Who Does Not Vote and Why? Implication for New Democracies

Abstract

With the attention of scholars already drawn to the decline in voter turnout in new democracies after the first wave of open/competitive elections, by relying on aggregate data studies have provided explanations for cross-national variations in turnout. Yet, the reliance on aggregate data makes it hard to establish what had lead individuals to abstain from the political process. Thus, in this chapter, by using individual-level analysis from the European Social Survey and the Afrobarometer we re-interrogate the determinants of non-voting in two new democracies of post-Communist Europe and sub-Saharan Africa. Having tested the various explanations for non-voting, first, our results show some consistency across the two regions, suggesting non-voters are those who lack any form of psychological engagement with politics, who are isolated from the recruitment networks and live in urban areas. Second, our result tends to be contradictory, in which while in post-Communist Europe non-voters are men and those with lower level of education in sub-Saharan Africa they are women and those with higher level of education. Third, pertaining to country level indicators, apart from the fact non-voters in both regions are those who have no trust in elections and who lived in countries with disproportional electoral systems, the results tend to be varied.

1. Introduction

Although the first wave of open and competitive elections in a number of new democracies, most especially those in post-Communist Europe and sub-Saharan Africa, was marked by high rates of voter turnout, in recent decades, the attention of scholars has been drawn to the decline of voter turnout in both regions. Despite the difference between the two regions, which includes a history of communism, colonialism, and economic and social development, both regions are similar in that they experienced the transition to democracy almost at the same time (i.e. the early 1990s), but more importantly, they are comparable in terms of their current trajectory with respect to the decline in voter turnout. In fact, an observation of national elections in both regions (i.e. Central/Eastern European countries and sub-Saharan Africa) shows the percentage of people who abstain from voting has gradually risen.

For example, in post-communist European countries, from initial rates of 80%, average turnout rates have reduced to 50–66% in some of these countries [1, 2, 3]. With regards to sub-Saharan Africa, data from the International Institute for Democracy and Electoral Assistance (i.e. International IDEA) suggest Africa’s average turnout was the lowest, at 64% compared to the world average ([4], p. 77). Moreover, even more pronounced are the country variations in turnout, with countries such as Bulgaria, the Czech Republic, Slovakia and Slovenia experiencing a significant downward trend of more than 32% ([5], p. 26). On the other hand, sub-Saharan African countries such as Cape Verde, Nigeria, Mali, Sao Tome and Principe, Zambia, and Senegal have an all-time downward trend in voter turnout ranging from 35.5 to 57%.

Because of the presumed consequences of low turnout for democratic theory and practice¹ and by relying on aggregate data, scholars in established democracies and to some extent those in emerging democracies have done a great deal to provide explanations for cross-national decline as well as regional variations in turnout. However, as cited by Karp and Milazzo [1], the dependence on aggregate data makes it extremely challenging to establish what has led individuals to abstain from the political process, which therefore makes the question of who does not vote and why in new democracies of post-communist Europe and sub-Saharan Africa an interesting and relevant puzzle which requires individual-level analysis. That said, at the core of this chapter is the need: (1) to establish who the non-voters are and what characterises them; (2) to see if the determinants of non-voting are generally similar across both regions or if each region is unique. To do this, we rely on a dataset drawn from the European Social Survey (ESV) and the Afrobarometer (AB). In the section that follows, we begin by explaining why it is important to study non-voting in the context of new democracies, followed by establishing trends in voter turnout (i.e. by comparing the turnout data from the first elections and most recent elections across both regions), then a theoretical review of determinants of voting to deduce explanations for non-voting. Finally, we proceed with a discussion of the research design, present our results, while concluding remarks round off the empirical findings.

1.1 Why studying non-voting matters

Although, we can argue that the issue of non-voting in new democracies does seem to have potentially important implication for democracy and its expansion, however, this does not seem to be clear at first sight. Thus, we find it important to begin by asking why studying the phenomenon of non-voting most especially in the context of emerging democracies in post-Communist Europe and Sub-Saharan Africa countries seem to matter. That said, a brief review of the literature reveals three crucial justification for studying non-voting:

First, as cited by Tambe [6] electoral politics or voting is generally considered or judged as an important corner-stone for representative democracy. Thus, the fact that certain groups or section of the population do not engage or participate poses a genuine problem to representative democracy as the fundamental principle of one-man-one-vote is being violated. To make matters worse, the implication of having a large group or section of people not voting is that they might be a risk of biased representation with groups that turnout having a greater influence on policy outcomes, government composition and issues that get to be debated at the national or political agenda ([7], p. 276). Second, Hadjar and Beck ([8], p. 522); argue non-voting does not only constitute a severe problem of lack of democratic representation but more notably it does reduce the legitimacy of an elected government which goes a long way of decreasing the degree of acceptance of governmental decisions. With the newness of democracies across our regions of interest, this seem to be an important rationale for studying non-voting considering that in such societies it is imperative that people turnout in their numbers as this will directly give consent to the winning candidate and or parties to exercise governmental control without opposition from the losing candidate or parties as they have met the criteria of being a legitimate government chosen by the people. Third, and final reason for studying non-voting is based on the argument that voting is supposed to strengthen citizenship and the quality of democratic civic life [9]. Building on this, Kymlicka [10] cites this justification goes back to classical political thinkers such as Jean J. Rousseau and John S. Mill, who advance the view that political participation tends to enlarge the minds of individuals, thus encouraging them to see and acknowledge that public concerns are the proper ones to which they should pay attention. Moreover, Putnam [11] suggest that voting is important seeing that it encourages social capital, volunteering and other forms of good citizenship. In summary, by building on the representative, legitimacy and citizenship and democratic civic life argument, it is reasonable to expect that the phenomenon of non-voting would have serious implication for democracy and its expansion considering the newness or transition of democracy in post-Communist Europe and sub-Saharan Africa. It is therefore natural as earlier mention to ask who the non-voters are and what characterises them; while equally establishing if the determinants of non-voting are generally similar across both regions or if each region is unique.

1.2 An empirical mapping of voter turnout trends in new democracies: evidence from central and Eastern Europe and sub-Saharan Africa

Voter turnout has been declining in most societies, particularly those in post-communist Europe and sub-Saharan Africa. In this section, we rely on five typologies, which center on:

• Countries that have high voter turnout and turnout remains high.

• Countries that have high turnout and turnout later declines.

• Countries that have low voter turnout and turnout remains low.

• Countries that have low voter turnout and turnout increases.

• Countries that have a stable level of turnout over time.

This is done in order to ascertain if both regions are experiencing a decline in voter turnout. Moreover, by relying on these five typologies, first we focus only on elections where the most important figure of the executive is being elected (i.e. national elections).² Second and most importantly we only considered countries across both regions, where multipartism is the norm and which have held at least four consecutive competitive elections. Tables 1 and 2 show the difference in turnout between the first elections (i.e. third wave of democratisation) and the most recent elections held in each country across the two regions.

Countries	First election		Most recent election		Difference in turnout (%)	Number of elections conducted
	Year	Turnout (%)	Year	Turnout (%)
Albania	1991	98.9	2017	46.7	−52.2	9
Bulgaria	1991	83.8	2017	53.8	−30	9
Czech Republic	1990	96.3	2017	60.8	−35.5	9
Croatia	1990	84.5	2016	52.6	−31.9	9
Estonia	1990	78.2	2015	64.2	−14	8
Poland	1989	62.1	2015	50.9	−11.2	9
Slovenia	1992	85.9	2014	51.7	−34.2	7
Slovakia	1990	96.3	2016	59.8	−36.5	9
Hungary	1990	65.1	2018	69.7	4.6	8
Latvia	1990	81.2	2014	58.5	−22.7	9
Lithuania	1990	71.7	2016	50.6	−21.1	8
Romania	1990	79.7	2016	37.8	−41.9	8
Ukraine	1994	75.8	2014	52.4	−23.4	7

Table 1.

Difference in turnout between the first election and the most recent election: post-communist countries.

Source: https://www.idea.int/data-tools/data/voter-turnout. Notes: Data for Central/Eastern European countries based on parliamentary elections.

Countries	First election		Most recent election		Difference in turnout (%)	Number of elections conducted
	Year	Turnout (%)	Year	Turnout (%)
Benin	1991	64.1	2016	66.1	2	6
Botswana	1989	68.2	2014	84.7	16.5	6
Burkina Faso	1991	35.4	2015	60.0	24.6	5
Burundi	1993	91.4	2015	73.4	−18	3
Cameroon	1992	78.2	2015	64.2	−14	4
Cape Verde	1991	75.3	2016	65.9	−10.1	6
Côte d’Ivoire	1990	70.0	2015	52.8	−17.2	5
Gabon	1993	88.1	2016	59.5	−28.6	5
Ghana	1992	50.2	2016	68.6	18.4	7
Guinea	1993	78.5	2015	68.3	−10.2	5
Kenya	1992	66.8	2017	79.5	12.7	6
Lesotho	1993	72.8	2017	46.4	−26.4	7
Malawi	1994	80.5	2014	70.7	−9.8	4
Mauritius	1991	84.1	2014	74.1	−10	6
Mozambique	1994	88.0	2014	48.6	−39.4	5
Madagascar	1992	74.9	2013	50.7	−24.2	6
Mali	1992	n/a	2013	45.7		5
Namibia	1994	74.2	2014	71.7	−2.5	5
Niger	1993	35.2	2016	59.7	24.5	6
Nigeria	1993	n/a	2015	43.6		6
São Tomé and Príncipe	1991	60.0	2016	46.0	−14	6
Senegal	1993	51.5	2012	57.1	5.6	5
South Africa	1994	86.8	2014	73.4	−13.4	5
Togo	1993	n/a	2015	60.9		6
Tanzania	1995	76.6	2015	67.3	−9.3	5
Uganda	1996	72.6	2016	67.6	−5	5
Zambia	1991	44.4	2016	56.4	12	8
Zimbabwe	1990	53.9	2013	54.3	0.4	5

Table 2.

Difference in turnout between the first election and the most recent election: sub-Saharan African countries.

Source: https://www.idea.int/data-tools/data/voter-turnout and http://africanelections.tripod.com/.

Note: Data for sub-Saharan African countries is based on countries’ parliamentary/presidential elections.

Looking at the data displayed in Tables 1 and 2 and drawing on our empirical mapping of voter turnout based on the five typologies permits us to make the following observations. First, beginning with countries that have high voter turnout after their founding elections and where turnout remains high, the following countries in sub-Saharan Africa (i.e. Burundi, Mauritius, Namibia and South Africa) could be placed in this category. Surprisingly, no country in Central and Eastern Europe could be found in this category. Second, moving to countries that experience high turnout rates after the first competitive elections and later suffer a drop in turnout, most of the post-communist countries could be placed in this category. Third, looking at countries that experience a very low level of turnout rate and their turnout remains low, the result reveals very few countries across both regions could be placed in this category except for Poland, Senegal, Mali and Nigeria. Fourth, turning to countries that experience a very low turnout in their first democratic elections and later experience an increase, like the third typology, very few countries across the two regions could be placed under this category except for Ghana. In summary, what is observed is that not only have new democracies in Central/Eastern Europe and sub-Saharan Africa experienced significant decline in voter turnout, but even more pronounced are the country variations across both regions, with some countries experiencing a significant downward trend in voter turnout, and more so than others.

2. Theory: determinants of non-voting

There is a huge array of factors that have been postulated to explain what influences people’s decision to participate in or refrain from politics (for an overview see: [6, 12, 13]; but see also [14, 15, 16]); to account for non-voting, the theoretical perspective that we adopt is derived from the classical work of Verba, Schlozman and Brady titled Voice and Equality: Civic Voluntarism in American Politics. To explain individual political participation, Verba et al. [17] began by posing the following question: Why do some people not take part in politics? To answer this question, their framework consists of providing three answers to the above question: (1) They cannot; which suggests a paucity or lack of necessary resources needed for political participation. (2) They do not want to; this points to the absence of psychological engagement with politics such as a lack of interest in politics, minimal concern with public issues and or a sense that activity makes no difference. (3) Nobody asked; this implies isolation from the recruitment networks through which citizens are mobilised to engage in politics. By standing on the shoulders of these giants, our explanations for non-voting in post-communist countries and sub-Saharan Africa therefore consist of a theory triangulation that rest on three main factors and or models: individual resources (i.e. capacity), motivation (i.e. political-psychology) and network recruitment.

2.1 Individual resources and non-voting

First, beginning with individual resources, we consider two salient resources that are said to be important for vote choice: socio-economic status (SES) and political experience (i.e. age). Socio-economic status consists of a voter’s educational level and/or income, with studies suggesting those with a higher SES are by and large considered to have a higher propensity to participate [18]. Also, Verba et al. [17] suggest individuals who are more educated tend to participate at a higher rate because they can understand the issues at stake in an election, thus making them more politically interested. In the context of new democracies, studies by Orvista et al. [19] with regards to post-communist Europe do provide evidence supporting that those with higher education and income are most likely to participate. However, when the effect of SES on vote choice is tested in sub-Saharan Africa the results are contrary to the general theoretical expectation, in that findings show those with a lower material status and lower level of education tend to participate at a higher rate compared to those with a higher socio-economic status [20, 21, 22]. Notwithstanding these findings from sub-Saharan Africa, the theoretical assumption adopted here is that the higher an individual finds themselves in terms of SES the more likely they are to engage in electoral politics. Based on this, our theoretical proposition suggests that the probability of non-voting in post-communist Europe and sub-Saharan Africa will decrease with higher socio-economic status. Second, another valuable resource which we consider is political experience, which is better operationalised as age of voters. Carreras and Castañeda-Angarita [23] argue political experience is generally considered to be acquired over time, most especially as voters face concrete policy issues. Following this line of argument, in the context of new democracies, Bratton [20], Kuenzi and Lambright [24], Resnick and Casale [25], Isaksson [22], and Tambe [12, 13] provide empirical evidence confirming those with more political experience are much more likely to engage with or participate in politics. By building on these studies, we therefore expect the probability of non-voting to be higher among those with lower political experience that is younger cohorts.

2.2 Motivation or voter political psychology and non-voting

Another key variable of Verba et al.’s [17] model for explaining why people do not participate in politics centres on motivation or a voter’s psychological disposition, which is measured by political efficacy, political interest, political trust and satisfaction with democracy. First, political efficacy refers to the degree in which voters believe they can understand national politics and the belief or perception that their actions generally have an influence on political institutions. Campbell et al. [26] and Abramson and Aldrich [27] show that lack of political efficacy is a major source of low voter turnout. Moreover, recent studies by Karp and Banducci [28] and Norris [29] reveal that individuals who are considered efficacious tend to be much more involved in politics. In our previous studies [13], although we have been able to confirm a positive significant relationship between political efficacy and electoral participation in post-communist Europe and Western democracies, however, with respect to tropical Africa this relationship was directly contrary to what we expected in that the relationship was non-significant. However, for the sake of comparability we would expect the probability of non-voting will tend to decrease with a higher level of political efficacy. Next, we include political interest, which is defined as the degree to which politics or political affairs arouse curiosity or attention among citizens. As expected, findings across established and emerging democracies do show that individuals who declared to be more politically interested are more likely to engage in politics, most especially in terms of voting [9], with lack of political interest being argued as a cause of lower voter turnout [30]. Thus, we expect the probability of non-voting to decrease with higher political interest. Third, we examine the relationship between political trust and non-voting. Political trust is broadly defined as voters’ or citizens’ evaluation of their political system. According to Putnam [11] trust is the basis of democratic society; this therefore means people will be more willingly to vote if they believe the political system is responding in some way to their voting behaviour. Relying on his work, we expect the probability of non-voting to decrease with increase in political trust. Fourth, we evaluate the relationship between satisfaction with democracy and non-voting, with Norris [29] suggesting that citizens who do not trust their political institutions are least likely to participate. However, in Central and Eastern European countries as in other regions, studies show the level of satisfaction with democracy is generally low [31]. Thus, we will expect the probability of non-voting to be higher among disenchanted voters.

2.3 Network of recruitments and non-voting

Apart from individual resources and motivational factors that we have examined above, Verba et al. [17] argue the only way we could explain why people do not take part in politics is based on the idea that nobody asks, or simply because individuals are outside of network of recruitments. The implication of this is that for us to explain non-voting it is important to look above individuals and include social networks such as family, friends, co-workers, politicians, parties, church, voluntary associations and interest groups, as these social networks can be considered important channels for mobilising individuals because they help nurture political interest and awareness on politics and issues at stake in an election through political discussion. Also, La Due Lake and Huckfeldt [32] argue social networks help provide expertise and free political information (i.e. social capital) which therefore increases the likelihood that citizens will participate in elections. Moreover, Kuenzi and Lambright [24] and Klesner [33], in the context of new democracies do suggest membership of voluntary organisations and or non-political organisations has a significant positive effect on voting. Based on these studies, we would expect the propensity of non-voting to be higher among those who are not members of social networks. Finally, we examine individuals’ place of residence and this is justified from the fact that scholars are still undecided if social networks or parties tend to be more effective in mobilising voters in urban or rural areas. For example, while a study by Karp et al. [34] argues cities are more attractive locations for parties to canvass due to their higher population, Hoffmann-Martinot [35] argues that urbanisation on the other hand tends to reduce interpersonal bonds and social networks thus making it less likely for people to participate. However, considering that scholars are still undecided, we shall only be able to decide on this once our empirical analysis is concluded.

3. Data, method and measurement

To examine comparatively the determinants of non-voting across countries in post-communist Europe and sub-Saharan Africa, one of the initial challenges that we encountered was the lack of comprehensive survey data that included a multitude of countries across both regions. However, to overcome this hurdle, we rely on two separate datasets: European Social Survey³ (ESS-2012; Round 6) and Afrobarometer⁴ (AB-2011-2013; Round 5 [36]) for Central and Eastern European and sub-Saharan Africa countries, which allows us to embark on an international comparison, as the measures of both the dependent and independent variables do not vary very much across the two geo-political regions. Moreover, considering that our outcome variable (i.e. non-voting) is binary, we use multivariate or logistic regression modelling to estimate the probability of non-voting across the two regions. Also, we introduce robust standard errors in our model estimation so as to reduce the variance of fluctuation across our data or samples and proceeds by building a step-by-step model that takes into consideration three theoretical perspectives, while adding a final model that takes into consideration our individual indicators and a few of the country-specific variables that we have added to our data.

Our main dependent variable of interest is non-voting. We employ a dichotomous measure of the respondents who did not vote in the most recent elections. Turning to the independent variables, our key variables are organised into three groups: individual resources, motivation and networking factors. First, concerning individual resources, we include the following: educational level, income and two demographic factors (age and gender). Second, for the motivational variables, we include: political efficacy, interest, trust and satisfaction with democracy. Third, for the networking factors, we include: membership in associational or voluntary organisations, and place of residence (i.e. whether respondents live in an urban or rural area). But more importantly, considering that our two geo-political regions have differing institutional characteristics, but even more so based on current research by Franklin [37], Gallego [38] and Van Egmond et al. [39] which reveals that politics or political action is influenced by the context in which individuals find themselves, in this study we include a number of country-specific contextual variables (i.e. electoral system, concurrent elections, closeness of elections and trust in elections) in predicting non-voting. Appendices 1 and 2 show which countries are included in our study and how the different variables have been operationalised.

4. Results

To explain the probability of non-voting in the new democracies of post-communist Europe and sub-Saharan Africa, we estimate a multivariate logistic regression which combines micro- and macro-level variables and build four models as guided by our theoretical strategy. Moreover, considering our main objective is to evaluate the genuine effects of the variables at different levels and using different models, we began by breaking down citizens’ propensity for non-voting along the lines of each of our key theoretical perspectives (i.e. individual resources, motivation, networks, and contextual or country-level) by building very parsimonious models step-by-step, before rounding up by pulling all the variables from each of the four theoretical perspective into an expanded logit model that incorporates the effect of individual- and country-level variables. An overview of the results from the logistic modelling is summarised in Table 3, which shows which variables was statistically significant in the two regions.

Table 3.

Determinants of non-voting in new democracies: overview of two geo-political regions.

Significance: *** p<.001,** p<.01,* p<.05.

Direction: + = positive relationship, given the coding used. − = negative relationship. 0 = no effect given coding used.

Yellow highlighting: indicates a variable for which it has not been possible to use identical question wording across both regions (so there is a chance that any differences might simply be artefactual rather than substantive).

Note: Analysis produced based on logit regression, with robust standard errors. Detailed tables showing actual coefficient can be obtained be seen in the appendences.

Sources: European Social Survey (Round 6, 2012) and Afrobarometer Survey (2011-2013).

First, we began by modelling the effect of individual resource variables on the probability of non-voting in post-communist Europe and sub-Saharan African countries, with the results revealing the following. Looking at education, in Central and Eastern Europe, people with higher education (i.e. secondary or higher education) turn out to have the lowest probability of being non-voters compared to those with low education, who have the highest rate of non-voting. However, this is directly contrary to our sub-Saharan Africa countries, where those with higher education rather turn to have a higher rate of non-voting compared to individuals with lower education. Next, regarding socio-economic status, across the two regions, we are able to confirm that those with a lower income or socio-economic status are more likely to have a higher rate of non-voting, which comes as a surprise with respect to the African countries. Additionally, looking at age of respondents, across both regions, our results indicate that younger cohorts have a higher rate of non-voting compared to their older citizens. Similarly, turning to the last individual resource variable, that is, gender, in post-communist countries gender points to a significant effect with non-voting, albeit men appearing to have a higher rate of non-voting compared to women. With regards to sub-Saharan Africa, we can confirm women have a higher rate of non-voting compared to men.

Second, pertaining to the motivational or political psychology variables, we can make the following extrapolations. In post-communist European countries, what is observed is that political interest, political trust and political efficacy tend to reduce the probability of non-voting. With respect to sub-Saharan Africa, we can equally confirm that political trust and political interest do in fact reduce the probability of non-voting, but this is not true for political efficacy, which surprisingly shows that those who are declared to be politically efficacious are more likely to be non-voters. Finally, with respect to satisfaction with democracy, across the two regions, our results indicate no substantial effect between being satisfied with a country’s democracy and the probability of non-voting.

Third, turning to the networks of recruitment, we modelled two types of variable (i.e. associational networks and place of residence). In Central and Eastern Europe as well as countries in tropical Africa, what we observe is that those who declared not to be members of associational or voluntary organisations have a higher rate of non-voting compared to those who are members of such organisations. Similarly, looking at one’s place of residence (i.e. urban or rural) what we can deduce is that individuals living in urban areas across both regions are more likely to be non-voters.

Fourth, looking at contextual-level variables, we can make the following remarks. Beginning with electoral system, in both regions, what our data tells is that the rate of non-voting is supposedly higher in countries with a disproportional electoral system (i.e. majoritarian, mixed or plurality). Additionally, turning to closeness of elections, our results tell of a statistically significant relationship with non-voting in Central/Eastern European countries, indicating that the probability of non-voting tends to be higher in countries where voters perceive the elections are closed or competitive. This result is directly opposite to that of our sub-Saharan African countries, where the results points to a non-significant relationship although in the expected direction (i.e. citizens living in countries where the elections are less competitive tend to have a higher rate of non-voting). Moving to trust in elections or electoral integrity, our data reveals that across both regions, the rate of non-voting tends to be higher in countries where voters perceive the elections are not considered to be free and fair. Lastly, we equally evaluate the effect of concurrent elections on non-voting. Although we could test the effect of this variable only in tropical Africa, what can be said is that the probability of non-voting is said to be higher in countries that do not concurrently held elections.

4.1 Who are the non-voters and how much consistency do we find across the two regions?

To explain the determinates of non-voting in the new democracies of Central and Eastern Europe and sub-Saharan Africa, we have relied heavily on cross-sectional survey data that is derived from the European Social Survey and the Afrobarometer. By employing a theory triangulation that consist of individual-level variables, and by adding a few country-specific indicators, we have tested the distinct types of effect on the individual decision to abstain from electoral politics. All done, we are now left with two important question which constitute the core objective of this chapter: Who are the non-voters and How much consistency do we find across the two regions?

Beginning with the first question, (i.e. who the non-voters are?), with respect to the individual resource variable, education and income tend to be a very important indicator in the post-communist countries that we analysed in this study, as those with a lower level of education and material status are more likely to be non-voters. With respect to sub-Saharan African countries this is quite the opposite, as those with a higher level of education are considered most likely to be non-voters. Another important characteristic of non-voters across both regions is age, with data suggesting young cohorts are more likely to be non-voters compared to older citizens. Additionally, moving to the motivational variables, our study reveals non-voting is greatly influenced by political efficacy, political trust and political efficacy. To be precise, in Eastern Europe as in tropical Africa, non-voters tend to display lower interest, trust and efficacy, albeit political efficacy appears to have an indirect effect with respect to sub-Saharan African countries. Similarly, looking at networks of recruitment, across both regions, we can confirm non-voters appear to be citizens who are not members of voluntary or associational organisations and who tend to live in cities or urban areas. Finally, putting the macro-level indicators into perspective, country electoral integrity or trust in elections and electoral formula appears to be influential in our context analysis, with non-voters being those who perceive elections are not free and fair and who live in countries where is electoral system is disproportional. Apart from this, what we could add is that country-level variables tend to have a varied effect across both regions. So far, we have identified who the non-voters are, but even more important is for us to interrogate how much consistency we find across the two regions. This question is important because it will enable us to ascertain if the determinants of non-voting are generally similar across both regions or if each region is unique. By relying on Table 3, we systematically cross-checked each of the four main models across the two geo-political regions, which informs us of the following: First, beginning with the individual resource variables, age proves to be consistent across post-communist European and sub-Saharan African countries, with younger cohorts far more likely to be non-voters. Second, looking at motivational indicators, our data reveals non-voting is positively influenced by factors such as political interest and political trust. Across both regions what we observe is that non-voters have a lower trust in political institutions and a lack of interest in politics or public affairs.

5. Conclusion and implications

So far, our analysis of the determinants of non-voting across new democracies in post-communist Europe and sub-Saharan Africa carries a number of implications. First, we begin by exploring the implication of what the analysis reveals in light of the theory used (i.e. individual resource, mobilisation and networking) and in order to do this, we ask a very simply question: does our theoretical model works equally well across both regions? Answering this question prove challenging because some variables or theories seem to very important in predicting people’s decision not to engage in electoral politics while other do not. That said, we can confirm individual resource, mobilisation and networking models seems to work well across both regions (i.e. most especially in Central and Eastern European countries) with a few exceptions in sub-Saharan Africa countries (i.e. relating to education and political efficacy). Additionally, the country, contextual or institutional variables seem harder to evaluate in general terms because of the inconsistency of the findings. Overall, the full picture in relations to non-voting in new democracies suggest that political behaviour of voters in these regions is determine by many of the same factors that influence political participation in older democracies. Second, this paper also carries certain implications for both future research and democracy: (1) The first relates to how voters could be brought back to the polls or encouraged to participate in electoral politics. In fact, as we have observed in this study, this could be achieved by increasing by raising citizens’ interest in politics and improving their trust in political institutions. For this to be attained, parties must be able to provide voters with clear alternatives, and invest in electoral campaigning, most especially in urban areas. (2) Next, one of the practical implication of this study relates to the conduct of future elections and democracy in these regions. In an era of declining turnout, this study points to the need for electoral integrity and transparency, that is, the need for countries to organise free and fair elections that will not only get voters back to the polls but will equally improve the overall quality of democracy in these regions.

Acknowledgments

The author would like to thank the editor for the valuable comments and suggestions to improve the quality of the paper.

Conflict of interest

No potential conflict of interest was reported by the author.

Appendices

1 Sample of countries in the European Social Survey and Afrobarometer Data.

Regions	Number of countries	Countries covered
Sub-Saharan Africa	27	Benin, Botswana, Burkina Faso, Burundi, Cameroon, Cape Verde, Côte d’Ivoire, Gabon, Ghana, Guinea, Kenya, Lesotho, Malawi, Mauritius, Mozambique Madagascar, Mali, Namibia, Nigeria, São Tomé and Príncipe, Senegal, South Africa, Togo, Tanzania, Uganda Zambia, Zimbabwe.
Post-communist countries	10	Albania, Bulgaria, Czech Republic, Estonia, Hungary, Lithuania, Poland, Slovakia, Slovenia, Ukraine
Total	37

2 Operationalisation of the variables.

Variable	Operationalisation: central and Eastern Europe (European Social Survey)	Operationalisation: sub-Saharan Africa (Afrobarometer)
Dependent variable
Non-voting	Some people do not vote nowadays for one reason or another. Did you vote in the last national election? (1) for non-voting and ‘voting’ (0), thereby omitting all the people who had not been eligible to vote at the last election.	With regard to the most recent national election, which statement is true for you? Voting in the most recent national election? (1) for non-voting and (0) for voting.
Individual resource variables
Age	Age is recoded into four-age categories: young adults (i.e. consisting of those aged 15–33), adults (i.e. those aged 34–49), middle-aged (i.e. those aged 50–59) and finally elderly people (i.e. those aged 66 and above).	Age is recoded into four-age categories: young adults (i.e. consisting of those aged 15–33), adults (i.e. those aged 34–49), middle-aged (i.e. those aged 50–59) and finally elderly people (i.e. those aged 66 and above).
Gender	Dummy: 1 for a woman, 0 for a man.	Dummy: 1 for a woman, 0 for a man.
Education	People were asked for the highest level of education they had achieved. The different educational systems & degree allows us to create three educational categories: Primary education, secondary education and higher education	People are asked for their highest educational level of education. Regarding education, respondent’s educational levels are recoded into four categories (i.e. no formal education, primary education, secondary education and higher education).
Income	Into which of the following income ranges does the total monthly income of this household fit: [10 deciles based on the currency and distribution of the country] (lowest income = 1…highest income = 10)	We use occupational status as a proxy to measure income. Occupational status is measure by a question which asks about the citizen’s occupational status. A dummy variable is therefore created, with 1 assigned to individuals having jobs, while a coding of 0 is assigned to individuals who declared having no job.
Motivation or political psychology variables
Political interest	Political interest is capture by a question that asks how interested would you say you are in politics? We recode this into a four-point with not interested coded as 0, not very interested 1, somewhat interested assigned a value of 2 and very interested coded as 3.	Political interest is capture by a question that asks how interested would you say you are in public affairs? We recode this into a four-point with not interested coded as 0, not very interested 1, somewhat interested assigned a value of 2 and very interested coded as 3.
Political trust	Trust is measured by an evaluation regarding trust in the parliament, the legal system and politicians (i.e. please tell me on a scale of 0–10 how much you personally trust each of the following institutions.	How much do you trust each of the following, or have not you heard enough about them to say: The President/Prime Minister? Trust parliament/national assembly, Trust courts of law. Recoded into 0 = no trust, 1 = little, 3 = some, 4 = a lot)
Satisfaction with democracy	How satisfied with the way democracy works in country on a scale of 0–10 (i.e. 0 extremely dissatisfied, 10 extremely satisfied)	Overall, how satisfied are you with the way democracy works in the country? 0 Not satisfied, 1 not very satisfied, 2 fairly satisfied, 3 very satisfied.
Network of recruitments variables
Associational/informal networks	Membership in voluntary or informal networks is measure by evaluating membership in different social and political organisation such as religious, recreational, environment, labour, professional and humanitarian organisations.	Membership in voluntary or informal networks is measure by evaluating membership in different social and political organisation such as religious, recreational, environment, labour, professional and humanitarian organisations.
Residence	Recoded into 1 = urban area, 0 = rural area	Recoded into 1 = urban area, 0 = rural area
Country level indicators
Electoral system	Dummy: 0 for Proportional system, 1 for plurality, mixed or majoritarian systems	Dummy: 0 for Proportional system, 1 for plurality, mixed or majoritarian systems
Concurrent elections	Dummy: 0 for elections not concurrent, 1 for concurrent elections	Dummy: 0 for elections not concurrent, 1 for concurrent elections
Closeness of elections	I measure closeness of election as the margin of victory for the winning candidate or over the runner-up in presidential elections. While for parliamentary democracies, we measure closeness of election as the difference in seat shares between the top two parties. That said, I coded the variable in such a way that a winning margin of less; 5% = 1, and a margin greater than 5% = 0.	I measure closeness of election as the margin of victory for the winning candidate or over the runner-up in presidential elections. While for parliamentary democracies, we measure closeness of election as the difference in seat shares between the top two parties. That said, I coded the variable in such a way that a winning margin of less 5% = 1, and a margin greater than 5% = 0.

3. Logistic analysis of non-voting: results for Central and Eastern European countries.

Models/variables	Model I				Model II				Model III				Model IV
	B	S.E	Sig	ExpB	B	S.E	Sig	ExpB	B	S.E	Sig	ExpB	B	S.E	Sig	ExpB
Constant	.069	.072	.34	1.07	1.369	.124	.000	3.932	1.443	.126	.000	4.232	1.715	.143	.000	5.557
Individual resource
Gender (Male)	.024	.041	.56	1.02	−.138	.043	.001	.871	−.132	.044	.002	.876	−.141	.044	.001	.869
Age (reference: young people)
Adults (34–49)	−.55	.056	.000	.574	−.478	.058	.000	.620	−.470	.059	.000	.625	−.470	.059	.000	.625
Middle-aged adults (50–59)	−.80	.064	.000	.449	−.608	.067	.000	.545	−.594	.067	.000	.552	−.606	.067	.000	.546
Elderly people (60+)	−1.05	.058	.000	.348	−.754	.061	.000	.470	−.751	.009	.000	.472	−.770	.062	.000	.989
Educational level (reference: primary education)
Secondary education	−.358	.052	.000	.699	−.247	.054	.000	.781	−.287	.005	.000	.750	−.294	.055	.000	.745
Higher education	−.882	.070	.000	.414	−.569	.074	.000	.566	−.639	.075	.000	.528	−.631	.076	.000	.532
Income	−.013	.008	.09	.987	−.005	.009	.59	.995	−.011	.009	.29	.989	−.011	.009	.27	.989
Motivational variables
Political interest					−.710	.028	.000	.492	−.709	.028	.000	.492	−.680	.028	.000	.506
Political trust					−.067	.010	.000	.936	−.069	.010	.000	.934	−.072	.010	.000	.930
Satisfaction with democracy					.002	.010	.84	1.02	.003	.122	.85	1.03	.002	.010	.85	1.002
Political efficacy					−.065	.010	.000	.937	−.064	.010	.000	.938	−.024	.012	.05	.976
Network variables
Associational network									−.150	.045	.001	.861	−.110	.046	.02	.895
Residence (Urban)									.230	.047	.000	1.25	239	.047	.000	1.27
Country-level variables
Electoral system													.125	.046	.000	1.133
Closeness of election													.243	.044	.007	1.133
Trust in elections													−.091	.013	000	.913
Chi-Square Improvement (df)	467.9 (7)				1341.0 (11)				1376.5 (13)				1462.5 (16)

Significance levels: ***p < .001, **p < .01, *p < .05.

4. Logistic analysis of non-voting: result for sub-Saharan African countries.

Models/variables	Model I				Model II				Model III				Model IV
	B	SE	Sig	ExpB	B	SE	Sig	ExpB	B	SE	Sig	ExpB	B	SE	Sig	ExpB
Constant	−1.356	.038	.000	.258	−.761	.054	.000	.467	−.762	.056	.000	.467	−.447	.068	.000	.647
Individual resource
Gender (male)	.140	.031	.000	1.150	.098	.031	.002	1.103	.082	.037	.000	1.086	.085	.031	.007	1.089
Age (reference: young people)
Adults (34–49)	−.811	.036	.000	.445	−.801	.036	.000	.449	−.780	.037	.000	.458	−.807	.037	.000	.446
Middle-aged adults (50–59)	−.947	.059	.000	.388	−.920	.059	.000	.398	−.904	.059	.000	.405	−.944	.060	.000	.389
Elderly people (60+)	−.961	.062	.000	.383	−.922	.063	.000	.394	−.924	.063	.000	.397	−.971	.063	.000	.379
Educational level (reference: no education)
Primary education	.197	.037	.000	1.217	.178	.038	.000	1.195	.137	.038	.000	1.146	.071	.039	.06	1.073
Secondary education	.326	.041	.000	1.385	.278	.042	.000	1.321	.196	.044	.000	1.216	.067	.045	.14	1.069
Higher education	.514	.080	.000	1.672	.457	.081	.000	1.579	.351	.083	.000	1.421	.209	.084	.013	1.233
Income	−202	.033	.000	.817	−.188	.033	.000	.828	−.198	.033	.000	.820	−.205	.034	.000	.815
Motivational variables
Political interest					−.164	.014	.000	.828	−.145	.014	.000	.865	−.135	.015	.000	.873
Political trust					−.127	.015	.000	.881	−.121	.015	.000	.886	−.057	.015	.000	.917
Satisfaction with democracy					−.073	.017	.000	.930	−.074	.017	.000	.929	.015	.018	.40	1.015
Political efficacy					.042	.012	.000	1.043	.043	.012	.000	1.044	.032	.012	.007	1.032
Network variables
Associational network									−.159	.046	.001	.853	−.192	.047	.000	.825
Residence (Urban)									.219	.033	.000	1.245	.236	.033	.000	1.266
Country-level variables
Electoral system													.264	.038	.000	1.302
Closeness of elections													−.013	.035	.70	.987
Trust in elections													−.244	.015	.000	.784
Concurrent elections													−.337	.068	.000	.640
Chi-square improvement (df)	1090.4 (8)				1391.0 (4)				1506.2 (4)				1957.4 (6)

Notes

1. By relying on the representative, legitimacy and citizenship and quality of democracy, we provide a justification for why studying the phenomenon of non-voting within the context of new emerging democracies seem to matter.

2. In the case of post-communist Europe these are generally parliamentary elections, while for sub-Saharan Africa it is a mixture of presidential and parliamentary elections.

3. The European Social Survey (ESS) is an academically driven cross-national survey that has been conducted across Europe since its establishment in 2001. Every 2 years, face-to-face interviews are conducted with newly selected, cross-sectional samples. The survey measures the attitudes, beliefs and behaviour patterns of diverse populations in more than 30 nations http://www.europeansocialsurvey.org/data/download.html?r=6 [Accessed: 18/06/2018].

4. Afrobarometer is a pan-African, non-partisan research network that conducts public attitude surveys on democracy, governance, economic conditions, and related issues in more than 35 countries in Africa: http://afrobarometer.org/data/merged-round-5-data-34-countries-2011-2013-last-update-july-2015 [Accessed: 18/06/2018].