SEARCH

SEARCH BY CITATION

Keywords:

  • ethnic catch-all;
  • consociationalism;
  • deeply divided places;
  • voting;
  • Northern Ireland

Abstract

  1. Top of page
  2. Abstract
  3. Ethnic Catch-All Voting
  4. Data
  5. Results
  6. Discussion and Conclusion
  7. References
  8. Biography

I elaborate a model of cross-bloc party support in deeply divided places. The model expects that the variation in the level of electoral support that citizens in Community A have for parties in Community B is a function of citizens' evaluations of the relative ability of parties in Community B to represent the interests of all communities. This ‘ethnic catch-all’ model of cross-bloc party support is tested in the context of consociational Northern Ireland, using data from a representative survey conducted directly after the 2010 Westminster general election. The findings are asymmetric: the model explains Protestant support for nationalist parties but not Catholic support for unionist parties. The findings, and their implications, are discussed.

Deeply divided places are characterised as having a number of party systems – one for each of the competing ethnic groups (Evans and Duffy, 1997; Horowitz, 1985; Rabuska and Shepsle, 1972). Party competition typically occurs within rather than between each ethnic bloc, and voting for a party in a different ethnic bloc – voting ‘across the divide’ – is extremely rare. Here, I focus on the potential for voting across the divide, in order to aid our understanding of how citizens in a deeply divided place view political parties in the rival bloc.

In order to explain potential cross-community voting, I elaborate the concept of ethnic catch-all voting, whereby citizens' levels of support for parties from a rival ethnic bloc are a function of citizens' evaluations of the ability of such parties to represent the interests of all of the competing communities. I argue that using the concept of ethnic catch-all voting to explain potential cross-community voting complements recent attempts to use ethnic tribune voting (supporting the party in your own ethnic bloc that you perceive to be the best at representing your ethnic group's interests [Mitchell et al., 2009]) to explain current within-community voting.

To identify which particular cross-bloc parties are likely to be perceived as more competent than others in terms of performing an ethnic catch-all function, I distinguish between issue-based and valence-base voting (following Stokes, 1963) and theorise the link between the two (building upon arguments made in Mitchell et al., 2009 and Sanders et al., 2011). I argue that valence judgements are likely to be driven by party positions. Specifically, I argue that parties that have consistently held moderate ethnonational positions will enjoy an ethnic catch-all valence advantage. In other words, if two parties in a rival bloc are currently adopting moderate ethnonational positions, the party that previously adopted an extreme position will be less credible as an ethnic catch-all party than the party that did not previously adopt an extreme position: hawkish party origins militate against, while dovish party origins facilitate, positive perceptions of rival bloc parties' abilities to perform the ethnic catch-all function.

I focus on the case of Northern Ireland, which is a particularly useful case to illustrate the power of ethnic catch-all voting to explain potential cross-community voting because, as a result of Northern Ireland's now functioning inclusive power-sharing consociational system, the role played by parties in representing (or not representing) particular community interests is highly salient (Mitchell et al., 2009).

The article is organised as follows. I begin by elaborating the concept of ethnic catch-all voting. Next, I describe the data used in the analysis, which come from a survey conducted in the aftermath of the 2010 Westminster general election in Northern Ireland. Subsequently, I report the extent of potential voting across the divide and I assess the degree to which this may be explained by ethnic catch-all voting (controlling for the other plausible explanations). I end with a discussion of the findings and their implications for our understanding of voting behaviour in Northern Ireland and other deeply divided places.

Ethnic Catch-All Voting

  1. Top of page
  2. Abstract
  3. Ethnic Catch-All Voting
  4. Data
  5. Results
  6. Discussion and Conclusion
  7. References
  8. Biography

One view of party competition and voting behaviour in deeply divided places is that fierce intra-community party competition leads to parties becoming more and more extreme (Horowitz, 1985; Rabuska and Shepsle, 1972). This is because any move in a moderate or conciliatory direction by a party leaves it vulnerable to being ‘outflanked’ by a competing party which accuses it of ‘selling out’ the ethnic group. Hence, parties compete with each other to offer the most hard-line defence of the ethnic group, leading to a spiral of moves in an extreme direction by parties in each of the ethnic blocs and, consequently, a lowering of the probability of any meaningful cross-community cooperation between parties of the different ethnic blocs.

This rather grim ‘ethnic outflanking’ reading of party competition is, as argued by Paul Mitchell et al. (2009) and John Garry (2009), conditional on at least two factors: voters' positions and institutional design. If voters are indeed hard line and become increasingly hard line then parties' movements in an extreme direction will be electorally rewarded (and, therefore, incentivised). Similarly, if there are no institutional features that minimise or offset incentives to extreme movement by parties, such movement may well flourish. On the other hand, if voters are, and remain, relatively moderate, movement in an extreme direction by parties will result in losses (and is hence disincentivised); and if political institutions are carefully set up to resolve the conflict by rewarding – with real political power – moves in a moderate direction by hard-line parties, party movement in a polarised direction is less likely to occur.

As argued by Garry (2009) in relation to Northern Ireland, the existence of relatively moderate voters plus the strong incentives to moderate party positions generated by consociational institutions led to ostensibly ‘extreme’ parties (the Democratic Unionist Party [DUP] and Sinn Féin in the Protestant/unionist and Catholic/nationalist blocs, respectively) adopting more conciliatory positions. Compromises were made by these parties relating to constitutional preferences and attitudes to power sharing, such that both parties appeared to encroach upon the relatively moderate positions of, respectively, the Ulster Unionist Party (UUP) and the Social Democratic and Labour Party (SDLP). The fact that both main parties in each competing bloc became very similar on these key ethnonational or ‘conflict’ issues led to a reduction in the role played by constitutional preferences, identity and attitudes to power sharing in driving vote choice in the 2007 Assembly election (at least in the unionist bloc) (Garry, 2009).

Mitchell et al. (2009) argue that these compromises made by Sinn Féin and the DUP on constitutional and power-sharing issues have indeed led to both parties in each bloc currently occupying the moderate centre ground, irrespective of the parties' formerly extreme positions. However, what does now differentiate the parties in each bloc is how capable the parties are perceived to be at representing the interests of their community in the executive – that is, their capacity to act as ethnic tribune parties. Once all parties in a deeply divided place have agreed to share power, and have thus, essentially, accepted the legitimacy of the state, constitutional issues recede in importance and are replaced by the now crucial issue of how resources are to be allocated by the new power-sharing government. Voters assess the relative competence of each of the parties in their bloc to ‘deliver’ for that community in the power-sharing government and the party best placed to ‘perform well’ in terms of representing their community's interests will be electorally rewarded (Mitchell et al., 2009).

Prior Position Cues Current ‘Ethnic Tribune’ Valence Assessment

According to Mitchell et al. (2009) – and similarly argued by Mitchell and Geoffrey Evans (2009) and by John McGarry and Brendan O'Leary (2009) – within-community party competition has moved away from being issue based (i.e. relating to parties' positions, in this case on constitutional matters and power sharing) and has instead become valence based (i.e. relating to relative evaluations of parties' reputations for efficiently performing a particular function,1 in this case parties robustly representing their community in resource-allocation decision making in the executive). Essentially, movement by the ‘extreme’ parties to the centre ground on constitutional issues is married to a robust reputation on community representation and it is this combination of issue moderation and valence robustness that explains the recent electoral success of the DUP and Sinn Féin.

An important aspect of Mitchell et al. 's (2009) argument is that this ‘combination’ is also causally related, in the sense that the extent to which a party has a reputation for robustly defending the interests of the community (i.e. its ‘tribune’ reputation) is a function of the party's former position on constitutional matters. For clarity, I sketch Mitchell et al.'s (2009) description of changes in party position that underpins their ‘ethnic tribune party’ argument (Figure 1). Essentially, Mitchell et al. (2009) argue that a voter in a given bloc (i.e. either a nationalist voter or a unionist voter) at time 2 is faced with parties that are close in terms of ethnonational position. Performance-based voting is likely to be important at time 2 given the rise in salience of the protection of community interests in the new power-sharing government. In order to identify which party is best able to represent their group interests a voter will take into account both past and present positions of parties on the ethnonational cleavage. Mitchell et al. (2009, p. 402) state that voters in a particular community seek to be ‘represented by their “strongest voice”. Typically this will be parties with reputations for tough bargaining, and such reputations will partly be based on their past records of less-moderate policy positions’.

figure

Figure 1. A Sketch of Mitchell et al.'s (2009) Discussion of Party Position

Note: The graph shows the movement in a moderate direction over time of the ‘extreme’ parties in each bloc (Sinn Féin and the DUP).

Download figure to PowerPoint

Hence a nationalist voter will reason as follows: because Sinn Féin has historically been extreme/hard line on the ethnonational dimension (as per time 1 in Figure 1) Sinn Féin is currently more credible as a robust defender of community interests. Similarly a unionist voter will reason: the UUP and DUP are very similar now in terms of ethnonational position but because the DUP used to have a more extreme position than the UUP it (the DUP) is best placed to defend unionist interests staunchly. In short, position drives valence – and more specifically prior ethnonational position drives current ethnic tribune valence.

The ‘position drives valence’ argument in Mitchell et al. (2009) is consistent with the argument later elaborated by David Sanders et al. (2011) in the context of British voting behaviour. They find that:

spatial calculations act as a source of valence judgements. The hypothesis considered is that Downs undergirds Stokes – voters use spatial calculations as heuristic devices when assessing party competence. Parties viewed as closer than their competitors to voters in spatial terms are more likely to be judged as credible vehicles for achieving widely shared policy goals such as economic prosperity, health care and public safety. Analyses are consistent with this spatial cueing hypothesis (Sanders et al., 2011, p. 312).

The interesting distinction between Mitchell et al.'s (2009) argument and Sanders et al.'s (2011) argument relates to the time point of position that performs the valence cueing. In Sanders et al. (2011), it is current positional factors that cue current valence judgements, whereas in Mitchell et al. (2009) it is prior (historical) positional factors that cue current valence judgements.

Prior Position Cues Current ‘Ethnic Catch-All’ Valence Assessment

The weak ‘tribune’ reputations of certain parties (namely the UUP and the SDLP) should not necessarily be an inevitable disadvantage for the parties in the medium term. This is because the UUP and SDLP's moderate positions (historically and currently) can leave them well placed in terms of non-tribune-type representation – namely, what is here termed ethnic catch-all representation. This is the reputation for being able to, at least to some significant extent, represent both of the competing communities; or, to put it slightly differently, to represent the other community in addition to representing their own community. The consistently non-extreme positions, over time, of the UUP and the SDLP on constitutional matters (as sketched in Figure 1) arguably give them reputational credibility in terms of an ethnic catch-all function.2

Using the term ethnic catch-all to describe a party is an attempt to convey the idea of a party broadening its appeal beyond the demographic/ethnic group from which it initially attracted support. Otto Kirchheimer (1966) described as catch-all those socialist parties which sought to broaden their appeal beyond the boundaries of the working class. A catch-all party seeks to reduce the priority placed on ideology and narrow social groups and instead seeks to appeal increasingly to a broader range of citizens (Wolinetz, 2002).

In their examination of party modernisation in deeply divided places, Cathy Gormley-Heenan and Roger Mac Ginty (2008) discuss the possibility of parties broadening their electoral appeal. The authors focus on the illustrative case of the DUP and, in a similar vein to Mitchell et al.'s (2009) argument, suggest that the DUP has moved from being a staunchly and comprehensively hard-line (pre-modern or non-modern) party to a party that subtly combines elements of intransigence with elements of progressiveness/modernism. As well as increased centralisation, professionalisation, engagement with the state and employment of technologically advanced election campaigning, Gormley-Heenan and Mac Ginty discuss broadening of electoral appeal within the Protestant community as a fifth facet of DUP party modernisation: ‘Clearly, in the context of divided societies, the likelihood of a “classic” catch-all party prospering is remote … In order to maximise support in ethnic elections, parties must reflect societal fissures and garner support in their own bloc’ (Gormley-Heenan and Mac Ginty, 2008, p. 56, emphasis added). Gormley-Heenan and Mac Ginty argue that the DUP, for example, has attempted to broaden its electoral base by appealing to both the fundamentalist wing (associated with Ian Paisley) and increasingly the more secular wing of Protestants (associated with, for example, Peter Robinson).

Given Gormley-Heenan and Mac Ginty's focus on the DUP up to 2007, their conclusion that broadening of electoral appeal is constrained to being within the ethnic community (rather than across ethnic communities) may be reasonable. However, I suggest here that the movement of ethnic parties towards becoming genuinely catch-all in the sense of appealing to out-group members may well be plausible, particularly for parties whose long-established reputations as moderates give them greater credibility to reach across the divide. Hence, the parties in each bloc with conciliatory reputations – the UUP and SDLP – may reasonably be regarded as having (at least some) ethnic catch-all appeal. In the Catholic/nationalist bloc, the SDLP has a strong reputation as a consistently constitutional party – very distinct from Sinn Féin which has emerged from a violent non-constitutional tradition. The SDLP has long fostered a policy of peaceful protest and engagement in dialogue with the representatives of the ‘rival’ community (McLoughlin, 2009). In the Protestant/unionist bloc, the UUP is typically regarded as the more conciliatory party. The party's perceived moderateness led to the establishment of the Democratic Unionist Party in the early 1970s by Ian Paisley as a bulwark against any possible unionist compromise (see, for example, the discussion in Evans and Duffy, 1997). Along with the SDLP, the UUP led the way in negotiating the 1998 Good Friday/Belfast Agreement, leading to the receipt of the Nobel Peace Prize by John Hume and David Trimble, respective party leaders.

A nationalist voter assessing the extent to which they support (or do not support) parties in the unionist bloc will, this article suggests, be driven by their evaluations of the capabilities of the rival bloc parties to represent all groups in society. In assessing the ethnic catch-all abilities of the parties in the rival bloc, nationalist voters will be influenced by the party positions of the rival bloc parties. They may reason as follows: the UUP and the DUP currently hold moderate ethnonational positions. However, the UUP is a much more credible actor in terms of possibly representing fairly and competently the interests of all communities because the UUP has a history of acting moderately (or at least more moderately than the DUP). Similarly, unionist voters assessing rival bloc parties are likely to regard Sinn Féin and the SDLP quite differently, given that the former party arose from a violent paramilitarism in contrast to the consistent constitutionalism of the latter. Hence, the SDLP may be regarded as a much more credible performer of the ethnic catch-all function.

In essence, in the same way that Mitchell et al. (2009) argue that prior ethnonational position cues current (ethnic tribune) valence assessments and (current within-bloc) voting, this article argues that prior ethnonational position cues current (ethnic catch-all) valence assessments and (potential cross-bloc) voting. The precise hypotheses are specified as:

  • H1. Variation in Protestant citizens' likelihood of voting for the SDLP rather than Sinn Féin is a function of variation in Protestant citizens' relative evaluation of the capacity of the SDLP and Sinn Féin to perform the ethnic catch-all function.
  • H2. Variation in Catholic citizens' likelihood of voting for the UUP rather than the DUP is a function of variation in Catholic citizens' relative evaluation of the capacity of the UUP and the DUP to perform the ethnic catch-all function.

Controlling for Other Predictors

A convincing test of the relationship between ethnic catch-all evaluations and potential vote for parties in the rival bloc must control for the range of current ethnonational positional factors. This is to ensure that catch-all valence evaluations are not merely picking up, or acting as proxies for, factors relating to ethnonational position. This study did not include specific questions asking respondents to indicate the position of each party in the system on an ethnonational ideological placement scale. Hence, I follow previous practice in studies of electoral behaviour in Northern Ireland and operationalise citizens' ethnonational position by measuring respondents' national identity (British/Irish/Northern Irish), constitutional preferences (united Ireland/devolution/direct rule), attitudes to power sharing (in favour or against) and ethnonational ideological self-description (unionist versus nationalist scale).3

As well as the ethnonational arena, citizens' positions on other political cleavages – relating, for instance, to the EU or socio-economic left–right dimensions – may drive levels of support for rival bloc parties. On the EU, the SDLP has a distinct position as the most Euro-enthusiastic party in Northern Ireland while the other main parties – Sinn Féin, UUP and DUP – are all, to varying degrees, Eurosceptic. Peter McLoughlin (2009, p. 603), for example, refers to the SDLP as being ‘generally viewed as the most pro-European party in Ireland’ and the party has also been described as ‘the champions of European integration in Northern Ireland’ on the basis of a survey of party candidates (Gilland Lutz and Farrington, 2006, p. 727; see also the discussion in Garry, 2012). One might imagine that pro-EU Protestants might, in theory at least, be drawn potentially to the SDLP given the absence of a pro-EU party in their own ethnic bloc. On the economic left–right dimension, the two nationalist parties are often characterised as being to the left of the two unionist parties. This is particularly so for Sinn Féin but the SDLP also has clear social democratic roots. While the UUP has often been seen as being to the right of the DUP, in recent years the DUP – despite its left-wing origins – is defining itself as a pro-business party.4

Data

  1. Top of page
  2. Abstract
  3. Ethnic Catch-All Voting
  4. Data
  5. Results
  6. Discussion and Conclusion
  7. References
  8. Biography

Sample

A post-election quota-controlled telephone survey was conducted by Market Research Northern Ireland5 in the direct aftermath of the 2010 Westminster election (N = 1,000). The survey was a stand-alone study commissioned by the author using financial resources from Queen's University Belfast. The representativeness of the survey is investigated by comparing distributions on key variables to the results of the 2010 Northern Ireland Life and Times (NILT) survey which is based on random sampling.6 The NILT 2010 survey yields the following distribution on the political ideology question: unionist 34 per cent, nationalist 20 per cent, neither unionist nor nationalist 45 per cent.7 The comparable figures from the 2010 election survey analysed in this article are, respectively, 36 per cent, 22 per cent and 42 per cent. Additionally, in terms of constitutional preferences, the figures supporting a united Ireland are directly comparable across the election survey and the NILT and are very similar (17 per cent in the election survey compared to 16 per cent in the NILT). The election survey also represents very accurately citizens' actual vote choice in the 2010 election. The figures are (real world election result followed by survey estimate): Alliance (9.3, 6.3), DUP (25.8, 25.0), SDLP (15.8, 16.5), Sinn Féin (25.2, 25.5), UUP (14.8, 15.2), TUV (3.3, 3.9), others (5.8, 7.6). Thus, the survey analysed in this article, while based on quota sampling, replicates well real world voting behaviour and independently generated frequencies on key political variables derived from a random sampling-based survey.8

Measuring Potential Vote

Studies of voting behaviour typically use vote choice as a dependent variable. Hence, voters are divided into a number of categories: assigned a ‘1’ for the particular party they voted for and assigned a ‘0’ for all of the other parties for which they did not vote. This categorical operationalisation of voting behaviour is almost ubiquitous and seemingly unproblematic. However, a number of analysts9 have emphasised the limited information yielded by such an operationalisation of voting behaviour. The 1 versus 0 distinction does not capture the variation in the levels of support voters may have had for the range of parties in the system. One voter may well have no likelihood at all of voting for any of the parties for whom he or she did not vote. In contrast, another voter may have been torn between voting for parties A, B and C; only very marginally preferring A (for whom he or she voted) to the other parties. The conventional operationalisation of voting tells us nothing about the possibly nuanced set of views a voter may have about the range of parties in the system. In order to measure systematically voters' levels of support for all of the parties in the system a question wording has been developed, validated and used in national and cross-national research projects.10 Applied to the Northern Ireland setting, the wording is as follows:

There are a number of political parties in Northern Ireland, each of which would like to get your vote: How likely is it that you would ever vote for the following parties. Please indicate your views on this 1–10 scale, where 1 means not at all likely and 10 means very likely. You may choose any number between 1 and 10. How likely is it that you would ever vote for [each party is listed and each party is separately scored by the respondent between 1 and 10].

The question is specifically designed not to act as a predictor variable (such as a party ‘likeability’ or party sympathy scale). The wording directly relates to the act of voting rather than to any theoretical explanation of voting; hence it is designed as a nuanced measure of the variation in levels of electoral support that citizens have for the set of parties in their system (and is thus acting as a dependent rather than independent variable). Such measures (often referred to as ‘propensity to vote’ measures, or more simply measures of ‘party support’) have a range of advantages. Of particular import for present purposes is the detailed information such measures give us about parties not voted for. Crucially, in stark contrast to the conventional measure of vote choice, what these variables provide is information regarding citizens' levels of electoral support for parties in the rival bloc. Furthermore, these electoral support measures have previously been defended and applied to useful effect as the dependent variable in electoral behaviour research in the Northern Ireland setting (Garry, 2007 pp. 354–7; 2009, p. 463 fn 21).

For present purposes it is the difference in likelihood of voting for the two rival bloc parties that is important. Hence, for Protestant respondents, the likelihood of voting for Sinn Féin is subtracted from the likelihood of voting for the SDLP, resulting in a scale that theoretically runs from −9 to +9, higher scores representing a greater likelihood of voting for the SDLP than for Sinn Féin. For Catholic respondents, the likelihood of voting for the DUP is subtracted from the likelihood of voting for the UUP, resulting in a scale that theoretically runs from −9 to +9, higher scores representing a greater likelihood of voting for the UUP than for the DUP.11

Measuring Ethnic Catch-All Evaluations and Control Variables

In order to measure ethnic catch-all evaluations, respondents were asked: ‘How good is each party at representing the interests of both the Protestant community and the Catholic community [1–10 scale, where 1 is “not good at all” and 10 is “very good”]’. It is the difference in the scores for rival bloc parties that is important for present purposes. Hence, similar to the above discussion regarding the construction of the dependent variables, Protestants' relative evaluations of the capacity of rival bloc parties to act as ethnic catch-all parties are generated by subtracting their score for Sinn Féin from their score for the SDLP (and Catholics' relative evaluations are generated by subtracting the DUP evaluation score from the UUP evaluation score).

To measure ideological self-identification the following question was asked: ‘Would you say you are: very strongly unionist, fairly strongly unionist, neither unionist nor nationalist, fairly strongly nationalist, very strongly nationalist’. To measure attitudes to power sharing respondents were asked: ‘Were you in favour of the power-sharing government being established after the 2007 election or were you opposed to it? – Very opposed, fairly opposed, in favour, very much in favour?’ To measure national identity respondents were asked: ‘Which of these best describes the way you think of yourself – British, Irish, Northern Irish (or other)?’ To measure constitutional preferences, respondents were asked the following question: ‘In terms of the long term future of Northern Ireland, which would you prefer? Northern Ireland should … (a) Remain in the UK with a direct and strong link to Britain, (b) Remain in the UK and have a strong Assembly and government in Northern Ireland, or (c) Unify with the Republic of Ireland?’

In order to measure economic left–right ideological views, the following question was asked: ‘In terms of taxation and spending which of the following would you prefer? – (a) cut taxes and spend less on health and education, (b) keep tax levels and spending levels as they are now, or (c) increase taxes and spend more on health and education’. In order to measure attitudes to European integration the following question was asked: ‘Some people are in favour of the European Union and further European integration while other people are opposed. What about you? Are you: very strongly in favour, fairly strongly in favour, not sure either way, fairly strongly opposed, very strongly opposed?’ Respondents were also asked a range of demographic questions, including religion, age, education and social class (as measured by the standard market research distinction between non-manual [ABC1] and manual [C2DE]).12

Results

  1. Top of page
  2. Abstract
  3. Ethnic Catch-All Voting
  4. Data
  5. Results
  6. Discussion and Conclusion
  7. References
  8. Biography

Protestant Support for Rival Bloc Parties

Figure 2a and b show variation in Protestants' likelihood of voting for each of the nationalist parties. Very low levels of potential electoral support are evident from Protestants for Sinn Féin (73.5 per cent indicated the lowest category [1]). In contrast, there is a much greater well of potential support for the SDLP (23 per cent indicate a score of 6 or higher and only 27 per cent indicate the lowest possible score of 1). Figure 2c shows the variation in Protestants' relative support for the two nationalist parties, with positive scores indicating greater potential support for the SDLP than for Sinn Féin, negative scores indicating a preference for Sinn Féin over the SDLP and a score of ‘0’ indicating equal support levels for both parties. Clearly, there is quite a degree of variation in Protestants' relative propensity to vote for either Sinn Féin or the SDLP and the balance of potential vote very clearly favours the SDLP. The patterns in Figure 3a–c, which focus on ethnic catch-all capacity, are less stark but point in a similar direction. Thirty six per cent of Protestants indicate a score of 6 or higher for the SDLP and only 5 per cent give it the lowest possible score; in contrast, only 20 per cent indicate a score of 6 or higher for Sinn Féin and almost the same proportion give it the lowest score.

figure

Figure 2. Protestants' Likelihood of Voting for Rival Bloc Parties

Note: The more positive the score, the greater the likelihood of voting SDLP rather than Sinn Féin.

Download figure to PowerPoint

figure

Figure 3. Protestants' Evaluations of Catch-All Ability of Rival Bloc Parties

Note: The more positive the score, the more positive the perceived catch-all ability of the SDLP compared to Sinn Féin.

Download figure to PowerPoint

Do relative catch-all evaluations predict relative party support, controlling for other relevant factors? The regression reported in Table 1 suggests that it does. The catch-all predictor is highly statistically significant and substantively very meaningful. This suggests that, controlling for other relevant factors, relative evaluations of rival bloc parties' abilities to perform the ethnic catch-all function drive Protestants' relative likelihood of voting for the rival bloc parties, hence confirming Hypothesis 1.13

Table 1. Predicting Protestants' Propensity to Vote SDLP Rather than Sinn Féin: OLS Regression
 coeff.se
  1. * 0.05 level, ** 0.01 level and *** 0.001 level of statistical significance.

(ref = female)  
Male0.1650.183
(ref = 45 plus)  
Younger than 450.449*0.183
(ref = lower than degree)  
Degree0.2540.225
(ref = c2de)  
abc10.0430.188
(ref = British identity)  
Northern Irish identity0.1340.201
Unionist–nationalist scale (1–5)0.339**0.121
(ref = direct rule)  
Pro-devolved administration−0.1770.191
Pro-power sharing (1–5)0.0770.081
Pro-tax and spend (1–3)−0.2280.141
Pro-EU integration (1–5)0.0810.078
More positive evaluation of SDLP than0.552***0.046
Sinn Féin in terms of catch-all function  
Constant0.1920.505
n455 
Adjusted r-square0.28 

Catholic Support for Rival Bloc Parties

Figure 4a and b show that the distribution of Catholic likelihood of voting for the DUP and UUP is very similar and largely negative. The amount of variation in Catholics' relative support for rival bloc parties (Figure 4c) is considerably less than the Protestant equivalent (Figure 2c discussed above). More even distributions of opinion are evident in relation to the catch-all evaluations (Figure 5a and b). The relative catch-all evaluations of the two parties are illustrated in Figure 5c.

figure

Figure 4. Catholics' Likelihood of Voting for Rival Bloc Parties

Note: The more positive the score, the greater the likelihood of voting UUP rather than DUP.

Download figure to PowerPoint

figure

Figure 5. Catholics' Evaluation of Catch-All Ability of Rival Bloc Parties

Note: The more positive the score, the more positive the perceived catch-all ability of the UUP compared to the DUP.

Download figure to PowerPoint

Does variation in Catholics' evaluations of the relative ethnic catch-all ability of rival bloc parties predict Catholics' relative likelihood of voting for the UUP and DUP?14 The simple answer is no. The regression analysis – see Table 2 – shows no statistically significant relationship. Hence, Hypothesis 2 is not confirmed.

Table 2. Predicting Catholics' Propensity to Vote UUP Rather than DUP: OLS Regression
 coeff.se
(ref = female)  
Male−0.0670.238
(ref = 45 plus)  
Younger than 450.2220.245
(ref = lower than degree)  
Degree−0.1740.265
(ref = c2de)  
abc10.3630.248
(ref = Irish identity)  
Northern Irish identity0.1560.298
Unionist–nationalist scale−0.0940.166
(ref = united Ireland)  
Pro-devolved administration−0.2300.287
Pro-power sharing (1–5)−0.1010.138
Pro-tax and spend (1–3)0.2870.176
More positive evaluation of UUP than0.1290.077
DUP in terms of catch-all function  
Constant0.2171.063
n267 
Adjusted r-square0.041 

Discussion and Conclusion

  1. Top of page
  2. Abstract
  3. Ethnic Catch-All Voting
  4. Data
  5. Results
  6. Discussion and Conclusion
  7. References
  8. Biography

The asymmetric findings are interesting. Protestants clearly differentiate the two nationalist parties; they are considerably more likely to vote SDLP than Sinn Féin and they are also considerably more likely to view the SDLP (rather than Sinn Féin) as capable of representing both communities. Crucially, catch-all evaluations play a strong role in predicting relative propensity to vote for the two nationalist parties. In contrast, Catholics clearly do not differentiate much between the DUP and UUP in terms of likelihood of voting for either party, and actually slightly favour the DUP over the UUP in terms of the catch-all role. Crucially, catch-all evaluations play no role in explaining relative propensity to vote for the two unionist parties. Protestants' views of nationalist parties are thus very different from Catholics' views of unionist parties. For Protestants the two nationalist parties are very different; for Catholics the two unionist parties are very similar.

One possible explanation of the asymmetric results is that the historical differences between Sinn Féin and the SDLP have been of a much greater magnitude than the historical differences between the UUP and DUP, hence leading to Protestants clearly differentiating between the two nationalist parties and Catholics not clearly differentiating between the two unionist parties. In terms of attitudes to political violence and determination to achieve long-term constitutional goals, several analysts have noted that the Sinn Féin/SDLP distinction is more potent than the DUP/UUP distinction (Coakley, 2008; McGarry and O'Leary, 2009). The paramilitarism and irredentism of Sinn Féin in the pre-peace process period arguably enabled the SDLP to develop a particularly clear reputation as a moderate party; the SDLP's abhorrence of violence rendered it a qualitatively different kind of nationalist party from Sinn Féin. In contrast, the differences between the UUP and DUP were arguably a matter of degree (of unionism) rather than being based on fundamentally different positions regarding the acceptability of political violence that distinguished nationalist parties. Somewhat speculatively, this suggests a re-sketching of Figure 1 (as presented in Figure 6).

figure

Figure 6. An Asymmetric Sketch of Party Position

Note: The graph shows the movement in a moderate direction over time of the ‘extreme’ parties in each bloc (Sinn Féin and the DUP) but the initial position of Sinn Féin was particularly extreme.

Download figure to PowerPoint

The asymmetric finding may also be consistent with recent analysis of vote transfers in Northern Ireland's elections held under the Single Transferable Vote system. Sydney Elliot (2009), for example, finds evidence of cross-community transfers from the main unionist parties and from the main nationalist parties. However, the former tend to be stronger than the latter, and the transfers to the SDLP from the UUP were the most sizeable. Elliot analysed transfers in 1998, 2003 and 2007 Assembly elections and, in each case, reports higher unionist to nationalist transfers than the reverse, and particularly UUP to SDLP transfers in all three elections (noteworthy, also, is the DUP to SDLP transfers which were non-trivial in 2007) (Elliot, 2009, pp. 125–30).

The more general implication of the asymmetric findings is that they act as a caution against assuming symmetry in party competition in deeply divided places. In each competing bloc in a deeply divided system, there will be a party with the staunchest views and these parties are often described similarly as the ‘extreme’ parties in the system in contrast to the ‘moderate’ parties (those parties in each bloc with less staunch views). However, the two so-called ‘extreme’ parties may be looked upon very differently by citizens in the opposing community. The evidence in this article suggests that one community may clearly differentiate between rival bloc parties in terms of how likely they would be to support the different parties and this variation in support is driven by judgements about relative capacity of the parties to represent all groups. In contrast, another community may see very little difference between the so-called ‘moderate’ and ‘extreme’ parties in the opposing bloc, and this may perhaps be accounted for by the fact that the differences between opposing bloc parties were never so great (for example, there was not a distinction between a violent and non-violent party) and hence differences between the parties in terms of their relative credibility regarding the catch-all function were never likely to be large or electorally potent.

A final point relates to methodological implications. Most analysts of electoral behaviour use categorical vote choice as the dependent variable. Here, it is potential voting that was focused on. The survey questions asked how likely it is that a respondent would ever vote for each of the parties in the system, on a 10-point scale. This nuanced question has been used as the dependent variable in voting studies by a growing number of researchers (for overview, see Van der Eijk and Franklin, 2009). It is a particularly useful question in a deeply divided place because it captures the difference between some voters who would never vote for a particular ‘rival’ ethnic party and other voters who possibly would. Identifying the extent of potential cross-bloc party support and understanding the motivations of potential cross-divide voters is a prerequisite for understanding the possibilities for electoral change in deeply divided contexts. It is extremely difficult to conduct such an investigation with the category-based conventionally operationalised dependent variable in voting studies, given that such a variable does not directly capture variation in the propensity to vote for parties. Hence, the more widespread use of the nuanced measures used in this article is advocated in order to aid understanding of electoral politics in deeply divided places.

Notes

John Garry was a British Academy Mid-Career Fellow 2012–13, a position which very generously funded research leave to conduct research on electoral behaviour under consociational conditions.

  1. 1

    See Stokes' (1963) critique of Downs (1957).

  2. 2

    Given that the focus in this article is on voting across the divide, the primary focus is not on the bi-confessional Alliance Party which since its foundation has sought to straddle the ‘divide’ and represent both communities. Given its origins, Alliance can be regarded as a fully fledged ethnic catch-all party rather than as a potentially emerging catch-all party.

  3. 3

    See, for example, Evans and Duffy, 1997; Garry, 2009; 2012; Mitchell et al., 2009; Tilley et al., 2008.

  4. 4

    The SDLP and Sinn Féin are to the left of the UUP and DUP in the candidate survey and the expert survey (Benoit and Laver, 2006, p. 277; Gilland Lutz and Farrington, 2006, p. 730; see also the discussion in Garry, 2009).

  5. 5

    The company has conducted a number of election studies in Northern Ireland including the 2004 and 2009 European Parliament election studies; for details see, respectively, Garry, 2007 and Garry, 2012.

  6. 6

    Full details of the survey can be found at: http://www.ark.ac.uk/nilt/

  7. 7
  8. 8

    Gschwend (2005, p. 88) argues that in order to demonstrate the validity of using a quota-based sample (rather than a random selection-based sample) ‘scholars should gather as much external evidence as possible to argue that their achieved sample represents the population on as many dimensions as possible. The more evidence they are able to compile, the more confidence there is that their estimation results are robust even based on quota sample data’. To facilitate transparency and replication, the data set in SPSS format will be made publicly available, as will the syntax file detailing all coding decisions and analysis commands.

  9. 9

    See, for example, Van der Brug et al., 2007; Van der Eijk, 2002; Van der Eijk and Franklin, 1996; 2009; Van der Eijk et al., 2006.

  10. 10

    See, in particular, Van der Eijk and Franklin, 1996.

  11. 11

    This variable construction follows previous research using these 1–10 ‘party support’ measures in the Northern Ireland context (focusing on within-community voting in the nationalist bloc) (see Garry, 2009 p. 463 fn 21). Another (much more widely used) approach to using these dependent variables in models of voting behaviour involves creating a stacked data matrix in which each respondent appears as a case a number of times in the data set, once each for the range of parties in the system (the data set is made up of party*respondent records) (see Van der Eijk and Franklin, 1996). The stacked data set approach facilitates analysis of voting behaviour in a single model as opposed to separate modelling of each party, hence allowing generalisations to the system level. The stacked data set approach has been used in the Northern Ireland setting (Garry, 2007). Here, the ‘relative party support’ measures (as per Garry, 2009) are used as they are the simplest way of testing the precise hypotheses in this article.

  12. 12

    For a recent analysis of the socio-economic bases of voting in Northern Ireland, see Evans and Tonge, 2009.

  13. 13

    One of the ethnonational positional control variables is highly statistically significant: the less strongly unionist a Protestant is the higher the likelihood is of him or her voting SDLP rather than Sinn Féin. Hence, moderate unionists are likely to differentiate between the parties and offer a relatively high ‘likelihood to vote’ score to the SDLP. Also, younger Protestants are more likely than older Protestants to prefer the SDLP to Sinn Féin.

  14. 14

    As far as the Alliance Party is concerned, respondents rated Alliance relatively positively in terms of being able to represent both communities (mean 5.68, se = 0.08, n = 808), statistically significantly higher than the mean scores given to any of the other parties from all respondents (mean scores of other parties: SDLP [5.32], DUP [4.71], Sinn Féin [4.36] and UUP [4.36]). A regression model was also run in which all survey respondents were included and, controlling for religion, age, education, social class, attitudes to power sharing and satisfaction with the power-sharing government, the ethnic catch-all function statistically significantly predicted potential support for the Alliance.

References

  1. Top of page
  2. Abstract
  3. Ethnic Catch-All Voting
  4. Data
  5. Results
  6. Discussion and Conclusion
  7. References
  8. Biography
  • Benoit, K. and Laver, M. (2006) Party Policy in Modern Democracies. London: Routledge.
  • Coakley, J. (2008) ‘Ethnic Competition and the Logic of Party System Transformation’, European Journal of Political Research, 47 (6), 766793.
  • Downs, A. (1957) An Economic Theory of Democracy. New York: Harper Collins.
  • Elliot, S. (2009) ‘The Electoral Dynamics of the Belfast Agreement’, in B. Barton and P. Roche (eds), The Northern Ireland Question: The Peace Process and the Belfast Agreement. Basingstoke: Palgrave Macmillan, pp. 105130.
  • Evans, E. and Duffy, M. (1997) ‘Beyond the Sectarian Divide: The Social Bases and Political Consequences of Nationalist and Unionist Party Competition in Northern Ireland’, British Journal of Political Science, 27 (1), 4781.
  • Evans, J. and Tonge, J. (2009) ‘Social Class and Party Choice in Northern Ireland's Ethnic Blocs’, West European Politics, 32 (5), 10121030.
  • Garry, J. (2007) ‘Making Party Identification More Versatile’, Electoral Studies, 26 (2), 346358.
  • Garry, J. (2009) ‘Consociationalism and Its Critics: Evidence from the Historic Northern Ireland Assembly Election of 2007’, Electoral Studies, 28 (3), 458466.
  • Garry, J. (2012) ‘Consociationalism, Regional Integration and Vote Choice: Northern Ireland and the 2009 European Parliament Election’, Acta Politica, 47, 113127.
  • Gilland Lutz, K. and Farrington, C. (2006) ‘Alternative Ulster? Political Parties and the Non-constitutional Policy Space in Northern Ireland’, Political Studies, 54 (4), 715742.
  • Gormley-Heenan, C. and Mac Ginty, R. (2008) ‘Ethnic Outbidding and Party Modernisation: Understanding the Democratic Unionist Party's Electoral Success in the Post-Agreement Environment’, Ethnopolitics, 7 (1), 4361.
  • Gschwend, T. (2005) ‘Analysing Quota Sample Data and the Peer-Review Process’, French Politics, 3, 8891.
  • Horowitz, D. (1985) Ethnic Groups in Conflict. Berkeley CA: University of California Press.
  • Kirchheimer, O. (1966) ‘The Transformation of the Western European Party Systems’, in J. LaPalombara and M. Weiner (eds), Political Parties and Political Development. Princeton NJ: Princeton University Press, pp. 177200.
  • McGarry, J. and O'Leary, B. (2009) ‘Power Shared after the Death of Thousands’, in R. Taylor (ed.), Consociational Theory: McGarry/O'Leary and the Northern Ireland Conflict. London: Routledge, pp. 1584.
  • McLoughlin, P. (2009) ‘The SDLP and the Europeanization of the Northern Ireland Problem’, Irish Political Studies, 24 (4), 603619.
  • Mitchell, P. and Evans, G. (2009) ‘Ethnic Party Competition and the Dynamics of Power Sharing in Northern Ireland’, in R. Taylor (ed.), Consociational Theory: McGarry/O'Leary and the Northern Ireland Conflict. London: Routledge, pp. 146164.
  • Mitchell, P., Evans, G. and O'Leary, B. (2009) ‘Extremist Outbidding in Ethnic Party Systems is Not Inevitable: Tribune Parties in Northern Ireland’, Political Studies, 57 (2), 379421.
  • Rabuska, A. and Shepsle, K. (1972) Politics in Plural Societies: A Theory of Democratic Instability. Columbus OH: Charles E. Merrill.
  • Sanders, D., Clarke, H. D., Stewart, M. C. and Whiteley, P. (2011) ‘Downs, Stokes and the Dynamics of Electoral Choice’, British Journal of Political Science, 41 (2), 287314.
  • Stokes, D. (1963) ‘Spatial Models of Party Competition’, American Political Science Review, 57 (2), 368377.
  • Tilley, J., Evans, G. and Mitchell, C. (2008) ‘Consociationalism and the Evolution of Political Cleavages in Northern Ireland, 1989–2004’, British Journal of Political Science, 38 (4), 699717.
  • Van der Brug, W., Van der Eijk, C. and Franklin, M. (2007) The Economy and the Vote: Economic Conditions and Elections in Fifteen Countries. Cambridge: Cambridge University Press.
  • Van der Eijk, C. (2002) ‘Design Issues in Electoral Research: Taking Care of (Core) Business’, Electoral Studies, 21, 189206.
  • Van der Eijk, C. and Franklin, M. (1996) Choosing Europe? The European Electorate and National Politics in the Face of Union. Ann Arbor MI: University of Michigan Press.
  • Van der Eijk, C. and Franklin, M. (2009) Elections and Voters. Basingstoke: Palgrave Macmillan.
  • Van der Eijk, C., Van der Brug, W., Kroh, M. and Franklin, M. (2006) ‘Rethinking the Dependent Variable in Voting Behavior: On the Measurement and Analysis of Electoral Utilities’, Electoral Studies, 25 (3), 424447.
  • Wolinetz, S. (2002) ‘Beyond the Catch-All Party: Approaches to the Study of Parties and Party Organization in Contemporary Democracies’, in J. Linz , J. Montero and R. Gunther (eds), The Future of Political Parties. Oxford: Oxford University Press, pp. 136165.

Biography

  1. Top of page
  2. Abstract
  3. Ethnic Catch-All Voting
  4. Data
  5. Results
  6. Discussion and Conclusion
  7. References
  8. Biography
  • John Garry is Senior Lecturer in Comparative Political Science in the School of Politics, International Studies and Philosophy at Queen's University Belfast. He is the author of a range of articles on voting behaviour and public opinion formation. John Garry, School of Politics, International Studies and Philosophy, Queen's University Belfast, 25 University Square, Belfast BT7 1PB, Northern Ireland, UK; email: j.garry@qub.ac.uk