Introduction

The enduring coronavirus disease that emerged in 2019, the Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) (COVID-19), has harshly wedged both health and the economic systems. There have been worries that voter turnout during elections could increase the risk of spreading the virus, and casualties are likely to upsurge as more people participate in voting. This insinuation stems from the fact that the disease quickly spreads in crowded places as people will be in close contact permitting transmission. Thus, voter turnout is expected to decline due to the pandemic. The paper investigates the relationships between COVID-19 and voter turnout in Nigeria. Examining the determinants of voter turnout in Nigeria is pertinent because even though Nigeria is Africa's largest democracy (Nwankwo 2018), it has witnessed a consistent decline in voter turnout since 2003 (Nwankwo 2019). Since the "rule by the people" is the fundamental principle of democracy, it thus makes sense that higher levels of voter turnout confer legitimacy on the election and hence a measure of a democracy's performance (Nwankwo and Okafor 2017; Kuenzi and Lambright 2007).

Current studies have focused on determining whether turnout during the pandemic increased the risk of more cases of or deaths from COVID-19 and others focused on the potential impact of the epidemic on voter turnout and absentee voting. The existing studies were conducted mainly in the US and Europe. There is a need to explore the nexus of COVID-19 and electoral results in Africa and other regions to compare the results with that already being produced in the US and Europe. This present study focuses on the perception of measures to curtail the pandemic and their influences on people's likelihood to vote. This context of the COVID-19-voter turnout nexus has not been explored.

The aim is to examine the relationship between voter turnout and COVID-19 in Nigeria. It does this by examining public opinions on the safety measures for curtailing COVID-19, such as the use of facemask, social distancing, quarantine, home confinement, whether they think the virus is real. This research is, to the best of my knowledge is the first to examine the perception of COVID-19 effect on the likelihood of voter turnout in Africa. The study was done in Lagos State, the epicentre of the pandemic in Nigeria. The study employed a survey design and a quantitative approach. The data were collected in Lagos State that has the most cases and death from COVID-19 in Nigeria. The questionnaire survey was administered to individuals of voting age in randomly selected households in the three most impacted municipalities in Lagos State. Lagos State is located on the South-Western part of Nigeria, approximately longitude 20° 42′ E and 32° 2′ E, and latitude 60° 22′ N and 60° 2′ N (Lagos State Government LASG 2021). Lagos State is the commercial hub of Nigeria and the West African sub-region, even though it is the smallest state in Nigeria by landmass. The state has the highest urban population in Nigeria, which is 27.4% of the national estimate. In the 2006 National Census, Lagos State's population was estimated as 9,013,534 compared with the National total of 140,003,542. However, based on the UN-Habitat and international development agencies' estimates, Lagos State is said to have about 24.6 million inhabitants in 2015 (LASG 2021). Lagos State is a metropolitan state which reflects the diversity of the country.

The study's analysis began with descriptive statistics (such as frequency, standard deviation) of the variables. Then, multiple regression was used to determine the effect of classic political participation factors while controlling for opinions on measures to curtail COVID-19. The paper shows that there is a widespread propensity to believe that COVID-19 is not genuine in Nigeria and that most of the protection procedures in combating the disease are not vital. It also shows that people who think that the use of facemask and quarantining is essential for getting infected or spreading the virus and that COVID-19 is not real are significantly more likely to vote during the pandemic. Besides, educated people and those who trust the government are more likely to vote if elections were conducted during the pandemic. While these findings may have some implications, the primary conclusion from this study is that it indicates that political participation during the pandemic can result from how linked the individual citizen is to the broader social and political community.

This conclusion is made because the primary finding is that most people do not think that the virus is real, and hence quarantining or related measures, e.g., self-isolation, social distancing, home confinement, are not required. This orientation is significantly influenced by political distrust among many people. The result reveals that people who are likely to vote do not hold these views and trust the government. It thus follows that people who have the notion the virus is real and practice isolation and social distancing are likely to have limited social interaction with their friends and neighbors. The interaction between people can shape their attitude as friends and neighbors may shape each other's views on the pandemic; hence it can also influence their propensity to vote. It does make sense that a reduction in social interaction caused by COVID-19 can shape the participatory levels of individuals. Hence, as the finding suggests, people who believe the virus is real and adhere to safety guidelines for COVID-19 by quarantining and will likely interact less with their friends and neighbors, which may engender political and social alienation affecting their participatory disposition, are less likely to turnout. However, since most people think that the virus is not real and do not have trust in the government and hence may not vote, voter turnout in Nigerian elections are likely to be low because people who have trust in government and have higher odds of voting prefer not to turnout rather than contract the virus.

The result thus contributes to the emerging argument that people's negligence of government's directives on the pandemic, such as social distancing and self-isolation, is because of political distrust in Nigeria (Ezeibe et al. 2020). Thus, we may argue that political trust has a boosting effect on accepting government policies and actions in the fight against COVID-19. Thus, theoretically, the analysis supports the political efficacy model of political participation and the social connectedness model indicating their vitals for exploring the connection between COVID-19 and electoral or political participation more broadly. The argument is that these two models are vital for exploring the nexus of COVID-19 and political participation in Nigeria and perhaps other African countries having similar political settings like Nigeria.

Furthermore, the paper contributes to some elements of the literature. First, it aligns closely with the very new and sparse literature examining the effects of epidemics such as Ebola (Beall et al. 2016; Campante et al. 2020), HIV/AIDS (Mansour et al. 2020) and more recently, COVID-19 (e.g., Adam-Troian et al. 2020; Bisbee and Honig 2020; Fernández-Navia et al. 2021 among others) on electoral behaviour). Regarding COVID-19, Fernández-Navia et al. (2021) show that the turnout was between 2.6 and 5.1% lower in boroughs with positive cases of COVID-19, while Adam-Troian et al. (2020) found that areas where the perceived threat of COVID-19 was higher, showed more support for conservative parties in the first-round of 2020 French municipal polls. Remarkably, they could not establish that the real COVID-19 risk had any impact on vote shares.

More generally, this paper also contributes to the recent studies on the political effects of the lockdown caused by COVID-19. Most of the studies suggest that lockdowns boosted the intent to vote for the ruling executive, trust in the government, and satisfaction with democracy in several European and Western countries (Bol et al. 2020; De Vries et al. 2020; Giommoni and Loumeau 2020). Relatedly, lockdown heightened degrees of trust in science, politicians, and the police in New Zealand (Sibley et al. 2020) and expanded the backing of strong leadership and technocratic governments in Spain (Amat et al. 2020). As this paper shows, these findings seem contrary to the Nigerian case where COVID-19 lockdown and associated measures decreased trust in government and more likely to reduce voter turnout.

Review of the literature

The political effect of epidemics has been researched, but this body of literature is still developing and scarce (Fernández-Navia et al. 2021). Existing studies have focused on Ebola, HIV/AID and now COVID-19. Beall et al. (2016) show that voter intentions backing the Republican Party rose in locales with more strong worries about Ebola. Similarly, Campante et al. (2020) find that Ebola fears triggered a decline in the Democratic vote portion in the 2014 midterm polls and reduce voter turnout. Mansour et al. (2020) analyzed the impact of exposure to HIV/AIDS on vote shares for the US House of Representatives. They found a positive relationship between HIV/AIDS death and the vote share received by Democratic candidates.

The research exploring the relationships between COVID-19 and electoral outcomes is still very nascent and absent in Africa and many other developing countries and regions. Most of the few studies that have been done were conducted in the US (Cotti et al. 2020; Flanders et al. 2020; Morris and Miller 2020), except a very few studies done in France (Bertoli et al. 2020; Cassan and Sangnier 2020). The thrust of the existing studies has been on determining whether turnout during the pandemic increased the risk of more cases (Cotti et al. 2020; Flanders et al. 2020) or deaths from COVID-19 (Bertoli et al. 2020; Cassan and Sangnier 2020). Others focused on the potential impact of the pandemic on voter turnout (e.g., Morris and Miller 2020) and absentee voting (Yoder et al. 2020).

Adam-Troian et al. (2020) examined the first-round of 2020 French local polls and show places where the supposed threat of COVID-19 was more significant to have more backing for conservative parties. Still, the real COVID-19 threat had no impact on vote shares. Bisbee and Honig (2020) indicate exposure to COVID-19 reduced backing for Bernie Sanders in the US Democratic Party presidential primaries. Morris and Miller (2020) showed that COVID-19 reduced turnout in the US primary elections. Cotti et al. (2020) investigated the extent to which in-person voting increased the number of COVID-19 cases after the Wisconsin primary elections and found a statistically and economically significant association between in-person voting and the spread of COVID-19 two to three weeks after the election. Flanders, Flanders, and Goodman (2020) examined the relationship between US primary elections' voter turnout with county-level COVID-19 incidence early in the pandemic and found that COVID-19 risk was related to voter turnout most strongly in Michigan during the week starting three days postelection. For more extended periods, the relationship was increasingly weaker. They suggest that even though there was increased absentee-ballot voting in the primary election, there was a relationship between voter turnout in at least one state with an evident rise in risks related with and possibly because of more extensive exposures linked to the primary election.

Yoder et al. (2020) explored whether absentee voting will impact voter turnout and election results during the COVID-19 pandemic in the US and indicate that notwithstanding worries that COVID-19 would reduce turnout in the absence of absentee voting, the turnout gap between 64 and 65-year-olds did not surge during COVID-19, even as the absentee voting trebled relative to earlier runoffs signifying that increasing absentee voting in the pandemic will cause many voters to shift to a more health-preserving means of voting, without essentially altering election results even despite potential partisan variances in favor for absentee voting.

In France, a study by Cassan and Sangnier (2020) investigated the extent to which the 15 March 2020 municipal elections contributed to the COVID-19 epidemics in France and showed that there was a damaging consequence of the voting in locations that were at comparatively advanced stages of the epidemics at the time of the poll. However, voting did not contribute to the epidemics in places with lower infection levels by 15 March. Overall, they indicated that the elections contributed 15% of all hospitalizations but holding future polls may not be damaging. Bertoli et al. (2020) studied the same election. They looked at the relationship between voter turnout and casualties from the pandemic showed that a higher turnout was related to substantially higher death for the aged population in the five weeks after the polls.

COVID-19 spread engendered a fear that in-person voting will further diffuse the virus, which creates a further burden to voters and hence voter turnout is likely to decrease. Vázquez Carrero et al. (2020) confirmed this hypothesis by exploring the effects of COVID-19 on voter turnout in the Basque Country regional elections scheduled immediately after the zenith of the first wave of the pandemic. They found that COVID-19 led to a decline in turnout by roughly 4.7% in boroughs impacted by the virus compared with those not affected by it at that time. This result is also similar for death from the virus, with municipalities witnessing more deaths from coronavirus having lower turnout than those having only infected cases. Fernández-Navia et al. (2021) examined the causal effect of local exposure to COVID-19 on voting actions and electoral results in Spain using data from regional elections conducted on 12 July 2020. They indicate that turnout was between 2.6 and 5.1% points lower in boroughs that suffered positive cases of COVID-19.

Using data from the 2020 Croatian parliamentary election, a study by Sircar (2020) applied the difference-in-differences approach to examine the relationship between COVID-19 and electoral outcomes focusing on the vote shares for the dominant Croatian Democratic Union party, as well as the turnout. The study argues that county-level COVID-19 infection levels did not affect Croatian Democratic Union support or turnout. Nevertheless, the statistical power may have affected the results using the difference-in-differences approach, matters connected to causal identification and dependability of infection rate measures.

The paper by Giommoni and Loumeau (2020) investigated the electoral impact of the COVID-19 lockdown in France. Focusing on variance in constraint measures employed across French divisions, they explored voting behaviour in the pre-lockdown preliminary municipal elections and the second round after lockdown was executed. They showed that lockdown rules significantly impacted electoral results. First, in areas with a stiffer restriction, the incumbent's vote cut was higher, plus the consensus for Green parties. Second, voter turnout was larger in localities under more severe lockdown. Using an extensive survey conducted during the pandemic's first wave, Daniele et al. (2020) explored the impact of COVID-19 on socio-political attitudes in Italy, Spain, Germany, and the Netherlands. Their study indicates a significant decline in interpersonal and institutional trust, plus backing for the EU and a tax-financed welfare state. They further show that the results indicate "a rallying effect around scientific expertise and incumbent governments that—together with populist positions losing ground"—suggest rising demand for competence.

Some studies explore the effect of lockdown on electoral behavior. Bol et al. (2020) show that lockdowns boosted the voting intentions for the serving administration, trust in the government, and contentment with democracy in Western European countries. Amat et al. (2020) show the epidemic boosted backing for strong leadership and technocratic governments in Spain. De Vries et al. (2020) show that the backing of the ruling executive in many European countries rose after the introduction of lockdown in Italy. Giommoni and Loumeau (2020) studied the French municipal polls. They found that in areas with harsher lockdowns, the backing for the in-power executive and Green parties was more remarkable, and also, voter turnout was higher. In New Zealand, Sibley et al. (2020) demonstrate that individuals subjected to more lockdown rules had risen levels of trust in government officials, the police and science.

From the literature, the existing studies were conducted mainly in the US and Europe. There is no research on the connection between voter turnout and COVID-19 in Africa. The thrust of the existing studies has been on determining whether turnout during the pandemic increased the risk of more cases of or deaths from COVID-19 or whether the pandemic will allow for a genuine and transparent election because of the potential changes in the election cycle. Others focused on the potential impact of the pandemic on voter turnout and absentee voting and the likelihood for changes in election result in swing states in the US. There is a need to explore the nexus of COVID-19 and electoral results in Africa and other regions to compare the results with that already being produced in the US and Europe. Besides, the results of the existing studies tend to be conflictive. There is no general agreement on whether COVID-19 increases or reduces turnout. This present study joins this debate by focusing on the perception of measures to curtailing the pandemic and their influences on people's likelihood to vote. This context of the COVID-19-voter turnout nexus has not been explored. Thus, this study examines the relationship between voter turnout and COVID-19 in Lagos, the pandemic's epicenter in Nigeria.

Theoretical insights on voter turnout and the COVID-19 pandemic

What constitutes political participation is varied, but voter turnout is the primary way citizens participate in politics. Several theories abound, but the rational choice model, political efficacy model, and socio-economic model are the primary lens through which political participation is explored (Nwankwo and Okafor 2017). In a rational choice context, people select outcomes with better efficiency over results with reduced benefit and choose to engage in actions that produce more greatly priced endings (Aldrich 1993). Thus, rational voters gauge the prices and gains of voting, and if the prices surpass the gains, you can anticipate abstention or otherwise if the reverse is the case (Downs 1957; Aldrich 1993; Franklin 2004; Nwankwo and Okafor 2017). Thus, we will anticipate that people who consider contracting or spreading COVID-19 have far lesser consequences than voting will likely turnout more during the pandemic. Hence, the pandemic can reduce voter turnout significantly, especially in a country where mail-in or virtual voting is not allowed, such as in Nigeria. However, rationality does not often predict turnout because people still turnout despite knowing the cost implications is far greater than benefits (Jung 2017). This happens because of civic voluntarism, the idea that people vote in elections because they see voting as their civic duty, have a great interest in politics and public affair, and essentially partisanship–people's psychological ties to a political party– one of the firmest predictors of voter turnout (Dalton 2008; Lewis-Beck et al. 2008; Singh 2011).

Political participation has been seen to be influenced by political trust, which underlines the political efficacy model. In this sense, persons' alignment, such as trust in government, civic duty, and political interest, are vital factors for voter turnout. The steadiest conclusions from these factors are the connection between political trust and electoral politics participation (Hooghe and Marien 2013). Studies exploring factors of voter turnout in Africa and indeed in Nigeria have noted that political trust is an essential factor determining political participation (Kuenzi and Lambright 2007, 2011; Nwankwo and Okafor 2017; Nwankwo 2019). Political trust also boosts the acceptance of government policies and actions in the fight against COVID-19 in Nigeria (Ezeibe et al. 2020). Thus, we can expect political trust to be significantly related to voter turnout during the COVID-19 pandemic, although it depends on what the government is advising on voting procedures. In Nigeria, elections have been conducted in Edo and Osun States after the national lockdown was lifted, and the elections witnessed significant turnout. Ezeibe et al. (2020) have argued that people's negligence of government's directives on the pandemic, such as social distancing and self-isolation, is because of political distrust.

The socio-economic model indicates that individuals with a high socio-economic status (SES) such as education, occupation, and income are more likely to embrace alignments that encourage them to vote. This claim has been supported by prominent studies on political participation in Western democracies (Verba and Nie 1972; Wolfinger and Rosenstone 1980; Leighley and Nagler 1992) and continues to be found in newer studies (Blais 2000; Carreras and İrepoğlu 2013). Some findings in Africa have endorsed this idea of the SES paradigm of voting (e.g., Amoateng, Kalule-Sabiti and Heaton 2014; Larreguy and Marshall 2017) but refuted by others (e.g., Kuenzi and Lambright 2007; 2011; Isaksson 2014; Nwankwo and Okafor 2017).

It is not yet clear whether people with higher SES will vote more during a pandemic or not. A study in China indicates that most Chinese residents of a reasonably high SES, particularly women, are well-informed about COVID-19, have optimistic mindsets, and have proper practices towards COVID-19 (Zhong et al. 2020). This finding suggests that people with higher SES will likely not turn out to vote in a COVID-19 pandemic because they are more knowledgeable and are confident about COVID-19 warnings and thus tend to be more willing to adhere to appropriate practices towards COVID-19. This is contrary to the established notion outside Africa, which indicates that individuals with higher SES tend to vote more because they are more knowledgeable and have better information.

Since other SES factors are less comprehensible than educational indicators (Gallego 2010), education is the main SES element considered in this study. Some socio-economic literature highlight distress (e.g., socio-economic pain) as a political involvement factor (Thomassen 1989). Despite the contention that various kinds of stress shape participation (e.g., Klandermans, van der Toorn and van Stekelenburg 2008), there is a joint emphasis on relative deprivation or objective material conditions. Thus, dissatisfaction with life presents a basic logic for individuals' political involvement (Gurr 1970; Buechler 2004; Giugni and Grasso 2015).

Public health proposals and governmental actions during the COVID-19 pandemic imposed several constraints on daily living, comprising social distancing, quarantine, self-isolation, and home confinement, which put a psychosocial strain on people because of the decreased amount of social activity through interactions with family, friends, neighbors, or entertainment. These adverse effects on social participation were associated with lower life satisfaction during the confinement period (Ammar et al. 2020). Thus, we would anticipate life satisfaction during the pandemic to be positively associated with voting and dissatisfaction with life to reduce turnout. The research takes on these claims and incorporates variables evaluating levels of life satisfaction and economic well-being (income).

Demographic variables: gender, marital status, and age, which are predictors of turnout, are also included in the study. It has been argued that older people in Africa have higher voting rates than the younger population (Resnick and Casale 2011). This result is in line with studies in western democracies, but the difference is that younger Africans are less likely to protest than older folks. Age plays a vital role in people's likelihood to vote in an election because older people do not get confused when faced with various party policies and packages (Carreras and İrepoğlu 2013). They have better political knowledge, enabling them to scrutinize many programs political parties are offering (Carreras and İrepoğlu 2013). Thus, it is expected that older eligible voters tend to vote more than younger voters.

Contrary to this postulation, we do know if COVID-19 will affect older people than younger people. Health experts and authorities warn that older people are at a higher risk of more severe and likely deadly illness from COVID‐19. Brooke and Jackson (2020, p. 2044) indicated that death data from Oxford COVID‐19 Evidence Service (25/3/20) show a death risk of 3.6% for people in their 60 s, which surges to 8.0% and 14.8% for people in their 70 s and over 80 s, respectively. Thus, we would anticipate that due to COVID-19, the age gap in voting would be reduced if not reversed.

It has been long argued that differences exist between the sexes in voting: males having higher turnout levels (Wolfinger and Rosenstone 1980). Social norms, structural factors, and women's late suffrage have been linked to the lower levels of voting among women (Franklin 2004; Mayer 2010). Amoateng et al. (2014) show that overall, females' voting rates are two over three times less than male rates but vary widely across African countries over time. There is no consensus on the gender gap in Nigeria: some studies find men to vote more than women (e.g., Taiwo and Ahmed 2015) others find no significant difference (Nwankwo and Okafor 2017). The traditional idea that being male has a higher effect on voting than being female is followed in this study.

Marital status has been shown to influence the decision to vote; married people tend to vote more than unmarried people (Wolfinger and Rosenstone 1980; Verba et al. 1995). However, it has not been straightforward as this suggests as it has been argued that marriage makes no difference (Miller et al. 1996; Highton and Wolfinger 2001). It is argued that the influence of marriage drops when a spouse dies (Wilensky 2002), when there is divorce or separation (Wallerstein and Kelly 2008) because terminating a marriage brings trauma that upsets usual behaviors, e.g., voting. Thus, it is safe to expect that turnout between never-married people and formerly married people will not be substantially different and anticipate that married people will tend to vote more than not married people.

The study examines how these variables discussed above will influence voter turnout under the COVID-19 scenario. Thus, this study used COVID-19 safety measures as controls to see the effect of the primary models of participation on voter turnout. The variables considered are social distancing, COVID-19 is real, quarantine/isolation, home confinement, and facemask. Most of these variables seem to relate to the social connectedness model of political participation. Social connectedness considers the nature of an individual's connection to the larger society. Initial ideas on this model identify alienation, prejudice, estrangement, anomie, and apathy as factors that impede political participation (Leighley and Vedlitz 1999; Nwankwo et al. 2017). Present exponents of this model argue that political involvement is a product of how connected the individual citizen is to the broader social and political community. Social connectedness is also related to friends and neighbors; the notion that people who live together in the same neighborhood and interact with each other tend to have similar voting behaviors.

It also follows that the interaction between people can shape their propensity to vote as friends and neighbors may convince each other to vote or not to vote. It does make sense that a reduction in social interaction caused by COVID-19 can shape the participatory levels of individuals. Several COVID-19 guidelines such as social distancing, quarantine/isolation, home confinement can limit people's connectedness to their friends and neighbors and the broader political and social community, which may engender political and social alienation affecting their participatory disposition. The study explores individuals' likelihood to vote in an election during the COVID-19 pandemic while accounting for these variables. The associated hypothesis tested for these variables is indicated in Table 1. Generally, these variables describe the divided opinions between COVID-19 enthusiasts and skeptics, which has also played out in the realm of electoral politics, for example, in the just-concluded US election where COVID-19 enthusiasts prefer mail-in voting while skeptics opposed it. While physical voting or mail voting is not a hot topic in Nigeria compared with the US, the enthusiasts-skeptics divide features prominently in the political discourse of the pandemic in Nigeria. Thus, this study will offer insights into the connection between political behaviors and popular perceptions of curtailing the COVID-19 pandemic.

Table 1 Variables considered for the study
Full size table

Methodology

This study was conducted in Lagos State, the epicenter of COVID-19 in Nigeria. Three local government areas (LGA) were purposively selected as they are the hub of COVID-19 in Lagos State, namely Eti-Osa, Lagos Mainland, and Ikeja LGAs. Lagos is the commercial hub of West Africa and the largest city in Africa by population. The study engages a cross-sectional survey in the 31 wards in the three LGAs (Eti-Osa has 10 Wards, Lagos Mainland 11 wards, Ikeja has ten wards). Data for the study were collected using copies of a questionnaire during a field survey in all the wards. Trained research assistants who are resident in each LGA administered the questionnaire from 20 July 2020 to 10 November 2020. A paper-based survey was possible because lockdown and other COVID-19 restrictions have since been suspended in the State in May 2020.

Since the LGAs have an almost equal number of wards, we administered the same number of the questionnaire. A total of 450 copies of the questionnaire (150 in each LGA) was administered to residents aged 18 and above (i.e., those eligible to vote) in all the wards. However, 397 questionnaires were correctly filled and returned, representing 88.22%. One street was selected from each ward in the three LGAs studied. In a ward, one street was selected randomly from all the streets in that ward. In the street selected, the households are counted, and a maximum of 15 households are selected randomly. If there are 30 households in the street, we make a list of 1–30 and select 15 households at random from the list. A household consists of one or more people who live in the same dwelling and may consist of a single-family or some other grouping of people (O'Sullivan and Sheffrin 2003). Appendix 1 shows the questionnaire, and Appendix 2 details the sampling scheme. The conditions for the choice of participants were eligible to vote, availability at the time of the survey, and disposition to partake in the study in line with earlier studies (Nwankwo 2019).

It is vital to note that the study is focused on individuals and not households directly. The study only sampled households to avoid administering the questionnaire to individual multiple times. One person of voting age and willing to participate in the study was administered a copy of the questionnaire in each household selected. The study did not ask the individuals questions regarding the character of their household. This also is vital for ethical reasons because the individual may not have the permission of the members of the household to disclose information concerning them. Some ethical matters were considered before doing the interviews. Involvement was strictly voluntary, and written consent was demanded from the respondents concerning readiness to be interviewed. The goal and length of the study were explained before the interviews, and the participants were informed that they could leave at any time and could decline to answer questions they are unhappy to discuss. COVID-19 guidelines such as using a facemask in public and maintaining a two-meter distance were adhered to in administering the questionnaire. The participant's identifying information was not directly indicated in the research report. In line with earlier studies, the questionnaire's administration followed random sampling of selected households in the street to avoid haphazard sampling and to recognized areas covered (Nwankwo and Okafor 2017; Nwankwo 2019). In each household selected, one questionnaire was administered to ensure a broader coverage of participants from varied backgrounds (Nwankwo and Okafor 2017).

Hierarchical multiple regression was performed to determine the relationship between voting and the variables under consideration. The dependent variable was voting and was set as dichotomous. The independent variables were placed into a categorical scale or dichotomous depending on the nature of the variables. The control variables are COVID-19 safety measures. The participants were asked: "if the general elections are held today, will you vote?" with 'yes' and 'no' responses options provided in the questionnaire. Other questions that will reveal the study variables were also listed on the questionnaire with the response options to choose from. For instance, the question was posed: "to what extent do you agree with the statement that the use of facemask is important for safety from COVID-19" with four response options provided viz. Strongly Agree, Agree, Disagree, Strongly Disagree. The response options are coded as 1, 2, 3, and 4, respectively, because those who agree more to the statement are likely to be COVID-19 enthusiasts who would rather abstain from voting than risk contracting the virus. So, those who disagree more will have higher odds of voting. Other covariates are also rendered in this fashion and coded accordingly. The Statistical Packages for the Social Sciences (SPSS) was employed for the analyses.

Results

Table 2 presents the result of the responses on the views on COVID-19. All the measures have the responses of "strongly disagreed", having higher percentages. Most of the respondents (over 60%) tend to either disagree or strongly disagree, suggesting that the participants do not fancy most of the safety measures in fighting against the virus in Nigeria. A more significant percentage of the respondents strongly disagreeing with the measures think that COVID-19 is not real (44.08%), and social distancing is also not supported by 42.82% of the participants. Among the measures, quarantine/isolation happens to be the most preferred by the participants having about 35%, while facemask use is highly not supported (72.8%).

Table 2 Participants views on measures for curtailing COVID-19
Full size table

Further descriptive statistics for the control variables are presented in Table 3. The mean for quarantine is 2.77, suggesting that the average respondent would either disagree or strongly disagree that quarantining is necessary for curtailing COVID-19. This is because the responses are coded 4 (strongly agreed), 3 (agreed), 2 (disagree) and 1 (strongly disagree) as indicated in Table 3. Since 2.77 is not up to 3.00, we can conclude that on average, the respondents either disagree or strongly disagree that quarantining is necessary for curtailing COVID-19. Similarly, the mean for facemask is 3.045, which suggests that, on average, the respondents agreed that facemask is essential to combating the spread of the virus. Social distancing and home confinement mean of 2.95 and 2.81, respectively, and both suggest that, on average, the respondent disagreed with these measures of curtailing the pandemic. COVID-19 is real has a mean of 2.91, which indicates a disagreement. The standard deviations for all the variables are approximately 1.00, and since they are not low, they do not tend to be very close to the mean. The standard deviation scores signify that the data points are spread out over a large range of values. This suggests that the level of the respondents' acceptance of these COVID-19 measures vary significantly.

Table 3 Descriptive statistics for the control variables
Full size table

The respondents' opinions on whether the COVID-19 safety measures (use of facemask, quarantine, home confinement, social distancing) are essential for contracting or spreading the virus and whether COVID-19 is real was used as a control to see the performance of the traditional predictors of voter turnout during the pandemic scenario. It is expected that people who do not think the virus is not real and that these measures are not vital are important for contracting or spreading the virus will more likely vote during the pandemic than those who do not think so. The result of the regression analysis is shown in Table 4. The regression equation is: Y = a + b(X). Hence,

$$Y = - 1.076 + 0.170\;\left( {{\text{quarantine}}} \right) + 0.153\left( {{\text{facemask}}} \right){-}0.004\left( {{\text{home}}\;{\text{confinement}}} \right) + 0.006\left( {{\text{social}}\;{\text{distancing}}} \right) + 0.029\;\left( {{\text{COVID}}19\;{\text{is}}\;{\text{not}}\;{\text{real}}} \right) + 0.121\left( {{\text{education}}} \right) + 0.034\left( {{\text{age}}} \right) + 0.028\left( {{\text{gender}}} \right) + 0.170\left( {{\text{political}}\;{\text{trust}}} \right){-}0.018\left( {{\text{income}}} \right){-}0.058\left( {{\text{partisanship}}} \right) + 0.136\left( {{\text{married}}} \right) + 0.036\left( {{\text{life}}\;{\text{satisfaction}}} \right).$$
Table 4 Regression estimates of voter turnout during COVID-19
Full size table

The constant, − 1.076, is the estimated value of y when x equal zero. Quarantine, facemask, education and political trust performed excellently in the model as indicated the t-ratios and p-values. T-ratios of the absolute value of 2 or greater indicates that the variable significantly predicts the dependent variable. With t-ratios of 9.788, 6.846, 3.233 and 2.623, quarantine, facemask, education and political trust, respectively, are the significant predictors of voting in a pandemic scenario in Nigeria. The Adjusted R Square = 0 0.309, which indicates that the proportion of the variation in voting explained by the independent variables is 30.9%.

Further, the result shows that without accounting for the traditional predictors of voting, home confinement and social distancing do not significantly predict voting. Apart from home confinement, all other safety measures have a positive relationship with voting, suggesting that the more people do not think that these measures are not vital (bar home confinement), the more likely they will turn out on election day. Not believing in the use of facemask increases the likelihood of voting by 0.155 and 0.185 for people who do not believe in quarantining. Not believing in social distancing increases the odd of voting by 0.014 and by 0.044 for not believing COVID-19 is real. Not believing in home confinement reduces the likelihood of voting by 0.009. In this scenario, people who think that facemask and quarantining are essential for contracting or spreading the virus and that COVID-19 is not real are significantly more likely to vote during the pandemic (p < 0.05).

When the effect of the traditional predictor of voting is accounted for, the impact of these variables on voting decreased. In this case, not believing in using a facemask increases the likelihood of voting by 0.153 and 0.170 for people who do not believe in quarantining. Not believing in social distancing increases voting by 0.006 and by 0.029 for not believing COVID-19 is real. Not believing in home confinement reduces the likelihood of voting by 0.004. However, this reduction does not significantly reduce the performance of the model as the people who do think that the use of facemask and quarantining are important for contracting or spreading the virus and that COVID-19 is not real are still significantly more likely to vote during the pandemic (p < 0.05). In addition, accounting for the influence of respondents' perception of COVID-19, education, and political trust are robust predictors of voting among the traditional variables (p < 0.05). Their effect is positive–having a higher level of education increases the chance of voting by 0.121 and by 0.17 for political trust. This result suggests that more educated people and those who trust the government are more likely to vote if elections were conducted.

All other traditional variables of voting did not significantly predict voting. Still, income and partisanship have a negative effect suggesting that more wealth and support for political parties do not automatically increase the likelihood of voting. It then means that people may be wealthy and support a specific political party but may still not vote, perhaps because of the fear of contracting the virus. The demographic variables considered all increased the likelihood of voting with age having 0.034 odd, being married (0.136), being a male (0.028). Life satisfaction increases the odd of voting by 0.036. All these variables' effect is not significant at 95% confidence level. Overall, as Table 3 shows, the inclusion of these traditional variables of voting improves the model performance as there is a significant increase in (R = 0.564, p < 0.01) in model 1 to (R = 0.625, p < 0.01) in model 2.

Discussion

This study explores the powers of traditional variables to predict voter turnout while controlling for opinions on the safety measures for curtailing COVID-19. The results indicate that most of the respondents (over 60%) tend to either disagree or strongly disagree with most safety measures fighting against the virus in Nigeria. It also shows that people who think that the use of facemask and quarantining are unimportant for contracting or spreading the virus and that COVID-19 is not real are significantly more likely to vote during the pandemic. Moreover, educated people and those who trust the government are more likely to vote if elections were conducted during the pandemic. The general belief that the virus is not real and the safety measures are not required could mean that COVID-19 may not reduce in-person voter turnout in Nigeria, contrary to the finding that it reduced turnout in the US primary elections (Flanders et al. 2020; Morris and Miller 2020). However, this may lead to an increase in the number of COVID-19 cases, as Cotti et al. (2020) argued, and perhaps increased casualties for the aged population (Bertoli et al. 2020) because age increases the odds of voting, as this study's results indicate. Thus, in this scenario, increasing absentee voting may be encouraged as many voters shift to a more health-preserving means of voting (Yoder et al. 2020). There has not been concrete evidence to suggest that the election in Edo and Osun states in Nigeria, for example, is linked to an increase in COVID-19 cases in those states, which suggests that voter turnout may not contribute to the epidemics in places with lower infection levels as Cassan and Sangnier (2020) found in France municipal elections.

The result that people who think that the quarantining is unimportant for contracting or spreading the virus and that COVID-19 is not real are significantly more likely to vote during the pandemic supports the claims that political involvement is a product of how connected the individual citizen is to the broader social and political community. It also bolsters the friends and neighbors' notion—people who live together in the same neighborhood and interact with each other tend to have similar voting behaviors. It also follows that the interaction between people can shape their propensity to vote as friends and neighbors may shape each other's views on the pandemic. It does make sense that a reduction in social interaction caused by COVID-19 can shape the participatory levels of individuals. Hence, the finding suggests, people who believe the virus is real and adhere to safety guidelines for COVID-19 by quarantining and will likely interact less with their friends and neighbors, which may engender political and social alienation affecting their participatory disposition. This can reduce the chances of view change on the pandemic and change in deciding to vote.

Furthermore, the results also show that although not significant at 95% confidence level, life satisfaction increases the odd of voting by 0.036. This finding lends credence to the thinking that COVID-19 imposed constraints on daily living such as social distancing, quarantine, self-isolation, and home confinement may put a psychosocial strain on people because of a decreased amount of social activity through interactions with family, friends, neighbors, or entertainment. Thus, these negative effects on social participation may lower life satisfaction among the public (Ammar et al. 2020). Thus, life satisfaction during the pandemic would be positively associated with voting and dissatisfaction with life to reduce turnout, as this study suggests. This finding thus makes sense of the importance of some literature highlighting distress as a factor of political involvement (Gurr 1970; Thomassen 1989; Klandermans et al. 2008; Buechler 2004; Giugni and Grasso 2015) in theorizing the connection between COVID-19 and political participation.

The result that political trust significantly predicts voting even in the face of opinions on COVID-19 confirms already established ideas that political trust is the primary determinant of turnout in Nigeria and many African states (Kuenzi and Lambright 2007, 2011; Nwankwo and Okafor 2017; Nwankwo 2019). Ezeibe et al. (2020) have argued that people's negligence of the government's directives on the pandemic, such as social distancing and self-isolation, is because of political distrust in Nigeria. Thus, it also supports the argument that political trust has a boosting effect on accepting government policies and actions in the fight against COVID-19. Thus, we can expect political trust to be significantly related to voter turnout during the COVID-19 pandemic. We can infer that the significant level of voter turnout in the 2020 governorship elections in Edo and Osun States after the national lockdown was lifted because the public believes that the virus is not real and public distrust of government actions and messages regarding the severity of the disease in the country.

Moreover, since educated people are more likely to vote if elections were conducted during the pandemic contradicts the assumption that people with higher SES will likely not turn out to vote in a COVID-19 pandemic because they are more knowledgeable and are confident about COVID-19 warnings and thus tend to be more willing to adhere to appropriate practices towards COVID-19 as implied from the study by Zhong et al. (2020). In China, Zhong et al. (2020) indicate that most Chinese residents of a reasonably high SES, particularly women, are well-informed about COVID-19, have optimistic mindsets, and have proper practices towards COVID-19. The variable measuring economic well-being, i.e., income, affirm the established notion in Africa, which indicate that individuals with a higher SES tend not to vote more, e.g., Kuenzi and Lambright 2007; Nwankwo and Okafor 2017) because lower SES voters are encouraged to vote because they can get paid for voting a party in the election (Nwankwo and Okafor 2017; Nwankwo 2018).

Furthermore, the paper aligns closely with the very recent and sparse studies assessing the effects of epidemics on electoral behaviors such as Ebola (Beall et al. 2016; Campante et al. 2020), HIV/AIDS (Mansour et al. 2020) and more recently, COVID-19 (e.g., Adam-Troian et al. 2020; Bisbee and Honig 2020; Fernández-Navia et al. 2021 among others). Regarding COVID-19, the paper finds the contrary of Fernández-Navia et al.'s (2021) results that the turnout was lower in places with positive cases of COVID-19. It suggests that COVID-19 cases may not directly reduce turnout levels because even though COVID-19 skeptics are significantly more likely to vote during the pandemic, their distrust in government make them abstain. Also, because those who would have voted (i.e., those having trust in government) prefer not to turnout rather than contract the virus, the overall level of turnout will be very low. Thus, even though COVID-19 will likely reduce turnout in Nigeria, it does so indirectly.

More generally, the paper's findings negate the recent studies on the political effects of the lockdown caused by COVID-19, which most of them found lockdown to be positively related to trust in the government, and the political institution and governance system and hence increased voter turnout across several European and Western countries (Bol et al. 2020; De Vries et al. 2020; Giommoni and Loumeau 2020; Sibley et al. 2020; Amat et al. 2020). This paper finds the contrary to the claims of these studies– COVID-19 lockdown and associated measures did not increase people's trust in government and more likely to reduced voter turnout.

Conclusion

This study has examined the powers of traditional variables to predict voter turnout while controlling for opinions on the safety measures for curtailing COVID-19. It has shown that there is a general tendency to believe that COVID-19 is not real in Nigeria and that most of the safety measures in fighting against the virus in Nigeria are not essential. It also shows that people who think that facemask and quarantining are not crucial in curtailing the virus and that COVID-19 is not real are significantly more likely to vote during the pandemic. Besides, educated people and those who trust the government are more likely to vote if elections were conducted during the pandemic. While these findings may have several implications, the primary conclusion from this study is that it suggests that political participation during the pandemic can be a product of how connected the individual citizen is to the broader social and political community.

This conclusion is made because the primary finding is that most people do not think that the virus is real, and hence quarantining or related measures, e.g., self-isolation, social distancing, home confinement, are not required. This orientation is significantly influenced by political distrust among many people. The result reveals that people who are not likely to vote do not hold these views and trust the government. It thus follows that people who have the notion the virus is real and practice isolation and social distancing are likely to have limited social interaction with their friends and neighbors. The interaction between people can shape their attitude as friends and neighbors may shape each other's views on the pandemic; hence it can also influence their propensity to vote. It does make sense that a reduction in social interaction caused by COVID-19 can shape the participatory levels of individuals. Hence, as the finding suggests, people who believe the virus is real and adhere to safety guidelines for COVID-19 by quarantining and will likely interact less with their friends and neighbors, which may engender political and social alienation affecting their participatory disposition.

However, since most people think that the virus is not real and do not have trust in the government and hence may not vote, voter turnout in Nigerian elections are likely to be low because people who have trust in government have higher odds of voting but prefer not to turnout rather than contract the virus. The voter turnout in recent elections seems to buttress this point as turnout levels are low (31.6% for Ondo State) and (32% for Edo State). The result thus contributes to the emerging argument that that people's negligence of government's directives on the pandemic, such as social distancing and self-isolation, is because of political distrust in Nigeria (Ezeibe et al. 2020). Thus, we may argue that political trust has a boosting effect on accepting government policies and actions in the fight against COVID-19. Thus, theoretically, the analysis supports the political efficacy model of political participation and the social connectedness model indicating their vitals for exploring the connection between COVID-19 and electoral or political participation more broadly. The argument is that these two models are vital for exploring the nexus of COVID-19 and political participation in Nigeria and perhaps other African countries having similar political settings like Nigeria.