dispositions and vote choice. We focus on the role of national identities in shaping electoral outcomes because such influences have long been recognized as important in the vote choice literature, especially in those states in which competition is organized (at least in part) by a centre-periphery axis (see Schildkraut 2014;. While the degree to which national identity shapes vote choice has been contested (Curtice 2009;2011;Johns et al. 2009;Wyn Jones 2012though cf. Wyn Jones 2022), there is evidence of strong associations between national identities and vote choice in recent elections and referendums held in Scotland and Wales (Henderson et al. 2021;Larner et al. 2022).
We present two key arguments in this paper. First, we demonstrate that the diverging electoral performances of Labour and the nationalist parties in Scotland and Walesoften remarked upon in popular commentaryis integrally connected with the role played by national identity in the two territories. In Scotland, vote choice has become far more polarized on national identity grounds than was the case in 1999, with those who emphasize a sense of Scottish identity now overwhelmingly likely to support the SNP. The position is rather different in Wales. While national identity discriminates very clearly between those who vote Conservative or Plaid Cymru, unlike its Scottish counterpart, Welsh Labour continues to attract support from across the identity spectrum.
Second, and relatedly, we demonstrate that the diverging relationship between national identity and vote choice in both territories associates with popular perceptions of whether the major parties can defend Scottish/Welsh interests. Evidence from the respective election studies show that Scottish Labour have struggled on this measure, especially in the wake of the 2014 Scottish independence referendum, whereas Welsh Labour are still able to convincingly portray themselves as a party that stands up for Wales. Taken together, our findings further clarify the role and importance of national identity in influencing vote choice in historic substate nations in Europe such as Scotland and Wales, comparing data and honing explanatory theory over space and time.

The 2021 Scottish and Welsh elections
Ostensibly at least, very little changed as a result of the 2021 devolved elections. The SNP and Welsh Labour continued in government, and few individual seats changed hands. This lack of change is perhaps surprising given the tumultuous period between the two elections, which spanned the divisive 2016 EU referendum, Britain's resulting exit from the EU, the coronavirus pandemic, and a host of additional nation-and issue-specific challenges. 3 It might have been reasonable to assume that heightened public attention to devolved government actions during the pandemic, for example, might have prompted polarized views of their performance, and that such preferences (on an issue largely irrelevant to national identity) might in turn influence vote choice in 2021. This was not to be the case. As others have discussed, the results in Scotland (Johns 2021;McMillan and Henderson 2021) and in Wales (Awan- Scully 2021;Larner et al. 2022) showed little change in support for the governing parties as well as for opposition parties. In Scotland, because 'no party vote share changed by as much as two percentage points' (Johns 2021, 493), the SNP continued to dominate both the constituency and regional ballots (see Table 1), winning 62 of the 73 constituency seats 4 and increasing their seat share by one to return 64 representatives (one short of an overall majority). Continuity was evident across the board, as the Scottish Conservatives returned as the largest opposition party (31 seats), followed by Scottish Labour (22 seats).
The story was similar in Wales. (see Table 2). Like the SNP, Welsh Labour also gained one seat, winning half of the 60 seats available in the Senedd. While the Welsh Conservatives had their most successful devolved election ever with a gain of five seats (to 16), this was based on a vote share that represented only a marginal improvement on their 2011 performance ). These seat gains were a consequence of the total collapse of the UK Independence Party (UKIP)a party that had performed strongly in the 2016 election held only a few weeks before the Brexit referendum. Plaid Cymru saw a net gain of one seat (to 13), but failed to increase their share of the vote.
How has electoral behavior changed in Scotland and Wales since 1999?
From a longer-term perspective, however, there has clearly been significant change since 1999. Figure 1 plots the vote shares of the three largest parties in Scotland and Wales (Labour, Conservative, nationalist) on the constituency and regional list ballots between the first and the most recent devolved elections. Focusing first on the diverging fortunes of Labour  party, the party's Welsh branch has continued to dominate in Wales even as election-specific fluctuations have determined whether it governs alone or as the largest party in a formal coalition or some alternative arrangement. By contrast, between 1999 and 2021 Scottish Labour's vote share in both ballots declined at every election, at first gradually but thenin the wake of the 2014 Scottish independence referendumfar more precipitously (see Henderson et al. 2022, 124). Relatedly, the nationalist parties in Scotland and Wales have also enjoyed very different electoral fortunes, despite (as noted) achieving a very similar vote share at the first devolved elections in 1999. After a weaker showing in 2003, the SNP increased their vote share in 2007 (after which they formed a minority government) and 2011 (where they won a majority). The party drew further ahead in 2016 and 2021, continuing to serve in government as a result. Plaid Cymru, however, have been unable to match the high watermark of 1999 and have instead consistently attracted around 20 percent of the vote in both the constituency and regional ballots at each election since 2003.
As for the Scottish and Welsh Conservatives, while they are currently the second largest parties in both legislatures, they have arrived at this situation via rather different routes. In Wales, the Conservatives have increased their constituency and regional vote share election-on-election, with the sole exception of 2016 where the intervention of UKIP served to undercut their support (Scully and Larner 2017). This contrasts with the performance of the Scottish Conservatives which remained largely unchanged across the  Scotland andWales (1999-2021) first four devolved elections, before leapfrogging Labour into second place in 2016 election (that is, in the aftermath of the independence referendum)a position they then retained in 2021.
How does existing scholarship explain electoral behavior in Scotland and Wales?
Despite the relative infancy of Wales' devolved institutions, a growing scholarship is cataloging and exploring patterns of voting at each devolved election in Scotland and Wales. 5 This work has been facilitated by election studies that provide individual-level survey data to help scholars explain outcomes at these elections. Despite this apparent depth, however, theories of voting behavior that seek to explain patterns of support over multiple election cycles are much less developed. Seeking to close this theoretical gap, scholars have repurposed theories such as 'second-order elections' from other multilevel electoral contexts or have proposed innovative new explanations from close analyses of time-series data in each of the case countries 'Second-order elections' (SOE)a theory that was first proposed to explain voting behaviour in European parliamentary electionssurmises that certain electoral contests should be perceived by voters as less-important ballots than first-order, state-wide elections. Such second-order contests should be associated with lower turnout, higher levels of protest voting, and outcomes which are largely decided by the relative electoral fortunes of parties in the first-order arena of political competition (Oppenhuis, van der Eijk, and Frankline 1996;Reif and Schmitt 1980;Rohrschneider and Clark 2007;Schmitt and Teperoglou 2017).
Yet although SOE is a useful model of vote choice in the European parliamentary context for which it was first developed, applying this theory to sub-state electoral contexts is not straightforward. Put simply, not all 'regions' are created equal. Not only is it likely that the explanatory power of secondorder explanations weakens as the level of sub-state autonomy increases (Jeffery and Hough 2009), but there is now widespread international evidence that electoral outcomes do not adhere to the model's foundational assumption that voters in such contests should vote to punish the governing parties at the state-wide level (Schakel and Jeffery 2013). In Scotland, the relative performances of Scottish Labour and the SNP since the Conservatives took office at the UK level would not be predicted by the SOE model, and there is increasing evidence that voters can engage in separate territorial decision-making calculi in Scottish and Welsh elections (Curtice 2009;2011;Scully 2013;Trystan, Scully, and Wyn Jones 2003;Wyn Jones and Scully 2006). There is even evidence that a sizeable proportion of the Scottish electorate now view devolved elections as 'first-order' and UK parliamentary elections as the 'second-order' contest .
Alternative explanations to 'second-order elections' focus on socio-political cleavages to explain vote choice in Scotland and Wales. The earliest of these approaches were greatly influenced by the theories that identified social class as the dominant driver of political behavior in Britain (for a summary, see Larner 2019). Borrowing heavily from the 'Michigan model' developed in the 1950s and 1960s, early 'organic' explanations of vote choice in Scotland and Wales focused in particular on the relative numbers of working-and middle-class voters in Scotland and Wales compared with the rest of Britain (Blondel 1963;Butler and Stokes 1969). Yet class-based approaches struggle not only to explain support for the SNP and Plaid Cymru, parties that fit uneasily into the two-party structure of class-based politics, but also cannot explain decades of one-party electoral dominance by Labour in Wales.
Later work recognized this issue and incorporated political identities other than social class into voting behavior models, with some explanatory success. Scottish vote choice models began to incorporate religious and territorial identities (Agnew 1984;Bennie, Brand, and Mitchell 1997;Bochel and Denver 1970). 6 And in Wales, Balsom, Madgwick, and Van Mechelen (1983)'s Three Wales Modela harbinger of socio-political models in the twenty-first centurytheorized that support for Plaid Cymru and the Labour party's dominance could only be understood in reference to voters' sense of British and Welsh identity and strength of their 'cultural attachment' (Wyn Jones 2022).
After the first sets of Scottish and Welsh devolved elections, these emerging identity-based explanations were challenged by an apparent consensus that perceived government performance was the central factor influencing vote choice, an approach known as the 'valence model' (Clarke et al. 2004 see for example Curtice 2009;2011;Johns et al. 2009;Scully and Wyn Jones 2012). This model had explanatory power in statistical modeling and appeared to align with efforts by political parties to demonstrate 'competence' in office. Indeed, governing competence was emphasized in the 2007 elections by the SNP and Plaid Cymru as a means to avoid alienating voters who might not have otherwise aligned on the constitutional question (Bradbury and Andrews 2010;Johns et al. 2009). Yet recent studies note the potential limitations of this approach. 7 In particular, competence evaluations appear to be not independent of a voter's partisan and national identities (Henderson and McMillan 2022;León and Orriols 2019;Tilley and Hobolt 2011). Rather, voters' understandings of governing competence appear to strongly correlate with their pre-existing political party preferences and stated national identities. Such alignment has been found in voter evaluations of how governments handled the coronavirus pandemic in England, Scotland, and Wales (Griffiths and Larner 2021).
These developments in the literature have been aided by growing crossfertilization in the comparative literature that is returning to socio-political and identity cleavages as a critically important motivator of vote choice. Particularly in 'historic nations' of Europe that are today constituent parts of other nation states, scholars have focused on the role of 'statewide' and 'substate' national identities in shaping vote choice; identities which may interweave and overlap. National identities should have a particularly powerful role in influencing vote choice because of their connection with voters' memberships of groups in societies. At a basic level, individuals belong to many different social groups and may claim (or be assigned) membership of some of these groups (Chandra 2012), and group membership tends to foster a sense of similarity and therefore attachment (Hechter 1975). In turn, such attachments may encourage individuals to favor policies and political actors that aim to protect their own group's identities and goals. The nation is often one of the most important of such groups, and is one that individuals in Scotland and Wales regularly cite as being among the most important to their sense of self (McCrone and Bechhofer 2015). Thus, if national identities matter to voters and are salient to both their self-conception and the way they navigate their political worlds, then it follows that they may take such identities into account when voting.
Evidence for group-and identity-based understandings of vote choice have been found in the many election-specific studies of Scottish and Welsh elections. In Scotland, for example, even before the first devolved elections, scholars have found a link between voting for the SNP and Labour and a sense of Scottish identity on the one hand, and Conservative voting and British identity on the other (e.g. Brand et al. 1993). In Wales, analyses of the 2021 Senedd election have also re-emphasized the importance of national identity as a key influence of vote choice . At that election, Conservative support was strongest among those who prioritized their British identity and weakest among those who prioritized their Welsh identity; the opposite was true for supporters of Plaid Cymru. In contrast, not only was Labour strongest among those who felt both strongly British and Welsh (a group located close to the median national identity among the Welsh electorate), but the party was also able to attract support from across the national identity spectrum.
National identities may also have an indirect association with vote choice, perhaps because they are chronologically earlier in a 'causal chain' and therefore less apparent in model fit in statistical models of voting behavior. In Scotland, for example, national identity can influence vote choice indirectly via its relationship with constitutional preferences: A majority of Scottish identifiers support independence and a majority of British identifiers prefer remaining in the Union (Bond 2015). Therefore, although a voter's preferences on the constitutional question predominate in vote choicethe constitutional divide in Scotland is now widely acknowledged to be the primary political cleavage in Scottish politics (e.g. Henderson and Mitchell 2018;Johns 2021;Liñeira and Cetrà 2015;McMillan and Henderson 2021) such preferences are in no way independent from their underlying national identities. Even in Wales, where the constitutional question is less bifurcated, the presence of an indirect effect of national identity on vote choice via constitutional preference can be identified (Wyn Jones and .

Using national identity to explain divergence
Even if it is reasonable to conclude from this discussion that national identity plays an important role in shaping vote choice in Scotland and Wales, the question remains: how might knowing this help explain the considerable divergence in fortunes we have noted in the period since 1999? To answer this question, we analyze the relationship between national identity and vote choice in all six devolved elections. Using surveys from the SSA, SES, and WES, we introduce separate multinomial logistic regression models for each Scottish and Welsh parliamentary election. We run a separate model for each devolved election in both countries, as the data is cross-sectional and does not include the same respondents. We select vote choice in the constituency ballot in both countries as our dependent variable, recoded to indicate whether a respondent voted for the Scottish/Welsh Conservatives, Scottish/Welsh Labour, the Scottish National Party or Plaid Cymru, or an 'other' party. Due to the small number of respondents in the 'other' category, we do not present this group in our analysis. We also exclude non-voters due to vastly different levels of non-voting between early and recent editions of both the WES and SES, a pattern which is likely due to the different methods of recruiting participants (random probability versus online). 8 For our measure of national identity, 9 we use the Linz-Moreno measure, which captures whether someone feels Scottish/Welsh and not British (1), more Scottish/Welsh than British (2), equally Scottish/Welsh and British (3), more British than Scottish/Welsh (4), or British and not Scottish/Welsh (5). It is true to say that scholars now increasingly recognize the limitations of this measure; in particular that the scale is of limited use in measuring trade-offs such as the relative intensities of an individual's dual identities (see Griffiths 2023;Guinjoan and Rodon 2015). In recent work these limitations have been compensated for by using other measures including relative territorial identity (RTI) . 10 The data required to create RTI scales are, however, not available in the Scottish and Welsh datasets before 2016. This leaves the Linz-Moreno scaledeployed in both territories across the entire time seriesas the only viable measure with which to operationalize national identity in this study.
As an initial point of orientation, Table 3 presents the percentage of respondents in Scotland and Wales that select each category in the Linz-Moreno question, excluding non-responses, over the whole period covered by our analysis. As can be seen, in Scotland, aggregate levels of national identities appear very stable between 1999 and 2021, with most individuals prioritizing their Scottish identity and only small numbers prioritizing a British identity (although the proportion reporting an 'equal' identity has risen slightly since 1999). Not only are the various national identity groups more evenly sized in Wales, but there has been more change since 1999. Notwithstanding an increase between 2003 and 2007, the proportion prioritizing their Welsh identity is rather similar in 1999, 2011 and 2021. The change is in the proportion claiming an 'equal' identity, which has shrunk considerably since 1999, while the proportion prioritizing their British identity has nearly doubled since 2003.
In our analysis we combine the 'More British than Scottish/Welsh' and 'British not Scottish/Welsh' into one category due to low sample sizessample size limitations are a general issue as we analyze the first three elections, especially in Wales. 11 Alongside Linz-Moreno measures of relative identities we include important sociodemographic control variables, namely age (interval variable, normalized between 0 and 1), education (university degree or not) 12 , gender (male/female), and social grade 13 (binary, non-manual [ABC1] and manual [C2DE] occupations). These variables are recoded to ensure each model draws comparable data from across the different surveys. All data is weighted using post-stratification weights to improve the representativeness of our sample relative to the general population, and respondents with missing data are excluded. The results of this modeling are presented in Figure 2 for Scotland and Figure 3 for Wales, where we report the predicted probability of voting for a party by voters identifying with each national identity category. We include the full models in the supplementary appendix. Table 3. National identity in Scotland andWales (1999-2021 These results serve to highlight the extent of divergence between Scottish and Welsh elections over time, with the association between national identity and probability of voting for Labour in Scotland and Wales being perhaps the most obvious starting point for consideration of this divergence.  In the first devolved elections, those who prioritized their Scottish identity were only slightly more likely to vote for the SNP than Scottish Labour, with Plaid Cymru enjoying a substantially stronger lead over Labour among 'Welsh only' identifiers. Nonetheless, even if national identity did differentiate voters, in 1999 both Scottish and Welsh Labour were able to attract substantial support across national identity spectrum in their respective countries. Subsequently, however, we see a change in Labour's fortunes in Scotland, with the party losing support not only among those who feel Scottish only but especially between 2011 and 2016among those who identify as more Scottish than British or equally Scottish British. Not only that, but respondents who prioritize a British identity were also far less likely to vote Labour in 2021 than has been the case 1999 as these voters shifted towards the Scottish Conservatives (Convery 2020;Kenny and Sheldon 2021). In Wales, meanwhile, even if (since 2007) British identifiers have been more likely to support the Welsh Conservatives than Welsh Labour (if by a narrow margin), the latter party has continued to perform relatively strongly among Welsh-only identifiers while enjoying a completely dominant position among those who feel more Welsh than British and equally Welsh and British. In broad brush terms, therefore, the divergence in electoral outcomes since 1999 can be said to be the result of a fundamental change in the relationship between national identity and vote choice in Scotland in contrast to long-term continuity in the case of Wales.

Why has the relationship between national identity and vote choice changed in Scotland but not in Wales?
But if we now have a better sense of what has changed, we are left with the question of why this has been the case? Our argument is that it reflects changing perceptions among those who identify as Scottish of the ability of Scottish Labour to defend Scottish interests. Comparisons between Labour and their Conservative and nationalist rivals, as well as between Wales and Scotland are instructive in this regard.
While it would be a mistake to conflate the history of the Scottish and Welsh Conservatives, nonetheless, until very recently at least, Conservative weakness has been a recurring theme in the discussions of voting in both countries (e.g. Mitchell 2015;Wyn Jones, Scully, and Trystan 2002). 14 A sense that parts of both electorates have tended to regard the Conservatives as an 'English party' that is either unable or unwilling to defend Scottish of Welsh interests has been pervasive (Bennie, Brand, and Mitchell 1997;Brown et al. 1999;Butler 1985;Seawright and Curtice 1995;Wyn Jones, Scully, and Trystan 2002) and has indeed regularly troubled the party itself (Convery 2016;Mitchell 1990). More recently, the focus of the literature has shifted to the ways in which the Scottish Conservatives have sought to capitalize on the aftermath of the post-independence referendum polarization of Scottish politics by portraying themselves as 'the party of the union' (see Convery 2020; Whigham 2019)a development that has its echoes in Wales . The nationalist parties, of course, adopt the converse position, presenting themselves as the explicit and selfconscious champions of the territorial national interest against the state. But what of Labour?
At both a UK and devolved level, many Labour supporters and politicians have (at times, at least) struggled to engage with the concept of 'nation', preferring instead to frame arguments in terms of social class (see Brown Swan 2022; Moon 2016) and/or simply assume a Britain-wide territorial framing that apparently requires no further explication. Indeed, what was then widely regarded as the party's underperformance in the first devolved elections in 1999 was widely attributed to its reluctance to engage with Scottishness and Welshness (see Denver and MacAllister, 1999;Trystan, Scully, and Wyn Jones 2003). Paterson et al. (2001), for example, argued that 'the principal reason for Labour's 'under-performance' in the first Scottish election was that it was not thought sufficiently capable of standing up for Scotland's interests within the Union' (2001,41). Similarly, the lack of clear differentiation between Labour in Wales and Labour in London at that election meant that the field was left clear for Plaid Cymru to portray itself as 'more 'Welsh' and left-wing than New Labour' (Wyn Jones and Scully 2003, 129).
The Scottish and Welsh Labour responses to the result of that inaugural election were, however, very different. Both academic and popular commentary has stressed the way that Welsh Labour responded to initial shock of the 1999 result by making a conscious attempt to appeal to the Welshness that many voters in Wales identified with (Trystan, Scully, and Wyn Jones 2003;. This has continued to the present day with the party emphasizing what it claims to be a natural affinity between identifying as Welsh and voting Labour (Williams 2018) and portraying itself as defending the Welsh interest against London. Welsh Labour has cooperated on a regular basis with Plaid Cymru through a series of more or less formal arrangements while avoiding any and all entanglements with the Welsh Conservatives.
The contrast with Scottish Labour could hardly be starker. Despite broad ideological similarities between it and the SNP, differences between them on the preferred constitutional future of Scotland have been regarded as decisive (Curtice 2011;Johns et al. 2009;Johns, Mitchell, and Carman 2013). During the campaign that preceded the 2014 Scottish independence referendum, Labour cooperated closely with the Conservatives as part of the 'Better Together' campaign, tying the two parties together under the banner of support for the union and, by extension, British identity (Fieldhouse and Prosser 2018). Labour's subsequent difficulties after the referendum have been compounded by the party's inability to establish a distinctive position for itself on the future of Scotland as well as tensions between the Scottish and British party organizations (Bennett et al. 2021). 15 We can explore voter perceptions of all of this by examining the association between national identity and perceptions of which parties defend Scottish or Welsh interests in 1999 and 2021. In doing so, it is important to note that although both the 1999 and 2021 election surveys include relevant questions, they are unfortunately not identical either territorially or temporally. In Scotland, the 1999 survey asked respondents whether they trust a party to stand up for Scottish interests just about always, most of the time, only some of the time, or not at all. The 2021 survey asked whether they think a party stands up for Scotland's interests very well, quite well, not very well, and not at all. For Wales, the 1999 questionnaire asked whether a party looks after the interests of Welsh people very closely, fairly closely, not very closely, or not at all closely. In 2021, the WES asked whether a party cares about the problems of Welsh people a great deal, a fair amount, not very much, or not at all. Given this variation and for ease of interpretation, we recode each four-category variable into a binary variable (combining the first two positive responses into one category and the latter two negative responses into another), and exclude non-respondents. Figures 4 and 5 present the percentage of respondents who have a positive perception of a given party's ability to defend Scottish/Welsh interests in each national identity category, for Scotland and Wales respectively.  Scotland in 1999. Scottish Election Study (1999, N: 1355-1386, (2021, N: 2783-2935).
As far as the Conservative and nationalist parties are concerned, perceptions of which party best defends Scottish or Welsh interests are rather similar in both countries and very much in line with expectations. In 1999, those who prioritized their Scottish identity were unlikely to think that the Conservatives stand up for Scottish interests 'always' or 'most of the time'. Likewise, even in the context of a rather generalized distrust of the Conservatives' willingness to care about the interests of Welsh people in general, those respondents in Wales who prioritized their Welsh identity were most negative in their perceptions. By contrast, those who prioritized their sub-state identities were highly likely to regard the SNP and Plaid Cymru as defending Scottish or Welsh interests respectively.
The same patterns recur in 2021. Respondents who prioritize their Scottish identity are highly unlikely to think the Conservatives stand up 'well' for Scottish interests but, conversely, are overwhelmingly likely to believe the at the SNP does so. Similarly, those who prioritize their Welsh identity are very unlikely to think the Conservatives care about the problems of Welsh people in general while very likely to think this is true for Plaid Cymru.
As for Labour, while in 1999 voters who prioritized their Scottish or Welsh identities tended to believe that Labour defended Scottish or Welsh interests, by 2021 this was no longer the case in Scotlanda development that very much coincides with the collapse in Labour's vote share among this group of voters. In Wales, though, even among those voters who emphasize their Welsh identity we continue to find a very high proportion who regard Labour as being willing to defend Welsh interests. The Welsh Labour strategy Source: Welsh Election Study (1999, N: 1018-1045), (2021, N: 2395-2488). of denying Plaid Cymru the opportunity to present itself as the only champion of the Welsh interest has been and remains remarkably successful.
These data also reveal further significant differences among those who prioritize their British identity in Scotland and Wales. Strikingly, in Wales, we find that in both 1999 and 2021 alike, Plaid Cymru is perceived to be the party best able to defend Welsh interests right across the national identity spectrumincluding among those who feel most British. In Scotland, however, by 2021 this is emphatically not the case either among those who feel equally Scottish and British or those who prioritize their British identity. Exploring why this is the caseand indeed, why this state of affairs has not translated into greater success for Plaid Cymru at devolved electionstakes us beyond the scope of the current paper. It does, nonetheless, illustrate some of the potentially significant questions brought into focus by this first, over-time comparative view of devolved elections in Scotland and Wales.

Conclusion
To compare the results of the 2021 devolved elections in both Scotland and Wales with those of their predecessor elections in 2016 is to be struck by a distinct absence of change. In Scotland, while the governing SNP saw a marginal strengthening of its already dominant position, the vote shares of all the main political parties were remarkably close to those reported five years previously. The collapse of UKIP meant that there was a little more change in Wales, with the Welsh Conservatives the main beneficiaries. But, again, the impact on the overall balance of the Senedd was marginal, with the governing Welsh Labour party not only sailing serenely on, butlike the SNPeven managing to further cement its previous dominance.
But as we have sought to demonstrate in this paper, if we compare the 2021 results against those of the first devolved elections in 1999 then we are left with a very different analytical perspective that focuses, not on continuity, but rather on divergence. The results in the inaugural elections for the Scottish Parliament and then National Assembly of Wales were remarkably similar. In both countries, a dominant Labour party successfully managed to fend off challenges from insurgent nationalist parties. Since then, however, electoral politics in both countries have moved in very different direction. 1999 proved a false dawn for Plaid Cymru as Welsh Labour moved successfully to rebuild and reinforce its dominant position. In Scotland, in contrast, by the second decade of devolution the Labour party had been supplanted by the SNP as the country's strongest political force.
We have sought to throw new light on this process by viewing developments over time in both territories through a comparative lens focused on national identitylong identified as a key determinant of vote choice in the relevant literatures. We have shown that, while national identity remains a driving force of voting behavior, its relationship to different parties (or rather, to a particular party) has shifted dramatically. In Scotland, in 2021 support for Scottish Labour had collapsed among those who prioritize their Scottish identity, with the SNP being the primary beneficiaries. Support for Scottish Labour has also fallen (albeit to a lesser extent) among those who prioritize their British. By the time of the 2021 election this had left the party caught in what might be termed an electoral 'noman's land'.
Comparison with Welsh Labour only serves to underline the extent of Scottish Labour's challenge in the third decade of devolution (and the altered conditions that might underpin a turnround in the party's fortunes). For in Wales, while national identity continues to differentiate very clearly between Plaid Cymru and Welsh and Conservative supporters, with Welsh identifying voters tending to support the former and British-identifiers the latter. But, as was the case in 1999, Welsh Labour continues to enjoy significant support across all national identity groups. Many of those exclusively or predominantly Welsh identifying voters who may have flirted with Plaid Cymru have been brought back into the Labour fold. Meanwhile, even if the Conservatives have outpolled Labour among British identifiers in four of the six devolved elections held in Wales, nonetheless Labour is broadly competitive even here (emphatically so, compared with Plaid Cymru).
We have theorized that these divergent patterns of vote choice in Scotland and Wales reflect voter perceptions of the ability of different political parties to defend Scottish or Welsh interests. As far as Labour is concerned, in 1999 both Scottish and Welsh Labour were able to convince those who prioritize their Scottish/Welsh identities that they could defend Scottish/Welsh interests. While this continued to be case for Welsh Labour, at the 2021 Scottish parliamentary elections this was not the case for Scottish Labour.
Compared to Scottish Labour at least, perceptions of the Conservatives and the two nationalist parties have been consistent over time in both Scotland and Wales. Those who prioritize their Scottish or Welsh identities are very unlikely to believe that the Conservative party defends Scottish or Welsh interests. This contrasts sharply with the same groups sense that the SNP and Plaid Cymru's stand up for/care about Scottish or Welsh interests. One interesting difference between both countries, however, is that while British identifiers in Wales seemingly accept the sincerity of Plaid Cymru's claims to represent Welsh interests, by 2021 British identifiers in Scotland were emphatically not of the same view with regards the SNP's claim to be representing Scottish interests.
Notwithstanding this difference, the overall picture that emerges from our analysis is of Scotland and Wales as two devolved political systems in which national identity acts as a key cleavage in the context of voting behavior. Both nationalist and Conservative parties find themselves arranged on either of the dividea divide, it is tempting to add, that Scottish Labour had fallen into by the second decade of devolution. Largely because it has refused to cede the ground of 'standing-up for Wales' to Plaid Cymru, at the sixth set of devolved elections Welsh Labour stood alone in successfully straddling this key cleavage. Notes 1. Scotland and Wales use a variant of mixed-member proportional electoral system. 2. The longevity of Labour dominance in Wales is remarkable: it has emerged as the largest party both in terms of vote share and seats from every devolved and UK-wide election in Wales since 1922 ). 3. For example, the SNP had to deal with 'an explosive political scandal enveloping the incumbent First Minister Nicola Sturgeon and her predecessor Alex Salmond' (McMillan and Henderson 2021, 37), while Welsh Labour had to contend with the 'emergence of a substantial and energetic grassroots independence campaign, Yes Cymru' (Larner et al. 2022, 4). 4. Scotland's electoral system meant that the SNP's strong constituency performance prevented them from winning further top-up seats in the regional list ballot. 5. In Scotland: (Anderson 2016;Denver and MacAllister 1999;Denver 2003;Johns 2021;Johns et al. 2009;Johns, Mitchell, and Carman 2013;McMillan and Henderson 2021;Scully 2013). In Wales: (Balsom, 2000;Bradbury, Denver, and MacAllister 2000;Jones 1999;Larner et al. 2022;McAllister and Cole 2014;Scully 2013;Scully and Elias 2008;Scully and Larner 2017;Trystan, Scully, and Wyn Jones 2003;2012). 6. Support for Labour and (particularly) the SNP associated with feeling Scottish, with the opposite being true for the Conservatives. 7. See Larner (2019) and Wyn Jones (2022) for further discussion. 8. Levels of non-voting are significantly higher in earlier editions of the SSA and WES, which may be a product of how the data was collected. Earlier surveys (pre-2007(pre- in Scotland, pre-2011 in Wales) were collected using a random-probability sample, whereas more recent surveys have been collected online. There is some evidence that respondents are less likely to misreport turnout (due to embarrassment) when taking self-administered online surveys (Holbrook and Krosnick 2010). However, online surveys are more likely to oversample the politically engaged (Karp and Lühiste 2016), which will likely lead to higher estimates of turnout. 9. It should also be noted that Wales has a significantly higher proportion of English-born residentsand English identifiersthan Scotland (Bradbury and Andrews 2010;Mann and Fenton 2017). It is also the case that those who identify as British but were born in England have very different political outlooks than those who identify as British but were born in Wales (see Wyn . Unfortunately, previous election studies do not capture relative English/British identities in Wales. Instead, the most common measure of relative state/sub-state identity in the Scottish and Welsh election studies focus on relative Scottish/British and Welsh/British identities. Consequently, we focus on these identities. 10. Although relative territorial identity measures also have their own limitations (see Griffiths 2021).