The inspiration for Kwakol Research's election forecast comes from the desire to give the general public and other stakeholders a trustworthy source of information and analysis in order to aid them in better understanding the dynamics of the forthcoming election and making informed decisions.
The Kwakol Research Nigeria 2023 presidential election forecast exercise is a forecast on vote intentions. Thus, we have combined three procedures to forecast the 2023 presidential election in Nigeria. This includes a survey poll, historical data analysis, and endorsements from political groups and influential individuals in the Nigerian political space. Our survey poll tracked the INEC registration database to help design the sample size based on the socioeconomic distributional characteristics of our respondents. Our historical data analysis focused on past party performance in elections to capture party structure, incumbency to capture resource control, and a personality factor to capture the influence of each presidential candidate representing the major political parties. We considered endorsements from sociocultural and ethno-nationalist groups such as Ohaneze-Ndigbo, Miyetti Allah, the Pan Niger Delta Forum, and the Southern and Middle Belt Leaders Forum. Each procedure was assigned unequal weights using conventional statistical procedures. We also conducted a scenario-based analysis with three scenarios: electoral violence, the possibility of a run-off, and last-minute alliances between political parties.
Also, unlike a majority of the polls and forecasts on the forthcoming presidential election, we highlighted political parties contesting the election in the data design, collection, and reporting instead of the presidential candidates.
For the survey poll itself, whose results account for at least 60% of our forecast model, we deployed a purposive and stratified sampling method and carried it out through in-person interviews conducted by 21 enumeration agents. The poll was conducted in Nigeria's six geopolitical zones and focused primarily on voter intent. Only people who verbally attested to owning a permanent voter’s card (PVC) and a willingness to vote in the 2023 General Elections in Nigeria were picked as respondents by the field enumerators.
Overall, 20 survey locations spanning villages, towns, and cities across selected states in Nigeria were selected, of which 1008 respondents were selected and interviewed. This select number of survey locations per geopolitical region and rural-urban classification was picked based on a combination of the judgment of the enumerators and the principal investigators/supervising researchers. The sample size was designed to replicate the characteristics of the 93 million Permanent Voter Card (PVC) holders registered by the Independent National Electoral Commission (INEC). This includes the occupational background, age and gender distribution, average monthly earnings, rural-urban classification, and the respective states that the PVC holders were registered in.
It is important to note that, unlike the majority of the polls in Nigeria, we polled more registered voters in rural Nigeria than registered voters in semi-urban Nigeria and urban Nigeria, respectively.
Figure 3. Source: Kwakol Research (2023)
The poll survey was replete with questions that inquired about potential voters' intentions for the 2023 presidential, gubernatorial, and senate elections. These important questions were preceded by inquiries on the respondents' biodata, including their age, gender, occupation, level of education, and income. The questionnaire contained a total of 27 open-ended and closed-ended questions that were conducted in English, Pidgin English, Hausa, Fulfulde, Yoruba, Igbo, Tiv, Igala, Egbira, Idoma, and Jukun.
STATA was used for the analysis, which was both descriptive and predictive. Bar and pie charts were generated using Datawrapper to represent the opinions of the respondents regarding interesting topics. The analysis was based on the self-reported willingness of the respondents (self-reported PVC holders) to vote, with 100% of the study group being considered as likely voters. The findings are presented in detail, covering different modules of the survey instrument.
Aside from highlighting the vote intent of the respondents we have polled in the survey poll exercise, we have further disaggregated vote intent by age groups, occupational background, highest educational attainment, average monthly earnings, gender distribution, rural-urban classification, and other significant factors.
“All models are wrong, but some are useful.” - George E. P. Box
In statistics, the quote “all models are wrong” is a popular aphorism. It simply means that statistical models are not adequate enough to capture the complexities of reality and the eventualities of the future, but they are useful, regardless. For this reason, and in the context of this presidential election forecast exercise, we proposed a simple forecasting model with a simple explanation because we also wanted to adhere to Occam’s Razor, - Occam’s Razor suggests that the most straightforward and logical explanation should be chosen over more complicated ones. We also did this with the full understanding that although people like simple narratives, the science behind predictive analysis and models is not that simple.
To forecast the outcome of the 2023 Nigeria presidential election, we deployed three key variables: poll survey results, historical data/status-quo political patterns, and endorsements from socio-cultural, ethnonationalist, religious, and political groups. We also assigned points to these variables to reflect the unique elements of Nigerian politics and elections. The allocation of points to the respective variables in this forecast model was solely based on the research team’s subjective judgment and the team’s expert analysis therein.
It is important to note that according to the 1999 Nigerian constitution and INEC, a party must receive at least 25% of the total votes cast in two-thirds of the states and the Federal Capital Territory to win in the first round. If no party meets this threshold, then a run-off election between the two parties with the highest vote total will take place, and the winning party must receive at least 50% of the total votes cast.
Now, to our unique forecast model:
Polls are arguably the most important factor when predicting election outcomes. However, we strongly believe, that the right mix of data, nuance, and context is vital for a comprehensive forecast model.
We proceeded to deploy a maximum predictive forecast value of 100 points that can be generated by combining the allocated points to these three elements: poll survey results, historical data and status-quo political patterns, and endorsements from groups based on the subjective judgments of the researchers. Of the 100 points, 60 were assigned to the poll survey results. This is because this forecasting exercise is largely based on predicting vote intentions, and opinion polls that are properly stratified and nationally representative provide an accurate assessment of the population's voting intentions. Furthermore, a maximum value of 30 points was then assigned to historical data & status-quo political patterns. The maximum value of 30 points allocated to these historical data & status-quo political patterns is a full sum of 12 points for party performance in the 2015 & 2019 presidential elections, 8 points for a party’s incumbency in government during this current electoral cycle, and 10 points for the personality factor of the respective presidential flagbearers of the frontline political parties. For the final variable, the endorsements by social, cultural, religious, and political groups of the presidential candidate pairings of respective political parties, an additional maximum value of 10 points were allocated.
This means that based on our forecast model, the higher a political party’s predictive forecast value points are, the higher their chances of winning the forthcoming presidential election in the first round.
We believe that this forecasting model provides a more accurate prediction of election outcomes and is a valuable tool for understanding the dynamics of the Nigerian voting public.
n = P + H + E
“We will see that simple mathematics and statistics, lots of common sense, and a good understanding of the ground reality or domain knowledge together can be very effective.” - Rajeeva L. Karandikar
The fundamental arithmetic equation for our forecast model is n = P + H + E where n is the sum total of the predictive forecast value, P is the poll survey result, H is the historical data and status-quo political patterns, and E is the collated endorsements by the aforementioned social groups based on the subjective judgments of this exercise’s undertakers (researchers). Here, the maximum value of n = 100 points.
We then mathematically established the predictive forecast value points of the respective frontline political parties:
We further reiterate that based on our unique forecast model that accounts for poll survey results, historical data/status-quo political patterns, and collated endorsements by socio-cultural groups, ethno-nationalist groups, religious groups, and powerful actors in the Nigerian political space, the higher a political party’s predictive forecast value points are, the higher their chances of winning the forthcoming presidential election in the first round.
Kwakol Research forecasts that in the perfect scenario, the Labour Party has the best chance of winning the 2023 Nigerian presidential election in the first round. The All Progressives Congress is the party with the second highest chance of winning the presidential election in the first round. The People's Democratic Party is the party with the third highest chance of winning the presidential election in the first round, while the New Nigeria People’s Party has the fourth highest likelihood of emerging victorious in the forthcoming presidential election in the first round.
But elections in Nigeria do not always pose perfect scenarios, so we have accounted for a number of approximate scenarios with context-laden scenario-based analyses: run-off election round permutations, electoral violence and other forms of violence during the election, and prospects of a last-minute alliance between two of the four frontline parties.
Scenario 1 - Electoral violence: Terror violence, Sectionalist violence, and Voter Intimidation
We predict that if either of the six scenarios below of electoral violence plays out in any of the geopolitical zones, the probable political party to benefit is the APC, most likely due to a repeat of the 2019 voting patterns that were favorable to the party in hindsight.
North Central Nigeria: In the event of violent attacks in the geopolitical zone, voter turnout will decline significantly. However, we expect to see a repeat of the 2019 election patterns in the region, where both the APC and the PDP share similar levels in terms of votes obtained. Violence in the geopolitical zone will affect the votes obtainable for the LP and the NNPP.
North-East Nigeria: If an outburst of violence by non-state actors occurs in the northwest, this will significantly reduce voter turnout, and the only parties to benefit from traditional voting patterns will be the PDP and the APC, whose strongholds are in the geopolitical zone.
North West Nigeria: If an outburst of violence occurs in the northwest, all the frontline political parties will be adversely affected, but the APC will benefit from the 2019 voting patterns despite a significantly lower voter turnout.
South-East Nigeria: If there is an outburst of violence prior to or on the day of the presidential election in the zone, this outburst will most likely be perpetrated by IPOB through its violent paramilitary arm, ESN. These outbursts of sectionalist violence will significantly dampen voter turnout, and while all the frontline parties might likely be adversely affected, LP will rue this outcome the more because their probable strong dominance in the region would have been dampened.
South-South Nigeria: We do not expect to see an outburst of violence from the militants in the south-south geopolitical zone, but we suspect that hired political thuggery could be a recurring theme during the election, particularly in Rivers and Delta states. We think that this will likely be due to specific interests for the minimum 25% of votes cast in the two respective states. We also suspect that pro-LP strongholds in the region might be targeted in order to heighten voter anxiety, therefore dampening voter turnout. The LP, more specifically, will rue this probable outcome.
South-West Nigeria: If electoral violence breaks out in Nigeria's southwest geopolitical region, it will most likely take the form of political thuggery and herdsmen invasion. Political thuggery, particularly in Lagos, might escalate to levels higher than what surfaced during the 2019 presidential election (especially in traditional pro-PDP areas). Other parts of the southwest particularly the parts neighboring the northcentral could re-live attacks from armed herdsmen. The PDP, LP, and NNPP will be adversely affected, while the APC will likely benefit.
APC v LP
In the eventuality that the presidency has to be decided by a run-off presidential election round between the two parties with the greatest number of votes cast in their respective favor, and if these two parties happen to be the APC and LP, we predict a very close race with the APC as the eventual winner. The APC’s incumbency, reliable voting patterns in its strongholds in the northwest and southwest, and the further polarization of the electorate along time-honored ethnoreligious lines in the midst of reeling voter fatigue will be of advantage to the APC.
PDP v LP
We predict that while the PDP will be adversely affected by the expected low voter turnout in the Northwest, and while the Southwest will back the LP instead of the PDP, it will not be enough for the LP. The PDP will benefit from Nigeria’s time-honored voting patterns along ethnoreligious and sectionalist lines, where the proportion of registered voters across the north will be of utmost importance.
APC v PDP
We predict that the PDP will benefit from a section of the electorate that is disgruntled with the APC’s same-religion candidate pairing, especially in the south-south and southeast. The PDP will also benefit from a section of the LP strongholds in the south-south and the southeast. The APC will maintain its strongholds in the northwest and the southwest albeit with a reeling voter fatigue. This leaves the key battleground geopolitical zones in the context of this run-off pairing to be the northeast and the north-central zones. Thus, we predict that the PDP will just about edge out the APC in this very tight race.
APC + NNPP
If there is a last-minute alliance between the APC and the NNPP with the NNPP dropping out of the race, we predict that the APC will be able to gain more support from the NNPP stronghold in the north and northeast, and thus, it is likely that the APC will win the election in the first round.
PDP + NNPP
If there is a last-minute alliance between the PDP and the NNPP with the NNPP dropping out of the race, we predict that the PDP will be able to gain more support from the NNPP strongholds in the northwest. However, we cannot conclude if this is enough for the PDP to win the presidential election in the first round.
In our subsequent articles, we will highlight the distinctive peculiarities of our polling results from an array of perspectives: polled respondents’ perception of governance, polled respondents’ perception of INEC, run-off analysis, and new electioneering trends (social media).