Explaining extreme right support: where's the campaign?

The wider literature on populist radical and extreme right-wing support often draws on one of three relatively distinct theoretical approaches. According to an older and less persuasive sociological account, electoral support for these parties is interpreted mainly as a response to socio-structural change, mainly through the decline of heavy industries and the manufacturing sector, and the onset of a global post-industrial economy. Such changes, it is argued, render economically deprived, working class urban areas that once relied heavily on manufacturing increasingly receptive to campaigns that pledge to halt or reverse these changes, and ‘punish’ mainstream elites who are held responsible (Betz, Reference Betz1994; Kriesi et al., Reference Kriesi, Grande, Lachat, Dolezal, Bornschier and Frey2006). Past aggregate and multi-level research in Britain is generally supportive, suggesting that right-wing extremist parties poll strongest in urban and deprived working class districts, where average education levels are low and there are large numbers of skilled workers and ageing populations (Bowyer, Reference Bowyer2008; Ford and Goodwin, Reference Ford and Goodwin2010; Cutts et al., Reference Cutts, Ford and Goodwin2011).

At the same time, however, existing research also demonstrates that once attitudinal or ideological considerations are introduced the sociological model emerges as only a poor predictor of support for the populist radical and extreme right. In short, while necessary these sociological considerations can only take us so far in explaining electoral support for these types of parties (Norris, Reference Norris2005; Van der Brug et al., Reference Van der Brug, Fennema and Tillie2005). The emphasis on attitudinal drivers of support, though in particular xenophobic public hostility toward immigration, has led several studies to focus on a second alternative theoretical approach that is rooted in group conflict theories (Blumer, Reference Blumer1958; Bobo, Reference Bobo1983). Seen from this perspective, extreme right party support will be strongest in areas where there are larger than average numbers of immigrants and/or members of settled minority groups, with such groups triggering feelings of actual or perceived ‘threat’ among residents, whether in regard to scarce economic resources (i.e. jobs, social housing, etc.) or more symbolic concerns over threats to values, national identity, and ways of life (Ivarsflaten, Reference Ivarsflaten2005; McLaren and Johnson, Reference McLaren and Johnson2007; Rink et al., Reference Rink, Phalet and Swyngedouw2009; Goodwin, Reference Goodwin2010). More recent studies have built on these findings to demonstrate the crucial importance of disaggregating immigrant and minority groups; support for the extreme right in countries such as Belgium and Britain is significantly and positively associated with the presence of large Muslim communities, a relationship that does not appear to hold in regard to non-Muslim groups (Coffé et al., Reference Coffé, Heyndels and Jan2007; Bowyer, Reference Bowyer2008; Ford and Goodwin, Reference Ford and Goodwin2010).

However, the above approaches tend to portray the electorates of the populist radical and extreme right as largely passive, with voters either switching to these parties in response to the effects of destabilizing socio-economic change, or to the arrival of highly salient issues, such as immigration or Islam. Like parties more generally, those on the extreme right-wing may assume an active role in attempting to ‘get out the vote’ at elections. Yet while the question of ‘how’ these parties seek to mobilize support has received attention (Kitschelt, Reference Kitschelt1995; Mudde, Reference Mudde2007; Zaslove, Reference Zaslove2008), the subsequent question of whether these campaigns have a demonstrable impact on electoral performance is largely ignored. This lack of research is striking given that there are good reasons to expect that campaigns can have important effects. In the wider literature on party politics, the proposition that campaigns can have a clear and significant impact on support is now the ‘accepted truth’, with numerous studies demonstrating how – with all things equal – the harder a party campaigns the greater its share of the vote (Denver and Hands, Reference Denver and Hands1997; Fieldhouse and Cutts, Reference Fieldhouse and Cutts2009; Pattie and Johnston, Reference Pattie and Johnston2009). Furthermore, broader trends such as the declining importance of party identification, the growing professionalization of parties, and the increased hesitancy and tendency of voters to make choices closer to elections underscore the potential for local campaigning to exert a significant impact on electoral performance (Cutts, Reference Cutts2006; Fisher et al., Reference Fisher, Cutts and Fieldhouse2012). To examine the validity of these divergent approaches, and investigate the electoral impact of extreme right party campaigning, we now turn to explore the context of the 2010 British general election.

Electoral context: the BNP and the 2010 general election

While the extreme right in Britain is associated with failure, over the period 2001–2010 the BNP achieved a series of advances in second-order elections to local, devolved, and European public office. As part of a ‘ladder strategy’ the party had prioritized local implantation and activism as important prerequisites to achieving parliamentary representation. From 2005, the party sought to bolster rates of activism among members by implementing a ‘voting membership scheme’ (modelled on a scheme developed by the Sweden Democrats), which provided various incentives for activism and financial donations (Goodwin, Reference Goodwin2011). As membership of the BNP approached 14,000, electoral support for the party was concentrated most heavily in the North West, Yorkshire, Midlands and South East regions of England, and in areas where local politics had long been dominated by the centre-left Labour Party (Ford and Goodwin, Reference Ford and Goodwin2010).

At the 2010 general election, the party sought to build on these gains by securing either elected representation in Westminster or second place finishes in its two target seats of Barking in outer-east London, and Stoke Central in the Midlands. Fielding a record 338 candidates, the party aimed to secure a larger share of the national vote than the radical right-wing UK Independence Party (UKIP), and emerge as the fourth largest party in British politics. From the outset, however, the BNP campaign was undermined by internal problems that included television footage of a BNP councillor brawling in the streets with Asian youths, allegations that a party official had made death threats to BNP Chairman Nick Griffin, the temporary closure of the BNP website by a disgruntled activist, and factionalism in local branches in Bradford, Kirklees, and Stoke-on-Trent. Most of these problems stemmed from wider discontent among the grassroots that was triggered by changes to the BNP's ethnically defined ‘whites-only’ membership, the employment of a business consultant to help ‘professionalize’ the party's internal operations, and allegations that Nick Griffin was politically incompetent following an appearance on the popular television programme, Question Time.

Aside from internal problems, the BNP campaign also met strong opposition in the core seat of Barking from Labour and anti-fascist activists. The latter ‘Hope Not Hate’ network claimed to recruit ∼142,000 supporters and 20,000 regular online campaigners to support a highly visible and active campaign against the BNP (Lowles, Reference Lowles2010). This focused on mobilizing groups traditionally under-represented among the extreme right electorate, such as employing less aggressive and more positive anti-BNP themes to rally women and minority ethnic voters. Though difficult to assess, typical of this activity was one ‘day of action’ that claimed to mobilize over 500 volunteers to distribute over 90,000 newspapers across Barking, Dagenham, and Havering (while on polling day it was claimed that over 150 volunteers mobilized 6000 non-BNP voters in key areas). This anti-BNP activity was evident in national media, such as the tabloid Mirror newspaper that ran a story entitled ‘five reasons not to vote BNP’.

Yet beyond Barking, the wider issue agenda of the 2010 general election remained favourable for the extreme right (Ford, Reference Ford2010). Since 2001, increased British public concern over immigration, rising ethnic diversity, and dissatisfaction with the three major parties enlarged electoral potential for anti-immigrant and populist parties. As several studies highlighted, anti-immigrant sentiment, anxiety over settled Muslims, and the increased importance of security-related issues had altered the broader issue agenda in British politics (McLaren and Johnson, Reference McLaren and Johnson2007; Clarke et al., Reference Clarke, Sanders, Stewart and Whiteley2009; Voas and Ling, Reference Voas and Ling2010). This remained evident during the 2010 campaign, where the issue of immigration was the second most important issue for voters, dominated one of the televised leadership debates, and attracted further attention when incumbent Labour Prime Minister Gordon Brown described a Labour voter who had raised concerns about immigration as a ‘bigot’.Footnote ⁵

On the supply-side, the BNP sought to take advantage of ‘Bigotgate’ by distributing a tailored newspaper in its target seats of Barking and Stoke Central, as well as parts of Barnsley, Leicestershire, and Manchester. More broadly, the BNP campaign sought to mobilize support by emphasizing traditional extreme right themes: an immediate halt to further immigration; the deportation of illegal immigrants; and the enactment of a voluntary repatriation scheme, whereby ‘immigrants and their descendants are afforded the opportunity to return to their lands of ethnic origin, assisted by generous financial incentives both for individuals and for the countries in question’ (BNP, 2010: 20).

In contrast to previous campaigns, stronger emphasis was placed on mobilizing anti-Muslim sentiment by devoting an entire section of the manifesto to ‘confronting the Islamic colonisation of Britain’ (BNP, 2010: 5). BNP policies included a ban on the burka, ritual slaughter, the building of further mosques, and immigration from Muslim countries. They also advocated the immediate deportation of radical Islamist preachers, their supporters, and ‘any other members of their community who object to these reasonable security measures’ (BNP, 2010: 5; see also Copsey, Reference Copsey2012). While the shift toward anti-Muslim sentiment was partly a reflection of the broader issue agenda, it also marked a continuation of the BNP's transition away from crude biological racism toward the ethno-pluralist doctrine (Goodwin, Reference Goodwin2010). Yet the BNP was not the only organization attempting to mobilize anti-Muslim prejudice. Since June 2009, the EDL had been staging confrontational demonstrations against militant Islamism, and attempting to mobilize opposition to the perceived threat from Islam, though not in the electoral arena. Indeed, one online survey suggests that the BNP is the favoured political party among EDL sympathisers (Demos, 2011).

Data and measurement: capturing campaign effects

To examine the electoral impact of campaigning on extreme right party support, we employ a range of aggregate and individual-level data. Table 1 reports summary statistics for all the variables used in the analysis. Our party campaign data are derived from several sources. First, at least since the 1980s political scientists have demonstrated that the more a political party spends on a campaign relative to its opponents, the stronger its electoral performance (Johnston and Pattie, Reference Johnston and Pattie1995, Reference Johnston and Pattie2008, Reference Johnston and Pattie2011; Pattie and Johnston, Reference Pattie and Johnston2009). As a surrogate measure of campaign effort, and when measured against individual-level data and other self-reported measures of party campaign activity, spending data have an established record of reliability, largely universal coverage, and validity. As in other studies that utilize spending data, we express the amount spent in each constituency as a percentage of the maximum possible amount allowed. Spending data from the ‘official’ or ‘short’ campaign are also utilized. These data reveal that at the 2010 general election the BNP devoted a total of £159,388 to the short campaign, and hence was unable to compete financially with the three main parties or more affluent minor parties such as UKIP. Over the entire campaign, the average BNP candidate spent only £750, compared with an average spend of £1278 per UKIP candidate, and £1839 per Green Party candidate (Johnston and Pattie, Reference Johnston and Pattie2011).

Table 1 Variable definitions and descriptive statistics

BNP = British National Party; UKIP = UK Independence Party; EDL = English Defence League.

Our second measure of underlying campaign effects is the extent of local electoral success by the extreme right prior to the 2010 campaign. Some scholars argue that implantation at the local level is important in determining whether or not populist radical and extreme right parties can sustain electoral success. Mudde (Reference Mudde2007), for example, contends: ‘Undoubtedly, the most important factor to decide whether or not a party fails or succeeds in persisting electorally is party organization and local implantation’. Gaining local office may enhance prospects of a wider breakthrough by raising the party's profile, promoting an image of electoral credibility, and fostering relations with voters. Local elected representation might also improve the prospect of wider success ensuring that basic electoral duties and party-voter contact is undertaken. Indeed, it is for these reasons that the BNP prioritized securing local council seats as a prerequisite to electoral success at the national level. While there is some evidence to suggest that the BNP has advanced in areas where it has established a local presence [Joseph Rowntree Charitable Trust (JRCT), 2004; Borisyuk et al., Reference Borisyuk, Rallings, Thrasher and van der Kolk2007], the precise impact of this local success remains largely unknown. Therefore, we hypothesize that local electoral success, which we measure as the BNP winning at least one local council seat in the constituency since 2005, will have a significant positive effect on the party's performance at the 2010 general election.

Our third measure examines the impact of membership on electoral performance. It is plausible to expect that where a party has larger clusters of members it is more visible at the local level, has stronger infrastructures, and higher rates of activism (Seyd and Whiteley, Reference Seyd and Whiteley1992; Whiteley and Seyd, Reference Whiteley and Seyd1994). Clearly endogeneity may be inherent in this relationship given that members may be contributing to success while success might also attract new members. Separating out this dynamic is difficult given that it is conceivable that both effects may be operating. Our measure of BNP membership is adjusted to take account of the fact that the BNP operated a ‘whites-only’ membership until a change of policy in 2010. Here we use the list of BNP members that was leaked onto the Internet in 2008 by disaffected activists, and which has been confirmed as genuine by the party leadership. The leaked document contains over 10,000 names and addresses, and includes postcodes that allow individual members to be matched to parliamentary constituencies. The inclusion of BNP membership, together with party spending and prior local election success, provides a thorough measure of local campaign effects at the aggregate level.

A fourth measure included relates to ‘legacy effects’, and the question of whether support for the extreme right at the 2010 election is positively related to previous cycles of extreme right party activism in earlier decades. Two variables are included: the first takes account of where support for the BNP's main predecessor in the 1970s, the National Front, was spatially concentrated; and the second draws on a coding of party literature to take account of where BNP branches were spatially concentrated during the 1980s.Footnote ⁶

However, while the above measures provide important and innovative insights into the impact of campaigning on extreme right support, they are at best surrogate measures. Therefore, to complement our approach and establish whether campaign effects are significant after taking account of attitudinal variables, we utilize BES panel survey data that provides information on party-voter contact during the campaign. Individual exposure to BNP election broadcasts, telephone and door-to-door canvassing, and party ‘knock-up’ are utilized. Even though such data have the advantage of measuring the direct link between campaign exposure and voter behaviour, it is reliant on respondents making accurate and conscious recollections of such exposure. If this assumption is not met and voters are affected by campaigns without attributing that as such, then the impact of campaigns may be underestimated. If campaigns have an effect on only a small number of voters in selected areas, national survey methods are more unlikely to successfully detect the impact. Hence, we complement our evidence from survey data with evidence provided by aggregate data. Therefore, a significant finding at both individual and aggregate level would reveal the importance of examining campaign effects when explaining support for populist radical and extreme right parties.

To address our core question, we first investigate the impact of the BNP campaign on its support by analysing a range of aggregate-level data that are collected from several sources. The campaign data outlined above are combined with 2001 census data for the newly redistricted constituencies in England. Specifically, socio-economic variables are derived from the census. Given that collinearity between these variables was found, a principal components analysis (PCA) was run and two factors were extracted that describe the characteristics of the parliamentary constituencies.Footnote ⁷ Factor 1 captures the urban character of the area and the level of deprivation, whereas Factor 2 reflects the class structure and education levels in areas that depend on the manufacturing sector. Also included is the percentage share of the constituency population that are retired and a measure to take account of migration patterns from outside the United Kingdom. Electors may also vote for the extreme right in response to perceived economic and/or cultural threats posed by ethnic out-groups, and immigration more generally (Klandermans and Mayer, Reference Klandermans and Mayer2006; Tolsma et al., Reference Tolsma, Lubbers and Coenders2008). Yet following recent research, we undertake a more nuanced examination of this relationship by disaggregating minority groups and including measures of the percentage of the constituency population, that is, Muslim, non-Muslim Asian, and Black (Bowyer, Reference Bowyer2008; Ford and Goodwin, Reference Ford and Goodwin2010). Recent studies also reveal a significant northward shift in the geographical concentration of BNP support (Ford and Goodwin, Reference Ford and Goodwin2010; Cutts et al., Reference Cutts, Ford and Goodwin2011). To take account of these regional variations in BNP support, our models include six regional dummies.

To take account of local political context we include a measure of local Labour Party hegemony: local authorities where Labour has held majority-rule since the re-organization of local councils in the early 1970s. Our analysis also includes a number of additional predictors that were largely specific to the 2010 general election. First, we measure the impact of the presence of a minority Black or Asian parliamentary candidate, which might enhance subjective perceptions of threat among local electorates and consequently increase support for the extreme right. We account of this possibility by including ethnic candidate selection for each mainstream party in the analysis. Second, we take account of whether the BNP reaped any electoral rewards from EDL rallies in those seats where the EDL held rallies over the period June 2009–May 2010. Third, recent research suggests that UKIP and the BNP draw from similar constituency bases of support (John and Margetts, Reference John and Margetts2009; Ford et al., Reference Ford, Goodwin and Cutts2012). This suggests that the presence of UKIP support or the decision of the party to field a candidate where the BNP stands would have an impact on BNP support, particularly given that UKIP is seen as being more politically skilful in how it presents contentious policy issues and has the greater credibility to appeal to a wider electorate (John and Margetts, Reference John and Margetts2009; Ford et al., Reference Ford, Goodwin and Cutts2012). Finally, we take account of a parliamentary expenses scandal that dominated media headlines for at least 18 months before the election, and affected each of the three mainstream parties. It appears plausible, therefore, that minor parties such as the BNP benefited from a subsequent rise in anti-establishment sentiment. Here we include two variables: one takes account of constituencies where MPs decided not to seek re-election but claimed moderate or excessive expenses; and the other includes constituencies where the incumbent MP did stand again and had made moderate or excessive expenses claims.

Modelling BNP support

One of the problems in measuring BNP support in the 2010 general election is that this measure is based solely on the number of votes received by a BNP candidate and, therefore, is only observed in those parliamentary constituencies where a BNP candidate is on the ballot. In this case, the dependent variable is censored, making a standard Ordinary Least Squares (OLS) regression inappropriate. A number of scholars have applied a Tobit regression to address the data censoring problem (Jackman and Volpert, Reference Jackman and Volpert1996; Jesuit and Mahler, Reference Jesuit and Mahler2004; Bowyer, Reference Bowyer2008) given that it is particularly useful where a dependent variable has the value of zero for some part of the population and is roughly continuously distributed over positive values (Wooldridge, Reference Wooldridge2002; Cameron and Trivedi, Reference Cameron and Trivedi2005). In the Tobit regression model, the dependent variable is a continuous latent variable, with BNP support in the 2010 general election only observed if y* is greater than the specified threshold (0 in our case to reflect those constituencies where the BNP did not stand and, therefore, got no votes; Long, Reference Long1997; Wooldridge, Reference Wooldridge2002; Bowyer, Reference Bowyer2008).Footnote ⁸ Of course, if these constituencies where the BNP either did not stand or received no votes were omitted from the model, then we would be introducing selection bias into our analysis (Golder, Reference Golder2003; Cameron and Trivedi, Reference Cameron and Trivedi2005). Using the same set of variables, the Tobit regression, therefore, estimates both the probability that the value of the dependent variable is greater than the threshold (in our case 0) and the expected value in those cases where it is observed. Put simply, it estimates the probability that the BNP fields a candidate in a constituency and its percentage 2010 vote share in those constituencies that it did stand. Here, like numerous scholars in the extreme right party (ERP) literature, we adopt this Tobit I estimation technique (Jackman and Volpert, Reference Jackman and Volpert1996; Golder, Reference Golder2003; Swank and Betz, Reference Swank and Betz2003; Jesuit and Mahler, Reference Jesuit and Mahler2004; Bowyer, Reference Bowyer2008).Footnote ⁹

The extreme right campaign: did it matter?

In political and media debate, the 2010 general election was largely framed as a disaster for the extreme right. In the aftermath of the election, it was revealed that the BNP had more than doubled its number of votes to over 564,000 and increased its share of the total vote to 1.9%. However, despite the overall increase in support the party failed to achieve a noticeable breakthrough in its target seats. In Barking, the BNP finished third with 14.8% of the vote.Footnote ¹⁰ It is important to note, however, that in 126 constituencies that were also contested by the BNP at the previous general election in 2005, the party improved its share of the vote by an average of 1.9%. Furthermore, BNP candidates met or surpassed the 5% threshold that is required to retain financial deposits in 73 seats, compared with seven saved deposits at the general election in 2001. BNP candidates polled over 10% in three seats and 8% or over in a further nine seats.

But to what extent did campaigning by the extreme right have an electoral impact? Utilizing the aggregate data that are outlined above, we first examine the factors that shaped BNP support at the 2010 general election and detail the types of areas where the extreme right polled most strongly, and relative to its performance at the previous general election in 2005. The first multivariate Tobit regression model examines the underlying pattern of support for the BNP in the 2010 general election (Table 2) while the second model incorporates prior BNP support in 2005 (Table 3).Footnote ¹¹ In both tables, the first column includes the Tobit beta coefficients that are related to the latent dependent variable (BNP support in 2010) and their respective standard errors. These show the effect of change in a given predictor variable (x) on the expected value of the latent variables, holding all other predictor variables (x) constant. So in terms of the latent variable, y*, these Tobit betas can be interpreted in just the same way as the betas from a standard OLS regression model. Yet while it is common practice in the ERP literature to report and discuss these Tobit coefficients (Jackman and Volpert, Reference Jackman and Volpert1996; Golder, Reference Golder2003; Bowyer, Reference Bowyer2008) – and we report these for brevity and comparability purposes – they are not entirely useful because the dependent variable is unobserved. Alternatively, by reporting other forms of marginal effects at the means of the independent variables, it is possible to measure the substantive significance of these coefficients and more intuitively determine the linear relationship between our predictor variables and BNP vote share in the 2010 general election. In the second column, we therefore report the marginal effects for the conditional expectation of y, given that y is positive (greater than zero) and the unconditional expected value of the observed dependent variable (the probability that y is positive times the expected value of y given that y is indeed positive).Footnote ¹²

Table 2 Tobit model and marginal effects of BNP (2010) General Election support in England

BNP = British National Party; UKIP = UK Independence Party.

Notes: We only model England because we don't have membership and legacy variables in Wales and Scotland. It was decided to create two factors from eight socio-economic variables. No qualification was included here because this was 0.84** correlated with semi-routine and routine occupations (working class).

**P<0.05; *P<0.10.

Table 3 Tobit model and marginal effects of BNP (2010) General Election support in England accounting for BNP prior support

BNP = British National Party; UKIP = UK Independence Party.

Notes: We only model England because we don't have membership and legacy variables in Wales and Scotland. It was decided to create two factors from eight socio-economic variables. No qualification was included here because this was 0.84** correlated with semi-routine and routine occupations (working class).

**P<0.05; *P<0.10.

Table 2 reports the first set of Tobit estimations.Footnote ¹³ Consistent with our hypotheses, the extreme right polled strongest in working class areas that are more dependent than others on a weakened manufacturing sector, and have large numbers of residents who lack educational qualifications. The party also performed better in those constituencies where there are larger Muslim communities, although this effect does lose its significance once we introduce the prior BNP vote in 2005. Moreover, the BNP improved its vote share in London – where its leader Nick Griffin contested the constituency of Barking and attracted considerable publicity – and also the Yorkshire and Humber region, where the party has been especially active at the local level since 2001 (Goodwin, Reference Goodwin2011). Yet after controlling for these demand-side factors, and also factors that are specific to the 2010 general election, the findings underscore how party activism and campaigning had a positive electoral impact on BNP support. The marginal effects conditional on being uncensored indicate that a 1% increase in BNP campaigning (party spending) results in a 0.11% increase in BNP support. Opposition campaigning, particularly the Conservatives and Liberal Democrats, did harm BNP support, although the reported marginal effects suggest that they had a largely negligible impact.

The Tobit regression results also suggest that prior local electoral success had a significant and positive impact on BNP support at the general election.Footnote ¹⁴ In seats where the extreme right party had previously enjoyed success at local level, it polled stronger. While establishing local credibility is an important task for minor parties generally, our results indicate that the BNP's strategy of community-based activism has played an important role in the party's ability to gain support.

There is also substantial evidence that the BNP performed stronger in constituencies where the party had larger clusters of members. The marginal effects conditional on being uncensored indicate that a 1% increase in party membership results in a 1.6% increase in BNP support. Membership acts as a proxy for activism but it is undoubtedly intertwined with local electoral success. The presence of large numbers of members (particularly in Labour-held seats) is likely to encourage higher rates of activism that, in turn, improve the prospects of local success and strengthen general election campaigns. Electoral breakthroughs at the local level are also likely to attract media publicity, foster an image of credibility and attract new recruits. While both effects appear to be operating in a ‘virtuous circle’, it is evident that local activism boosts local representation, which in turn heightens electoral profile and legitimacy. Therefore, it appears that in 2010 the BNP reaped electoral dividends where the extreme right party established an image of local credibility. Reflecting BNP activism in Labour strongholds, the party also improved its vote in seats where Labour has dominated power at the local level since 1973. In contrast, we find no evidence that the party performed stronger in areas where the extreme right had previously been electorally active in the 1970s and 1980s. Where UKIP stood a candidate, BNP support increased, although, as expected, BNP support did decline in those constituencies where UKIP polled strongly. Where UKIP was strong the BNP struggled to make headway suggesting that voters perceived the former as a more credible party to support than the BNP (John and Margetts, Reference John and Margetts2009; Ford et al., Reference Ford, Goodwin and Cutts2012).

Consistent with the findings above, even after taking account of the party's prior vote, similar campaigning effects are observed. The Tobit beta estimates and the other marginal effects are reported in Table 3 and remain largely similar. Where the party campaigned more intensely it reaped electoral rewards, while the party seemingly gained a boost in support in those seats with high numbers of members and those constituencies where Labour has dominated power at the local level since 1973. Once the prior BNP vote is controlled for, support for the party in 2010 did not significantly increase in areas where the extreme right had achieved local election success or in areas where the EDL had staged street-based protests and rallies in opposition to perceived threats from Islam and settled Muslim communities. Other 2010 general election-specific factors such as whether the candidate was Asian or Black, or where incumbent MPs either stood down or fought on despite excessive expenses claims also failed to have a significant impact on BNP support. The demand side remained important. Even after controlling for prior support, the reported marginal effects – conditional on being uncensored and unconditional on the expected value of the observed dependent variable – both indicate that the relationship between those living in working class manufacturing areas and BNP support is not only significant but very strong.

Modelling the extreme right vote at the individual level

The Tobit regression models presented above suggest that campaigning by the extreme right had a significant and positive electoral impact on its overall performance. Yet while these findings are informative, they remain constrained by a lack of individual-level data. The importance of encompassing individual-level data is underscored by the observation that it is possible to gauge the effects of campaigns by examining how this activity is perceived by individual voters. Also, past research has demonstrated how support for populist radical and extreme right-wing parties is driven strongly by voters’ instrumental motives and/or their ideological affinity with the policies and ideas on offer (Van der Brug et al., Reference Van der Brug, Fennema and Tillie2005; Cutts et al., Reference Cutts, Ford and Goodwin2011). For these reasons, we supplement aggregate-level data with an examination at the individual level.

How important are campaigning effects once we control for a battery of attitudinal predictors on extreme right party support? To address this question we utilize data from the post-election campaign panel of the 2010 BES. Party campaigning is measured by aggregating respondents’ responses to viewing a party election broadcast, being canvassed, receiving direct mail and being contacted by e-mail, text message, social networking sites, and ‘knocking-up’ on polling day.Footnote ¹⁵ These were evaluated and included separately for all four parties. In addition, we also test for the influence of attitudinal motivations on support for the extreme right. First, we measure anti-establishment sentiment that was heightened by an outbreak of a parliamentary expenses scandal in 2009 by assessing respondents’ perceptions of trust in politicians. Following the wider literature, we might expect that citizens who are more distrustful of politicians are more likely to support the extreme right. Second, given the BNP's programmatic emphasis on it, we include Euroscepticism that is measured by opposition to closer co-operation with the European Union (Hobolt et al., Reference Hobolt, Spoon and Tilley2009). Third, we examine whether feelings of economic insecurity as a result of the financial crisis motivated citizens to vote for the BNP (Citrin et al., Reference Citrin, Green, Muste and Wong1997). Fourth, we measure concern over the core extreme right issue of immigration as those who perceived this to be the most important issue facing the country (Van der Brug et al., Reference Van der Brug, Meindert and Jean2000, Reference Van der Brug, Fennema and Tillie2005; Rydgren, Reference Rydgren2008; Cutts et al., Reference Cutts, Ford and Goodwin2011). Unfortunately, there were no explicit measures of anti-immigrant sentiment in the campaign panel of the BES. Finally, we also take account of public attitudes toward the war in Afghanistan, opposition to which was a prominent BNP policy at the election, and so we expect those who disapprove of the war to be more liable to support the extreme right.

Table 4 shows the logistic regression of the BNP vote in 2010. A large number of variables perform as expected, with coefficients statistically significant and correctly signed. Our findings provide further evidence that concern over immigration is a key driver of support for the BNP: respondents who ranked immigration as the most important issue facing the country were five times more likely to support the extreme right, than other parties. However, political dissatisfaction and Euroscepticism were also important drivers of support: extreme right voters were far less likely to trust politicians while respondents who supported European integration had a lower probability of supporting the party. In contrast, economic security as a result of the financial crisis, and disapproval with the war in Afghanistan were insignificant. However, these results also underscore the crucial importance of the campaign: even after controlling for these individual attitudinal predictors, the BNP campaign still mattered. Respondents who said they were contacted by the BNP were 1.2 times more likely to vote for the party. While the BNP campaign had a positive and significant impact on the party's vote, campaigning by all other political parties appears to be ineffective in reducing the probability of voting BNP.

Table 4 Binomial Logit of BNP voting in the 2010 General Election in England

BNP = British National Party; UKIP = UK Independence Party; BES = British Election Study.

Source: 2010 BES Campaign Post-Election Panel.

Notes: Weighted data using post-election weight. Robust clustered standard errors (182 clusters). Binomial logit analysis of voting for the BNP vs. voting for all other parties.

**P-value 0.05.

Discussion and conclusions

Despite the fact that campaigning is integral to elections and their outcomes, the literature on populist radical and extreme right-wing parties reveals little about the electoral impact of campaigning by these types of parties. Albeit restricted to one case, our analysis nonetheless provides evidence that local campaigns run by the extreme right can have significant and positive electoral effects, and even in a case that is traditionally associated with electoral failure. By utilizing a combination of innovative data, we find that support for an extreme right party at the 2010 British general election was significantly higher in areas that were targeted by its campaigns: where the party ran intensive local campaigns, recruited larger clusters of members, had a history of success at local elections, and where local politics were historically dominated by the centre-left, support for the extreme right at a general election was significantly higher. There is also little evidence that the BNP benefited electorally in areas that had been targeted by the streets-based EDL social movement, suggesting that different forms of right-wing extremist politics may not be benefiting from each other in symbiotic fashion. More broadly, and consistent with past research, we find that electoral support for the extreme right is strongest within working class areas that are more dependent than others on the manufacturing sector and are characterized by low average education levels.

These findings provide evidence that, like other types of parties, where those on the extreme right-wing target their resources, build local electoral support and nurture their memberships they are able to increase their electoral dividends. Even in a case like the BNP, a party that has failed to mobilize a broad and stable coalition of voters, by investing in local campaigning and targeting particular types of areas the party significantly improved its electoral performance. Our aggregate level findings are confirmed at the individual level after controlling for a battery of established attitudinal predictors of extreme right voting. Those contacted by the BNP were significantly more likely to vote for the party, while campaigning by all other political parties was ineffective in reducing the probability of voting BNP.

Interestingly, however, and despite this positive electoral impact on support, in the aftermath of the election the BNP's failure to achieve a visible breakthrough fuelled internal factionalism. This culminated in an (unsuccessful) leadership challenge and led many key organizers to abandon the BNP, switch allegiance to the English Democrats, or establish new organizations such as British Freedom. In the aftermath of the BNP's failure to achieve a wider national breakthrough at the 2010 general election, and following the party's decade-long experimentation with a vote-seeking strategy, the extreme right milieu in Britain has become increasingly fragmented, with much attention now focusing instead on the non-electoral social movement, the EDL.