NigeriaPolls · Data & Analysis
How Accurate Were the 2023 Election Polls? A Data-Driven Retrospective
Which Nigerian polling firms got the 2023 election right? We analyzed predictions vs. actual results from NOI Polls, SBM Intelligence, Afrobarometer, and others.
Context
The 2023 Election Context
Nigeria's 2023 presidential election was one of the most consequential and unpredictable in the country's democratic history. For the first time since 1999, no incumbent president was on the ballot. Three major candidates competed in what became a genuine three-horse race:Bola Ahmed Tinubu of the All Progressives Congress (APC), Atiku Abubakar of the Peoples Democratic Party (PDP), and Peter Obi of the Labour Party (LP), whose candidacy ignited an unprecedented youth-driven movement known as the "Obidient" wave.
The election was held against a backdrop of economic distress — record inflation, a currency redesign crisis that caused cash shortages nationwide, and widespread insecurity including banditry in the North-West, separatist tensions in the South-East, and lingering Boko Haram activity in the North-East. The Central Bank's naira redesign policy, announced just weeks before the election, disrupted campaign logistics and may have affected turnout patterns.
INEC introduced the Bimodal Voter Accreditation System (BVAS) and the IReV result viewing portal, promising greater transparency. Yet technical failures on Election Day delayed result transmission, fueling disputes and legal challenges that would drag on for months. The final certified results gave Tinubu 8,794,726 votes (36.61%), Atiku 6,984,520 votes (29.07%), and Obi 6,101,533 votes (25.40%).
For pollsters, 2023 was a stress test. Could they capture a genuinely fluid electorate? Could they model youth turnout accurately when many young voters were first-time registrants? Could they anticipate the geographic realignment triggered by Obi's candidacy? This retrospective examines which firms succeeded, which struggled, and what the industry learned.
The Predictions
The Major Polls: Who Predicted What
In the months and weeks before the February 25, 2023 election, several Nigerian and international polling organizations published horserace polls. The table below summarizes the final pre-election predictions from the most prominent firms.
| Polling Organization | Fieldwork Dates | Sample Size | Method | Tinubu (APC) | Atiku (PDP) | Obi (LP) | Winner Predicted |
|---|---|---|---|---|---|---|---|
| Dataphyte | Jan–Feb 2023 | 2,400 | F2F + CATI | 29% | 23% | 25% | Tinubu |
| SBM Intelligence | Jan 2023 | Not disclosed | Mixed | ~30% | ~25% | ~22% | Tinubu (narrow) |
| NOI Polls / Gatefield | Dec 2022 | 2,000 | F2F | 21% | 23% | 13% | Atiku |
| Afrobarometer | 2022 | 1,600 | F2F | — | — | — | No horserace |
| We2Geda / Stears | Jan 2023 | 6,000+ | Online panel | 15% | 18% | 46% | Obi |
| Nextier / ADC | Jan 2023 | 3,000 | F2F | 23% | 20% | 27% | Obi |
| Anap Foundation | Feb 2023 | Not disclosed | IVR | 23% | 23% | 21% | Too close to call |
Note: Percentages may not sum to 100% due to rounding, undecided voters, and other candidates. Afrobarometer did not publish a horserace poll in 2023 but released issue and approval data. SBM Intelligence figures are approximate based on their published analysis. Some online-panel polls significantly overstated Obi's support due to demographic skew.
Reality
Actual Results vs. Predictions
Official INEC Results (Presidential)
Total votes cast: ~24.9 million
Turnout: ~27% of registered voters
Prediction Error Analysis
Rankings
Accuracy Ranking of Polling Firms
We scored each firm on four criteria: (1) correct winner prediction, (2) average absolute error across the top three candidates, (3) methodological transparency, and (4) timeliness (fieldwork close to Election Day). The composite score produces the following ranking:
1. Dataphyte
Correctly predicted Tinubu victory with the smallest average error. Mixed-mode design (F2F + CATI) provided geographic coverage and urban-rural balance. Published full methodology.
2. SBM Intelligence
Provided nuanced analysis that anticipated the three-way split and correctly identified Tinubu as the likely winner. Their security-focused research gave them unique insights into turnout risks. Did not publish full horserace numbers, limiting comparability.
3. Anap Foundation
IVR methodology produced high error margins, but the firm responsibly labeled the race "too close to call" rather than making a definitive prediction. Their caution was warranted.
4. NOI Polls / Gatefield
Face-to-face methodology is generally credible, but the December fieldwork was too early to capture the late Obi surge. Predicted Atiku to win, which proved incorrect.
5. Nextier / ADC
Predicted Obi to win, which was directionally wrong. The error may have stemmed from sampling in urban centers where Obi enthusiasm was visibly high but not representative nationally.
6. We2Geda / Stears
Online panel methodology severely skewed toward young, urban, educated respondents — exactly the Obidient base. The result was a 20+ point overestimate of Obi's support and a corresponding collapse in Tinubu and Atiku numbers.
Success Patterns
What the Accurate Polls Had in Common
The firms that performed best in 2023 shared several methodological and strategic characteristics:
1. Mixed-Mode or Face-to-Face Data Collection
The two most accurate firms — Dataphyte and SBM Intelligence — both relied on methods that reached beyond the digital elite. Face-to-face interviews in rural areas and smaller towns captured the perspectives of voters who were not active on Twitter or TikTok, where the Obidient movement was most visible. Online-only polls, by contrast, systematically excluded older, rural, and less-educated Nigerians, leading to massive overestimates of Labour Party support.
2. Geographic Stratification
Accurate polls ensured representation from all six geopolitical zones, including the North-West and North-East where APC maintained its base. Some less accurate polls appeared to under-sample the North or overweight Lagos and Abuja, missing the depth of Tinubu's support in the core North.
3. Turnout Modeling
The most accurate firms did not simply ask "Who would you vote for?" — they applied turnout probability models based on past voting behavior, PVC possession, and stated enthusiasm. Nigeria's historically low turnout (27% in 2023) means that raw voting intention is a poor predictor of electoral outcomes. Firms that treated every respondent as equally likely to vote produced inflated support for candidates with enthusiastic but less reliable bases.
4. Fieldwork Timing Close to Election Day
NOI Polls' December survey was simply too early. The final month before the election saw massive shifts: the cash scarcity crisis, last-minute defections, and intensified campaign activity. Firms that fielded in late January or early February captured more of the actual electorate's mindset.
5. Methodological Transparency
The best firms published their sampling frames, questionnaire wording, weighting procedures, and margin of error. This transparency allows external analysts to assess credibility and identify potential biases. Firms that released only headline numbers without methodology should be treated with skepticism regardless of whether their predictions were accurate.
Lessons
Why Some Polls Missed
The polls that significantly missed the mark — particularly those showing Obi winning — suffered from a combination of technical, methodological, and analytical failures:
The Online Panel Trap
Several polls relied on online panels or social media recruitment, which in Nigeria skew dramatically toward young, urban, educated, and politically engaged demographics — precisely the Obidient profile. These panels failed to reach the rural, older, and less digitally connected voters who ultimately delivered Tinubu's victory. Internet penetration in Nigeria is approximately 50%, and social media usage is even more concentrated.
Failure to Model Ethnic and Religious Voting Patterns
Nigerian voting behavior remains strongly correlated with ethnic and religious identity, even among voters who claim to prioritize competence or policy. Some polls treated the electorate as a homogenous national sample without adequately weighting for the enduring importance of regional, ethnic, and religious loyalties in Nigerian electoral behavior.
Underestimating APC's Ground Game
Tinubu's campaign invested heavily in grassroots mobilization, traditional ruler engagement, and party structure activation. These assets are hard to capture in standard polling but produce turnout on Election Day. Some pollsters may have overweighted visible enthusiasm (rallies, social media) at the expense of quiet but effective party machinery.
The "Shy Voter" Problem
Social desirability bias may have led some respondents to conceal their true voting intention, particularly if they supported a candidate perceived as controversial or establishment-aligned. This phenomenon — sometimes called the "shy Tory" effect — can systematically depress reported support for certain candidates.
Late-Breaking Events
The naira redesign crisis, fuel scarcity, and insecurity in the South-East all unfolded in the final weeks before the election. Polls fielded earlier missed these dynamics. Even polls fielded late may have struggled to capture how these crises reshaped voter calculus in real time.
Looking Ahead
Lessons for 2027
As Nigeria looks toward the 2027 general elections, the 2023 retrospective offers clear guidance for pollsters, journalists, and voters:
- Invest in mixed-mode designs. No single method can capture Nigeria's diversity. Face-to-face remains essential for rural coverage, while phone and online methods can supplement urban data.
- Weight by PVC possession and past turnout. Self-reported voting intention is unreliable. Turnout modeling must incorporate objective indicators of electoral participation.
- Field as close to Election Day as possible. Nigerian electorates are fluid. The final 2–3 weeks are critical.
- Be transparent about limitations. Pollsters should openly discuss what their methods can and cannot capture, rather than presenting point estimates as certainties.
- Avoid online-only panels for national horserace polling. The 2023 results should permanently discredit unrepresentative digital panels for predicting Nigerian election outcomes.
- Account for regional and ethnic dynamics. Cross-tabs by state, zone, ethnicity, and religion are not optional — they are essential for understanding how results will distribute geographically.
- Track the "Obidient" evolution. Whether Labour Party consolidates its 2023 gains or fragments will be one of the defining questions of 2027. Early tracking polls should monitor party identification shifts.
At NigeriaPolls, we are building on these lessons for our 2027 election forecasting program. Our methodology integrates mixed-mode data collection, turnout probability modeling, and real-time voter file updates to produce the most accurate 2027 election polling in Nigeria.
FAQ
Frequently Asked Questions
Which Nigerian pollster was most accurate in 2023?
Dataphyte and SBM Intelligence both produced notably accurate forecasts. Dataphyte correctly identified Bola Tinubu as the likely winner, while SBM Intelligence provided granular analysis that anticipated the three-way split and the role of youth turnout.
Why did some polls show Peter Obi winning?
Several polls captured Obi's remarkable momentum among young, urban, and educated voters — a genuine phenomenon. However, some polls overweighted these demographics or failed to model turnout accurately in rural and northern areas, leading to overestimates of his national support.
What is a good accuracy score for an election poll?
In competitive multi-candidate races, a poll that correctly predicts the winner and comes within 3-5 percentage points of the actual result is considered strong. Predicting the correct order of finish is the minimum standard for credibility.
Did the polls fail in 2023?
No — the polling was mixed, not uniformly failed. Several firms correctly identified Tinubu as the likely winner. The bigger challenge was predicting the precise margin and the geographic distribution of support, especially the Obidient surge and its limits.
Why was youth turnout lower than expected?
Despite massive online enthusiasm, actual youth turnout (18-35) in 2023 was lower than anticipated due to PVC collection barriers, transportation costs, security concerns in some areas, and the cash scarcity crisis that disrupted mobility in the final weeks.
Will 2027 polling be more accurate?
If Nigerian pollsters apply the lessons of 2023 — especially around mixed-mode design, turnout modeling, and demographic weighting — 2027 polling should improve. However, new dynamics (new candidates, shifting party structures) will create fresh challenges.
Keep Learning
Related Guides
Polling 101: How Public Opinion Research Works
A beginner-friendly introduction to how polls are conducted, from sampling to analysis.
IntermediateHow to Read a Poll Like a Data Journalist
Learn to spot methodological red flags and interpret poll numbers with confidence.
ReferenceThe Polling Glossary
75 essential polling and survey terms explained in plain English.
AdvancedMethodology Explained: How NigeriaPolls Works
Deep dive into our sampling, weighting, and quality control procedures.
StatisticsSample Size and Accuracy
How many people do you need to interview for a reliable poll? The math explained.
Data2027 Election Polls
Track the latest polling for the 2027 Nigerian general elections.
Stay Informed with NigeriaPolls
Get the latest polls, election forecasts, and analysis delivered to your inbox. No spam — just data-driven insights on Nigerian public opinion.