Religion and Voter Choice Homogeneity in the Nigerian Presidential Elections of the Fourth Republic

3.1 Data

In Nigeria, regional patterns of voting can easily be discerned, metropolitan and local patterns will be challenging to analyse due to the scarcity of data at these levels. Most studies in Nigeria have used the state-level data which is readily available. Thus, the analysis done in this study is at the state level not municipal or local level. The data consist of votes received by political parties in the presidential elections of 1999, 2003, 2011 and 2015. Data on 2007 election is mostly unavailable at the state level, and the 2007 election was widely replete with irregularities hence the reason it has been omitted (Adebayo and Omotola 2007). The data were collected from official results released by the Independent National Electoral Commission (INEC)–the electoral umpire. Data on socio-economic variables were obtained from the Annual Abstract of Statistics (National Bureau of Statistic NBS 2008, 2012, 2016). Due to socio-economic data limitations, the analysis is run at the state level. One might wonder about the pertinence of a study of voters’ choice homogeneity at this level of analysis (the population of Nigerian states being between 1.5 and 10 million). While the electoral data are available at a much more detailed level of study (at least for some of the ballots) that would offer the possibility of measuring the intra-state degree of VCH, it was difficult to access the data at the local level from the electoral commission. The major impediment of running the analysis at the local levels is the fact that the data on the explanatory variables are not available at the local level.

Up to data official data on Nigeria’s religious characteristics is not available hence the account of Ostien (2012) of University of Oxford was used. It is pertinent to note that the religious composition of the states dates to censuses in 1952 and 1963 (in Ostien’s paper). These data are the best data available. From the data, we can observe the diversity of the religious mixes in 1952 and 1963, as between Christians, Muslims and Others, both within the regions and between the regions. Also, the data show two missionary religions, Christianity and Islam were both trying to win converts from the Traditionalists, and both were succeeding, to different extents in different places. Thus, there is nearly even decrease in the percentages of “Others” from 1952 to 1963, across all regions and states except in Kano and Katsina where there were slight increases. Nevertheless, the Traditionalists were still holding their own in many places in 1963 and as Sampson (2014) indicates traditional religion still has a fair number of followers. The challenge is how to know whether and where the Traditionalists still exist in substantial amounts today.

An Afrobarometer map (Figure 1 ) that shows the result of a survey of the Muslim population in Nigeria perhaps provides a clue of the distribution of Muslims in Nigeria. As in Figure 1, Sampson (2014) notes that the states in North West and North Eastern Nigeria where Sharia practice is legal to have a majority Muslim population. These states are Borno, Bauchi, Gombe, Jigawa, Katsina, Niger, Kaduna, Kano, Kebbi, Sokoto, Yobe and Zamfara, except Taraba State that has an almost equal distribution of Christians and Muslims. The Muslim-Christian share of Adamawa population is contestable as both religions thrive. Figure 1 shows that there are about 60–79% Muslims in Adamawa but the fact that Sharia is illegal there is an indication that Christianity is a significant religion because most of the tribes who rejected Islam converted to Christianity. The states in the Middle Belt plus the Federal Capital Territory are religiously mixed except Benue and Kwara. In Benue, Christians dominate, followed by Traditionalists (Sampson 2014) and in Kwara, Muslims dominate (see Figure 1). According to Sampson (2014), the South-West zone is religiously mixed, and this is evident in Figure 1. Based on this comparison, we can say that the 1963 data to some extent is still reliable especially for states in Northern Nigeria and the Old Western Region except Delta and Edo States where Christianity and Islam have gained some grounds, respectively. Thus, for all the states in Northern Nigeria and the Western Region, the 1963 data and Figure 1 are compared to make a judgement of the likely distribution of the religion.

Figure 1:

Percentage distribution of the Muslim population in Nigeria.

Data Source: Afrobarometer Survey (2012).

For example, Figure 1 shows that Sokoto, Kebbi, Zamfara, Katsina, Jigawa, Kano and Gombe States have between 80% and 100% Muslim population based on a 2012 survey. By comparison, these states (except Gombe [75%]) have a similar distribution of Muslim population in 1963 as indicated in Table 1 that shows the distribution of religions in all the states in 1952, 1963 and likely present. So, we use the 1963 data for these states and update the Muslim population of Gombe State as 80% and update the Christian population based on the percentage change in Christianity share from 1952 to 1963 thereby reducing the share of traditional religion. Also, as an example, Sampson (2014) notes that Edo State now has an almost equal percentage of Christians and Muslims, but Figure 1 shows that Edo State has no more than 19% of its population as Muslim, but it was only 7.4% in 1963 from 7.1% in 1952. So, it is hard to confidently say Edo State has an almost equal share of the Muslim and Christian population, hence, its Muslim population is pegged at 19%. This sort of calculation was done for the Northern Region and Western Region because the 1963 data almost matches Figure 1.

However, it is essential to recognise that the 1963 data is not entirely reliable in the case of the Old Eastern Region because of a significant decline in traditional religions and the growth of Christianity. Controlling for these changes is a bit problematic as the subsequent censuses after 1963 did not collect data on ethnicity and religious make-up of the population. To bring the data up to date to a reasonable extent, we use the decadal percentage change in the Christian population in each state in the Old Eastern Region to a point where it is reasonably in tandem with Figure 1. For example, Enugu had 18.7% Christians in 1952, 0.5% Muslims. In 1963, it had 47.9% Christians representing a 29.2% increase in Christian population while Muslim population remained 0.5%. We assume the rate of conversion from Traditional Religion to Christianity was constant and we take that further by 10 years, we have 77.1% Christians. If it is projected further into the next census year, we have 106.3%. This figure is above 100%. So, we stop at 77.1% since, from Figure 1, Enugu has between 20% and 39% Muslim population. Overall, it is expected that the Christian population in Enugu will not be less than 77.1%. Thus, for Enugu State, we have 77.1% Christians, 20% Muslims (going further than 20% means the total will exceed 100%) and 2.9% other religions. This calculation was done for other states in the Old Eastern Region except for states where traditional religion has dramatically declined, e.g. Abia. The likely religion distribution of the states is provided in Table 1.

3.2 Data Analysis

The level of homogeneity of voters’ choice in each state is considered as the inverted fractionalized Herfindahl index (HI) expressed as (Dalton 2008: p. 903):

HI=∑Xi2

where HI represents the level of VCH, ∑Xi2 represents the summation of the square of the percentage share of votes received by the parties (for votes received by parties, see Table 2) in the election. In this way, the voters ’ choice homogeneity in each of the 36 states of the federation and the Federal Capital Territory (FCT) in each election year was calculated. The Herfindahl index can be interpreted as the probability that two voters selected at random from among the voters in a state would have voted for different parties. Hence the higher the HI values the less heterogenous the voter choice in that state. In other words, the higher the HI values, the more voters in a state tend to vote for a single party (i.e. homogeneous). If the HI is low, then voters in that state have dissimilar (heterogeneous) choices, i.e. they voted for different parties. The HI value does not tell which political party voters prefer but an indication of heterogeneity in voters’ choices or otherwise (see Table 3 for the results).

Multivariate regression analysis was done with the VCH as the dependent variable and the socio-economic variables, religious homogeneity as independent variables (explanatory variables) to determine if each variable explains VCH taking into account the other variables. Religious diversity is generally measured as the probability that two randomly chosen individuals are not of the same religion. Hence a Herfindahl-type fractionalization index seems to be a more robust measure of it. It is expected that the higher the religious diversity index, the less religiously heterogenous a state is, i.e. the more religiously homogeneous the state and the more similar the voter choice in that state and vice versa. See Table 3 for the result of the calculation of the religious homogeneity index (RHI).

4.1 The 1999 Election

As the regression estimates in Table 4 shows, there is a low negative and insignificant (at 95% confidence level) relationship between religion and voter choice homogeneity implying that as religious homogeneity increases, VCH decreases. A unit increase in religious uniformity leads to 0.126 decreases in VCH. If religious identity were a determinant of voter choice in the 1999 election, the coefficient would have been positive and strong. Unemployment has no significant relationship with VCH while absolute poverty shows a very weak negative association with VCH. A unit increase in absolute poverty is associated with 0.002 decreases in VCH in 1999 election, respectively. This result indicates that neither religion or socio-economic variables significantly predicted vote choice in the 1999 election. Some explanations can be made in respect of this result. First, the fact that the two candidates, Olu Falae of the coalition of the Alliance for Democracy (AD) and All People’s Party (APP), and Obasanjo of the People’s Democratic Party (PDP) are both Christians and Yoruba meant that religion is not a significant determinant of vote choice. Nonetheless, the Southwest states, i.e. the Yoruba region including Ekiti, Ondo, Osun, Oyo, Ogun and Lagos voted massively for AD/APP as shown in Table 2.

Secondly, is the fact that the PDP had a national appeal. The PDP’s widespread appeal is a function of the support it gets from many prominent politicians and military generals and power brokers from different regions of the country (Ihonvbere 1999). It is also argued that the Muslim dominated states in northern Nigeria were left with no choice than to choose between two Christians. However, the fact that they voted more for the PDP supports that argument that Muslims in the north considered Obasanjo as a friend of the north who will protect if not augment their interest (Ihonvbere 1999). Note that despite being a northern party, the APP did not receive substantial support in the north (see Table 2) except in Zamfara where it got 64.1%, Yobe (53%) and Sokoto (56.1%). Thus, the religious affiliation of the parties might not have been a major deciding factor of vote choice. One crucial element for this might be the membership of the prominent leaders of the north who are Muslims such as the Vice-Presidential candidate, Alhaji Atiku Abubakar, late Shehu Yar’Adua and other military generals in the military regimes.

4.2 The 2003 Election

The analysis of 2003 presidential election does not show a positive relationship between religious identity, the socio-economic variables and VCH. A unit increase in RHI, unemployment and absolute poverty is associated with a 0.036, 0.003 and 0.003 decrease in VCH, respectively. Compared with the 1999 election where the two candidates were Christians, the All Nigerian People’s Party (ANPP) had a Muslim as the candidate of the person of General Muhammadu Buhari, a former military dictator who was Head of State from 1983 to 1985. The ANPP is an offshoot of the defunct APP, so it has its base in the northern region. Therefore, the presidential contest was a faceoff between the incumbent Obasanjo of the PDP, a Christian and Buhari of ANPP, a Muslim. Before the election, Buhari has been a core advocate of spreading the Sharia principle across Nigeria and thus enjoys substantial support from the Muslims in northern Nigeria.

Indeed, compared with 1999 election, the 2003 election result (Table 2) shows that ANPP received a good share of the total votes in the North West and some parts of North East including Bauchi (62.1%), Gombe (52.9%), Jigawa (80.4%), Kano (74.9%), Katsina (76.2%), Kebbi (65%), Zamfara (80%), Yobe (64%), Sokoto (73.3%). Nevertheless, the regression result indicates that the relationship between religion and vote choice is very low and negative. Thus, even though it could be argued that religious identity was a determinant of vote choice in Northern Nigeria, it offers an unsatisfactory explanation in a general term. A significant reason for this is that the South West and the Middle Belt states which have a mix of both Christians and Muslims who voted for Obasanjo of the PDP meant that the RHI could not correlate positively with VCH.

The implication of the results so far is that although religion has some part to play it cannot alone provide an excellent explanation of vote choice in Nigeria. Some other factors might be considered here such as the popularity of the candidates and ethnicity although the influence of ethnicity on vote choice is not new in the literature. Regarding the popularity of the candidates, we could argue that the rise in the 2003 vote share of the ANPP in the north compared with the AD/APP alliance of 1999 is because Buhari is popular in northern Nigeria as a former governor and head of state. Whereas, Olu Falae was mostly known in the South West where he got his most substantial support in 1999. Another interpretation of this shift can be attributed to ethnicity because Falae who is Yoruba got massive support from the Yoruba States while Buhari who is Fulani got enormous support from the Hausa-Fulani States. This explanation is not new and adds nothing significant to the analysis.

Regardless of the ethnicity of the candidates, party-identification might be a significant predictor of vote choice. For example, even though some voters in Igbo dominated South East States voted for All Progressive Grand Alliance (APGA) such as in Abia (34.9%), Anambra (32.4%), Imo (27.7%) and Enugu (15.7%). The vast majority voted for the PDP whose candidate is Yoruba not least the South-South states where many minority ethnic groups exist despite the formation of APGA by the former leader of the defunct Biafran Republic, Odimegwu Ojukwu. Biafra was a secessionist state whose territory covers the old Eastern Region (presently, the South East and South-South geopolitical zones) attempted to break away from Nigeria leading to brutal war from 1967 to 1970.

Also, in 1999, the South West States voted for AD/APP particularly because they preferred a party, they considered their own, i.e. the AD than PDP. In 2003, the AD did not field a candidate for the presidential seat which perhaps switch the determinant of vote choice from party-identification to ethnicity which meant that the PDP’s candidate who is Yoruba was preferred to Buhari of ANPP. Historically, the leaders of the former Action Group and National Council of Nigeria and the Cameroons (NCNC) which have their strongholds in the Western Region and old Eastern Region were among the founders of the AD and PDP, respectively. Thus, as we shall see more clearly in the 2015 election, party-identification can be a good explanation of vote choice in Nigeria, but how party-identification can shape vote choice might not be entirely straightforward, it can correlate with religion, ethnicity, and the membership or non-membership of prominent political elites and leaders.

4.3 The 2011 Election

Goodluck Jonathan was the Vice President to the late President Musa Yar’dua who was declared the winner of the 2007 election. The death of Yar’dua in 2010 meant that Jonathan was sworn in as President and subsequently won the PDP primary election. The emergence of Jonathan as the PDP candidate in the 2011 election meant that Buhari will face yet another Christian from the south. Consequently, when the result of the elections is mapped as Figure 2 shows, party support became polarised in a north-south dimension with a large support for Jonathan’s PDP in the south and Buhari’s Congress for Progressive Change (CPC) in the north. Compared with previous elections, the regression coefficient shows a weak positive relationship between religion and VCH. A unit increase in religious identity is associated with 0.300 increase in VCH but not significant at 95% confidence level. Buhari left the ANPP to establish the CPC after disagreements with party members in the ANPP.

Figure 2:

States where parties won in the presidential elections.

Putting religion aside, for now, the tremendous support of the CPC in the 2011 election compared with the ANPP support in the 2003 election shows that Buhari’s popularity in northern Nigeria was a factor as earlier argued given that the CPC was newly formed just before the election. This argument can be substantiated by the fact that the Action Congress of Nigeria (ACN) which was a new party with its base in the South West did not receive the most substantial support of voters there not least the AD which has been known as the party of the region instead, the PDP did. This result puts party-identification argument earlier noted in question because if it were the case that party-identification is a strong determinant of vote choice, the South West would have voted the AD or ACN. However, we might consider that Jonathan was considered a closer candidate than Ribadu because he is from the south.

The ACN was a faction of the AD led by former Governor of Lagos State, Bola Tinubu. Nuhu Ribadu was the candidate of the ACN. Ribadu was a former Chief of the anti-graft agency, Economic and Financial Crimes Commission (EFCC) who became famous for having spearheaded the fight against corruption at the EFCC (Lewis 2011). The ACN, however, only managed to win two states namely Kebbi and Osun. The reason is perhaps that Jonathan’s reputation at that time was still considered an asset; having worked as a lecturer, many people thought his education would bring the much-needed good governance as opposed to Buhari who has a military background (Lewis 2011). Besides, Jonathan received massive support from the south-south and southeastern states because it is the first time someone from the region will have the opportunity to be president since the end of the civil war in 1970. So, massive support was needed to ensure victory. Nevertheless, there is no denying that religion did play a role in voters’ choices, but the influence it had was weak and insignificant.

4.4 The 2015 Election

Once again, the 2015 election was a contest between Buhari and Jonathan. This time, Buhari had more support compared with previous elections due to the merger of CPC, ACN and factions ANPP and APGA in 2013 to form the party of the All Progressives Congress (APC). As Table 4 shows, there is a significant and robust positive relationship between religion and VCH. The socio-economic variables show a weak negative association with VCH. A unit increase in religious identity is associated with 0.675 increase in VCH at 95% confidence level. Thus, at first glance, we can say that religion played a significant role in voter choice in the 2015 election more than any other factor compared with other polls. However, it is essential to put this in the context in which religion became a significant determinant of vote choice given that the two major candidates for the 2011 and 2015 election were the same. While Jonathan won in 2011, Buhari won in 2015. So, what are the reasons for this change?

One of the narratives put forward by the media and academics is that the refusal of Jonathan to step aside for a northern Muslim to become president after the death of Yar’Adua jettisoned the power-sharing arrangement between Muslims and Christians. Another argument is that the support Jonathan received declined due to his poor performance as issues of corruption and insecurity trailed his administration amid the growing popularity of Buhari as a man of integrity who can stem the tides of corruption and insecurity in the country. Also, it is argued that the merger of opposition parties into APC made the APC stronger to challenge the PDP.

If we put the narratives in the context of the result in Table 4, we might conclude that the refusal of Jonathan to step aside for a northern Muslim to become president after the death of Yar’Adua was entirely the case. However, I would assume other explanations for the significance of religion in 2015. One is that the south-west and north-central states are more religiously heterogeneous and thus may have sub-state level homogeneity. On the other hand, if religious voting would cause the voter homogeneity, we could also have assumed more homogeneous voting in the northern states in 2003 when Buhari was associated with support of the Shari’a agenda. In all, if the significance of religion in the 2015 VCH was due to “the attempt to go off track of the unofficial but acknowledged power-sharing arrangement” the effect would have been there already in 2011. Thus, we can argue that the merging of opposition parties to APC made VCH more likely as previous split votes now were combined.

Nevertheless, the mere merger of opposition parties may not directly translate to vote choice homogeneity. Placing this in the context of the preceding analysis is essential to have a better comprehension of the 2015 election. One important intermediate factor for this might be the decamping of many prominent leaders of the north and Southwest, e.g. former Vice President Atiku Abubakar, the Speaker of the House of Representatives, Aminu Tambuwal among others from the PDP to APC not least the resignation of membership of the PDP by the former President, Olusegun Obasanjo. So, the mass movement of leaders of the north and southwest perhaps sent a strong message to their people thereby shaping their vote choice.