Regions as primary political communities: A multi-level comparative analysis of turnout in regional elections

Despite the importance of regional democracy, comparative analyses of voter participation in regional elections are rare. We examine individual participation in regional elections in Canada, Spain, and the United Kingdom and make three arguments: (i) standard models of turnout devised for national contests only partly explain regional turnout, with the personal characteristics of voters more important than contextual variables; (ii) territorial identity and the level of jurisdictional authority wielded by the regional legislature are important determinants of electors’ willingness to participate in regional elections; (iii) when we contrast subjective perception and objective/aggregate findings we see that regional legislative authority matters more than perceived salience, and individual regional identity matters more than the views of fellow citizens, but that these effects are largely conditional on age.

Why is voter turnout higher in some regions and lower in others? The issue of varying political participation has vexed political scientists and practitioners alike, and especially so in the last two decades amid an apparent decline in voter turnout. However, voter turnout, and whether declining participation poses a crisis of legitimacy for democratic institutions, has principally been examined in the context of national and, more recently, European parliamentary elections (Jackman and Miller 1995; Blais and Dobryzynska 1998; Blais 2000; Franklin 2004; Flickinger and Studlar 2007). In contrast, participation in regional elections, that is, those elections to intermediary “meso-level” authorities between the municipality and the state, has rarely been the subject of political inquiry, despite considerable variation in regional turnout from one individual to another and from one region to another.

And yet regional democracy has become increasingly important in recent years. Across Europe and North America, many new elected regional legislatures and governments have been established while existing regional institutions have seen their power and responsibility extended. The regional political authority index compiled by Hooghe, Marks, and Schakel (2008) demonstrates that, between 1970 and 2005, twenty-nine of forty-two mainly EU and OECD states became more “regionalized,” while only two became marginally less regionalized. Regional institutions now make legislative, policy and spending decisions over a vast range of activities central to their populations. The legitimacy underpinning regional rule comes not so much from the central governments that have dispersed constitutional power and responsibility, but from the regional populations who express (or withhold) democratic consent when electing regional representatives at the ballot box. Given the increasing importance of the regional layer of democracy, it is perhaps surprising that there has been relatively little attention paid to political participation in regional elections, especially beyond single case studies (for the latter, see Hough and Jeffery 2006; Dandoy and Schakel 2013). A few studies have examined variation in inter-regional turnout using aggregate data (Horiuchi 2005; Percival et al. 2007; Henderson and McEwen 2010), but so far as we are aware, comparative studies of variation in individual-level turnout in regional elections across different regions and nation-states are rare, largely because of a lack of comparable data.1 Yet, it is not nations and regions that decide whether to vote, but those citizens resident within them. The risks of a methodological reliance on aggregate-level data to explain individual behavior are well-rehearsed in studies of the ecological fallacy (Robinson 1950; Kramer 1983). The choices individual citizens make are shaped by the context in which they live, but the meaning and significance of that context is inevitably subject to varying individual interpretations.

This article confronts the problem of data availability directly by using a bespoke dataset which merges individual-level data from different survey sources, and supplements the merged data with aggregate data pertinent to the regional level. The resulting dataset is used to explore variations in individual voter participation in regional elections across twenty-nine regions in Canada, the United Kingdom, and Spain.

The article’s objectives are three-fold: First, to examine the power of existing models developed from national election studies to explain voter participation in regional elections; second, to argue that variation in regional turnout can be understood better by supplementing existing models with variables particular to regional politics; and third, to contrast the capacity of individual and aggregate level variables to explain variation in individual voter participation. In line with earlier research (Percival et al. 2007; Henderson and McEwen 2010; Ragsdale and Rusk 2011), we expect that regional voter participation is affected by the power and influence of the institutions being elected, as well as the sense of attachment voters feel to the region in question. We also hypothesize that electoral behavior will be driven by individual perceptions of an institution’s importance, as well as the feelings electors have toward the region in question. We thus distinguish between the impact of subjective and objective indicators. We expect that individual perceptions matter, but that their effect will vary and be conditional.