At the last count, no fewer than eight polls have been conducted on the forthcoming presidential elections by different organisations. Notable organisations such as ANAP/NOI (three polls), Bloomberg/Premise, Stears, Nextier, EiE/SBM, to name a few, have made projections based on the views of their respondents. This is not to mention the countless independent and citizen-driven analysis making the rounds on social media.
This, according to experts, is a commendable signal that data has become a crucial consideration for election analysis in Nigeria.
Public polling, when done right, still remains the most reliable way to predict election results. But of course, a sheer enthusiasm for polling and – for that matter — massive collection of data do not preclude analysts from making consequential statistical errors that could render predictions inaccurate.
Admittedly, many of the polls have endeavoured to be statistically thorough, but there are still issues such as convenience sampling, ascertainment bias (e.g even with random sampling, more enlightened, urban-based Nigerians are likely to be polled by phone than rural voters), huge number of undecided voters, fewer number of sampled voters (and voting population) actually ending up voting, among others.
Politics is complex, and so is statistics. Needless to say, elections are a very unwieldy affair, and even the most sophisticated data models might fail to sufficiently account for the unpredictability of human factors.
As such, just before Nigeria goes to the polls, Dataphyte brings a new perspective to the polling conversation that does not predict voters’ behaviour but measures voters’ perspectives on both administrative and political factors that may influence how people vote. For instance, one of the questions is, what do Nigerians actually think of candidates’ popularity, party popularity and effectiveness of political campaigns at the local government level? This is important because elections in Nigeria are influenced by local politics — “politics is local”, they always say.
Methodology
In partnership with DEAN Initiative, we polled 1,356 Nigerians across the 36 states and the FCT using a simple random sampling method. A few facts about the data are noteworthy: first, the data do not measure respondents’ probable voting choice, rather it measures their perception of candidates, parties and election readiness, generally.
Secondly, due to timing and resources, samples seem to be disproportionately larger in some states than others. However this was resolved by weighting the data before analysis. More so, our data reflect more state-level analysis than national, and so, states with smaller population (and voters) tend to have fewer respondents overall. Measured against the population, our sample size gives us a margin of error of +/- 3 percent at 95 percent confidence interval (i.e. were our survey to be conducted again 100 times, 95 times our observed results would be within + or – 3 percentage points of the population value).
Thirdly, where we measured perceptions on political parties, the survey focused only on the top four political parties, based on public opinion: All Progressives Congress (APC), Labour Party (LP), New Nigeria Peoples Party (NNPP), and the Peoples Democratic Party (PDP).
For the survey, we asked a total of 25 questions, grouped into five core themes: Political Party Activities and Engagement (party popularity; campaign reach; campaign effectiveness; vote buying); Candidacy (questions on women and youth candidacy; and candidates’ popularity), Electoral Integrity and Security(questions on election security; election personnel; trust in INEC; campaign freedom; satisfaction on PVC collection; payment for PVC collection; proximity of polling units; effects of cashless policy on vote buying); and Voters Attitude (trading of PVC; vote selling; voters apathy).
See Table 1 below for a summary of the survey questions
However, for the purpose of analysis, we have selected only 11 most relevant variables for this piece.
Now, let’s dive into the poll results.
Political Parties and candidates
- How effective are political campaigns at the local government level?
As noted, elections in Nigeria are won at the local level – the very structure of voting in which the electorate vote at polling units (which each has less than 500 registered voters, according to the Electoral Act) attests to this. This means for a party’s political campaign to be judged effective – and by extension for its candidate to be truly popular - both must resonate strongly in residents’ immediate community beyond the usual off-the-top, macro and social media perception often used to judge candidates or party popularity. But often, most political parties – even the traditional ones with seemingly stronger financial power – concentrate on state level rallies, with lesser efforts directed at local governments specifically.
The majority of our respondents believe that local level campaigns by political parties are effective or very effective at mobilising citizens to vote for that party’s candidate.
However, these dynamics change a little when examined at the state level. For instance, compared to other states, Benue, Cross River, Ebonyi, Edo, Osun, Oyo, and Taraba have substantially high number of respondents who think LG level campaigns are ‘fairly effective’ – although overall majority in these states still consider local campaigns effective or very effective.
- Party and candidate popularity
Which candidates are most popular at local governments?
Overall, Bola Tinubu and Peter Obi closely trail each other in popularity among respondents, with Tinubu leading Obi by only 3.9 percent percentage points. Atiku comes third in popularity, while Rabiu Kwankwaso of the NNPP trailing them.
But can we really say Bola Tinubu is the most popular, given the very contestable and tiny difference between him and the next candidate, Peter Obi? State level results provide a clearer view. Peter Obi is considered the most popular candidate in 15 states and the FCT (mostly states in the South-West, South-South and North-Central regions), while Bola Tinubu takes the lead in 11 states across the South-West, part of North-Central and part of the North-West. Atiku Abubakar follows with a lead in nine states mostly in the North-East and North-West region. Kwankwaso is considered most popular only in Kano State.
This representation fits the popular narrative in other polls. But a common mistake as with many other polls is the tendency to ask respondents “Which candidate” they would vote for, rather than “Which Party”. Yet, this distinction is very important as we will observe below.
- Which party is most popular at the local government level? Does Peter Obi’s popularity translate to top-of-mind awareness of LP?
When examined overall, the majority of respondents (47.9 percent) consider APC as the most popular party at their local government. This is followed by the PDP (35.6 percent) and LP (14.2 percent)
At the state level, APC is the most popular party in 20 states, while the PDP is most popular in 13 states, and LP in 4. This difference is statistically significant, using a chi-square test to examine whether there is a difference in the distribution of responses to party popularity among the 36 states and the FCT (This test has been applied to all data presented at the state level in this piece, and the difference among states remains statistically significant for all responses presented at state/regional level).
A note of caution worth mentioning here is that presenting the data this way only tells us which party has the highest number of respondents, choosing it as most popular in a given state, not the proportion of respondents choosing each candidate per state. But in reality, party A might be the most popular across many states only among a small majority of 30-40 percent (while other three parties share the 60 percent) and then performs poorly across other states (e.g only. 5-10 percent considering it as popular). Its closest contender, party B, might be most popular in fewer states but with a large majority in those states (e.g. 60 percent-70 percent), and closely trailing party A in other states where party A has won (e.g 35 percent as against Party A’s 40 percent).
As such, a more nuanced way to consider party popularity in states is to look at what proportion of the population in each state considers a party as most popular. This differentiation is important, given that the victorious candidate must win the overall majority votes and an absolute majority of at least 25 percent of the total votes in two-thirds (24) of the 36 states. The table below shows party popularity in each state.
Table 2: Party Popularity
However, whichever the data on party popularity is presented, results show that the Labour Party is way less popular than the APC and the PDP across the states. LP is considered most popular in only four states, compared to the APC’s 20 and the PDP’s 13. What explains this? Why would a good number of Nigerians have knowledge of Peter Obi, but not the party? Compared to the number of respondents who consider Obi as the most popular candidate, a smaller number think of LP as the most popular. This theory holds even truer when we examine party popularity by states, and we see that LP as a party is less popular where Peter Obi himself is considered most popular.
This seems antithetical, but one possible explanation is that Peter Obi as a brand seems to have both Top-of -Mind Awareness and Salience among respondents, but it is quite likely that his party is not well thought of or noticed to the same degree by respondents. This is not far-fetched, given that unlike the other two prominent candidates who come from Nigeria’s most dominant political parties, Peter Obi comes from the LP – a relatively newer party, and ‘third-force’ whose relevance is driven by Peter Obi’s candidacy rather than existing party structures.
- Which party has highest number of followers at the local government level?
When asked which party has the highest number of followers in their local government, the majority of respondents believe the APC has the highest, followed by the PDP, while a much smaller fraction believe the LP and NNPP have the highest followership in their local governments. It is important to stress that this is a perception of public followership or popularity, rather than a test of which party respondents follow.
At the state level, both the APC and the PDP are considered to have the highest followership – leading in 15 states, respectively, followed by the LP with six states, and NNPP with 1 state.
Again, Table 3 below shows more detailed figures on followership for all the 4 parties in each state.
Voters' attitude
- Vote-buying
When asked what their perception of political parties engaging in vote-buying and other financial inducement activities on election day is, the majority of respondents opined that the tendency for parties to engage in such activities is moderate to very low. Perhaps, this is informed by the belief that the electoral process is now more credible and gives less room for such manipulative tactics to count.
*The difference in vote buying perception by state is not statistically significant, and as such has been excluded from our analysis.
- Vote-selling
Similarly, about 45 percent of respondents think that the tendency for voters to want to sell their vote on election day is very low or low, while 29 percent believe that this is moderate. Again, a smaller proportion of Nigerians believe that citizens could engage in vote-selling activities come election day
7) Do citizens think CBN naira redesign/cashless policy will curb vote-buying?
The last couple of months have seen a dramatic turn of events for many Nigerians, with the Central Bank of Nigeria (CBN) rolling out its currency redesign policy. Interestingly, what should ordinarily be a monetary policy and economic issue has spiralled into political tension across the country. The policy could affect the election in many ways, with the court having to mediate and some political parties threatening to pull out of the polls if the CBN extends the new naira notes deadline. Several public figures and political analysts have suggested that the policy will curb vote-buying.
But what do citizens think? A vast majority of respondents (71.2 percent) overall believe that the new CBN naira redesign and cashless policy could curb vote-buying activities on election day.
*The difference in the perception of the effect of cashless policy effect across states is not statistically significant.
8) Citizens’ preparedness and eagerness to vote
As seen with many election analysis reports, a very important factor in the 2023 elections is voter turnout. For one, there has been a massive increase in the registration and collection of new PVCs, which might suggest increased enthusiasm on citizens’ side. Also, an increased sense of democratic dissatisfaction in the current government may spur more people to come out to vote. Our analysis shows that quite a high proportion of Nigerians feel highly prepared and eager to vote on election day (69 percent).
However, state-level data present a mixed picture, with a slightly high proportion of respondents rating citizens’ preparedness as moderate.
See Table 4 below:
Election integrity & security
9) What is the level of trust citizens have in INEC’s administration?
INEC’s election administration has arguably improved over the years, with several new reforms initiated by the commission, including but not limited to the adoption of technology for accreditation and results transmission, the creation of new polling units, adoption of a new Electoral Act, innovation in voters’ registration and PVC collection, among others.
But what is the level of trust or credibility citizens place in INEC? At the aggregate level, more than half of respondents (54.35 percent) have a moderate to high trust in INEC’s readiness and credibility to conduct the elections, while about 15 percent hold very high trust in the commission.
At the regional level, most citizens hold a ‘moderate’ trust in INEC’s administrative credibility, with the exception of South-West, where trust is ‘high’.
10) Proximity of polling units – are polling units close to voters?
It is important to note that polled responses here may not accurately reflect the proximity of polling units to all voters, since, typically, respondents would answer based on proximity to their own residence, place of work or other familiar routes. However, two separate but closely-linked questions could be used to establish a premise for the nearness of polling units to citizens.
The first question, “Are polling units within close proximity to citizens?” measures the nearness of polling units to citizens, without factoring in whether this is a trekkable distance or whether citizens would require some means of transportation. The second question, “Are polling units within walking distance to citizens” accounts for the location of PUs relative to transportation. This is important, given that on election day, commercial forms of transportation tend to be limited, and citizens often either have to use private transportation means or walk where this is not available.
Both results show that the majority of citizens consider polling units to be very close to citizens. Whether those PUs are the ones where citizens have registered to vote or not is a different question.
11) What is the level of safety citizens feel going out to vote in general elections?
The majority of citizens feel that it is safe or very safe to go out to vote (72.35 percent combined).
Next, we present the data at the regional level rather than state, and the picture is quite similar to the national aggregate. However, there is a slightly lower perception of safety in the South-East compared to other regions. Could a slightly lower perception of safety affect voter’ turnout in the region? What might this mean for Peter Obi, the leading candidate in the region?
The 2023 presidential elections is yet the most keenly contested polls since Nigeria’s democratic transition in 1999. A much heightened level of democratic expectation from citizens has required that the leading candidates put forward their best political strategies, raising difficult-to-answer questions as to who would win the polls come Saturday, February 25, 2023. Our analysis has attempted to answer some of these puzzles, covering a range of issues — from voters behaviour to political party campaigns and party popularity. As we head to the polls on Saturday, we are eager to see how the data play out, especially in relation to candidates’ popularity versus party popularity, voter turnout, and vote buying occurences.