Life beyond the ballot box: the political participation and non-participation of electoral abstainers

ABSTRACT

Drawing on electoral participation and social movement studies, we develop a typology of abstainers on the basis of their forms of non-electoral participation, and explore the determinants that drive belonging to each of these sub-groups. Although there is a positive correlation between electoral turnout and non-electoral participation, through applying latent class analyses and regressions we find that there is a subset of abstainers who decide not to cast a vote but take part in non-electoral political activities. These ‘alternative voicers’ are critical of the institutional system and do not feel represented by it, but they are politically involved. Based on their patterns of non-electoral participation, we propose a more nuanced typology of alternative voicers (e-activists, super-activists, and consumerists) and explore their drivers relative to other abstainers. We use data from the original LIVEWHAT survey conducted in 2015 across nine European countries (N = 18,367).

KEYWORDS:

• Electoral abstention
• non-electoral participation
• latent class analysis
• European politics

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes on contributors

Lorenzo Zamponi is an assistant professor at the Scuola Normale Superiore in Florence (Italy), where he is part of the COSMOS (Centre on Social Movement Studies) research team. His research interests include memory, contentious politics and media analysis. His publications focus mainly on the recent wave of anti-austerity protest in Europe, on the cultural elements of social mobilisation and on the emergence of non-protest based forms of collective action.

Lorenzo Bosi is Assistant Professor in Political Sociology at the Scuola Normale Superiore. He received his Ph.D. in politics from Queen’s University, Belfast, and is the past recipient of the ECRC (University of Kent), Jean Monnet and Marie Curie (EUI) postdoctorate fellowships. He is a political sociologist pursuing comparative analysis into the cross-disciplinary fields of social movements and political violence.

Martín Portos is Postdoctoral Fellow at the Department of Political and Social Sciences, Scuola Normale Superiore (Florence, Italy). He holds a PhD from the European University Institute. Winner of the Juan Linz Best Dissertation Award in Political Science and the ISA's Seventh Worldwide Competition for Junior Sociologists, he has published on political participation, social movements, inequalities and nationalism.

Notes

1 Despite the slight differences in meaning, we use the terms ‘abstainers’ and ‘non-voters’ interchangeably throughout.

2 Note that for political process-oriented accounts, indicators related to electoral politics, such as party competition, the number of political parties, etc. tend to be considered as indicators of the political system’s degree of openness (Braun and Hutter 2016; Vráblíková 2014). Also, party membership is a usual predictor of individual non-electoral participation (Norris et al. 2005; Dalton et al. 2010).

3 Note that cost-benefit analyses are mediated by loyalty dynamics, i.e. how attached and committed agents remain to the organisation, which might make them opt for sticking to the status quo. When their voice is heard and they can reform the organisation, loyal members will be particularly devoted to organisational success.

4 Grant agreement N. 613237. Coordinator: Marco Giugni (University of Geneva). Funding institution: European Commission (7th Framework Programme).

5 In order to ensure that the right people are invited in the right proportions, YouGov conducts its public opinion surveys online using Active Sampling (LIVEWHAT 2014: 388–9). When it recruits a new panel member (through a host of different sources, including strategic partnerships with websites, standard advertising, etc.), YouGov records some socio-demographic information, so it can draw a sub-sample of the panel that is representative of the country’s adult population by using the abovementioned quotas (LIVEWHAT 2014: 388–9). Respondents in this sub-sample are invited to answer the survey— these Internet users can access only once with their username and password, and receive a modest cash incentive (see LIVEWHAT 2014: 388–9). Upon completion, the raw data is accompanied by weights (LIVEWHAT 2014: 388–9). For further information on the sampling and recruitment procedures of the survey, see the LIVEWHAT deliverable and technical appendix (LIVEWHAT 2014; see also https://yougov.co.uk/about/panel-methodology/). In order to access the full LIVEWHAT questionnaire, use the following link: http://www.livewhat.unige.ch/wp-content/uploads/2014/07/Deliverable-4.11.pdf. See also Grasso and Giugni (2016); Giugni and Grasso (2017).

6 In Laponce’s (1967) study, the sub-sample of voluntary non-voters consists of 76 individuals. 24 forced non-voters were likewise identified. The latter are those who could not vote for different circumstances (illness, displaced, etc.).

7 These authors find that factors such as demographic characteristics, general political interest and the electoral context mediate the impact of campaign attention on non-voting (Ragsdale and Rusk 1993).

8 These effects might be mediated by social interactions and relational mechanisms, as Laurison (2016) suggests.

9 Given our data structure (i.e. binary responses), the poLCA R package is used.

10 We have replicated our analyses with a restrictive approach to electoral abstention: abstainers are those who did not vote both in the last national and European elections. Considering abstention at two different types of elections protects our sample against a number of problems. First, EU elections have traditionally been considered second-order in character (Hix and Lord 1997: 87–90; Reif and Schmitt 1980). Citizens often perceive that EU elections have a lower impact in their interests than other types of elections, thus a lower turnout is expected in general. At least in the 1980s-1990s and early 2000s, neither campaigns were dominated nor voters were motivated by the institution that was being elected— the Europarliament— but by the domestic political cycle (Schmitt 2005). More recently, the ‘second order’ nature of purely legislative elections together with the ‘second rate’ features of elections with no clear agenda-setting role has been used to explain the success of Eurosceptic parties in the 2014 European election (Nielsen and Franklin 2017). Second, when it comes to turnout, European Parliament elections have a number of distinctive features, as abstention is not uniform between and within member states, and also varies over time (Blondel et al. 1998). Additionally, the gap between turnout in national and European elections— both between states and over time at the intra-state level— is far from constant (Blondel et al. 1998). Third, national elections take place in different contexts and settings. For example, neither the Great Recession nor the levels and character of popular response to it were constant across European countries. The domestic dynamics of electoral participation might be endogenous to these aspects. However, the criteria used to build the previous restrictive subset of abstainers are rigid, forcing us to exclude one country— namely Switzerland— and leading to a small sample size (n = 1,333, which represents 8.2% of the total sample). Moreover, some authors have emphasised that, in a context of multi-layered institutions, the trade-off between domestic preferences and voting behaviour in European Parliament elections is complex (e.g. Clark and Rohrschneider 2009; Hobolt and Wittrock 2011), and political campaigns leading up to the elections may make European Parliament elections less second order (Beach et al. 2017), weakening the justification for using the restrictive approach to abstentionism.

11 Even though there is some variation across countries in the ratio between response rate and real turnout, it falls within acceptable standards (see LIVEWHAT 2014). If we exclude the most deviant cases, the results reported throughout hold robust.

12 Alternatively, we build a simple summated scale that results from adding the (up to 16) forms of non-electoral political participation— measured through dummy variables— that a given individual might have engaged in. As a robustness check, we have used the simple summated scale as dependent variable, and our results do not change in any substantial way (not reported here).

13 Since we have dummy variables, standard Pearson’s r procedures that assume a normal distribution do not fit our data well. Cronbach’s α is 0.83, above the 0.7 reliability threshold (Eigenvalue = 4.58; 28.65% of the variance explained).

14 The eleven items of the summated scale on the evolution of job conditions come from summating alternative responses (1 = yes; 0 = otherwise) are the following: ‘I took a reduction in pay’, ‘I had to take a job I was overqualified for’, ‘I had to work extra unpaid overtime hours’, ‘I had to work shorter hours’, ‘I had to take or look for an additional job’, ‘my work load increased’, ‘the working environment deteriorated’, ‘I had less security in my job’, ‘I had to accept less convenient working hours’, ‘employees were dismissed in the organisation for which I work’, ‘I was forced to take undeclared payments’.

15 The three items capture the degree of agreement with the following statements: ‘I consider myself well-qualified to participate in politics’, ‘I feel that I have a pretty good understanding of the important political issues’, and ‘I think that I am at least as well-informed about politics and government as most people’, respectively. All of them are measured in 1–5 scales, ranging from ‘strongly disagree’ to ‘strongly agree’. The PCA conducted offers a one-component solution. It reports an Eigenvalue of 2.19, with 73.12% of the variance explained. The scale reliability coefficient falls within acceptable standards (Cronbach's α = 0.81).

16 The internal political efficacy index is not correlated with the indicator of external political efficacy (Pearson’s r = −0.03).

17 The ten included institutions are: the national parliament, politicians, political parties, the EU, trade unions, the judicial system, the police/army, the media, the national government and banks.

18 The scale we built meets the minimum threshold of reliability (Cronbach's α = 0.91; Eigenvalue = 5.60, one-item solution; 55.99% variance explained).

19 These three aspects allow us to assess horizontal power-sharing within institutions, vertical power dispersion and access to challengers through direct democratic mechanisms. Braun and Hutter (2016) rely on three different indicators. First, the Lijphart’s (1999) executives-parties index has information on cabinet duration, the proportionality of electoral systems, the number of effective parties in parliament and the absence of minimal winning and single-party majority cabinets. Second, the percentage of local and regional government as percentage of total taxation measures fiscal decentralisation. Finally, a direct democracy index ranges from zero (no referenda) to four (required referenda plus three types of non-required referenda available).

20 However, the effect of political trust on non-electoral participation is not statistically significant at the 5% level in models 1–2 (Table 5).

21 Although small, note the effect of education holds among abstainers when we use the restrictive criteria to abstentionsim (models 2, 5 and 8 in the Table A2, Appendix).

22 Excluding the turnout predictor and the aggregate-level variables.

23 (AIC(2) = 5220.04; BIC(2) = 5454.96), with 52 estimated parameters, 625 residual degrees of freedom (max. log-likelihood = −2,558.02).

24 (AIC(4) = 16,791.43; BIC(4): 17,173.23), with 67 estimated parameters, 2,138 residual degrees of freedom (max. log-likelihood = −8,328.72).

25 As a robustness check, we have run a multinomial logistic regression with country fixed effects (not reported here), and our results hold.

Additional information

Funding

This work was supported by FP7 Socio-Economic Sciences and Humanities: [Grant Number 613237].