NigeriaPolls
All articles
methodology

How NigeriaPolls Conducts Rigorous Public Opinion Research

How NigeriaPolls designs a poll: the phone sampling frame, who gets asked, weighting, quality control, and what every published result shows, so you can judge any number you see.

NigeriaPolls Research14 April 20268 min read
How NigeriaPolls Conducts Rigorous Public Opinion Research

Published: February 2026 | Updated: October 2026 | Reading time: 6 minutes

A poll is only as good as the way it was built. This guide explains how NigeriaPolls designs a poll, who gets asked, how answers are checked, and what every published result shows, so a journalist, a researcher or an ordinary reader can judge any number on this site, and any poll anywhere in Nigeria.


1. Why many Nigerian "polls" mislead

The social media poll. A page posts a question to its followers and a few thousand people tap an answer. That result gets reported as public opinion. Followers of one account are not Nigeria. They skew urban, online and already interested in the topic.

The open link survey. A link goes out by email and social media, and the people who answer are the people who chose to. The motivated and the angry take part. The busy and the disconnected do not. The result describes who had the time.

The unadjusted national number. A poll gathers a thousand answers, most of them from Lagos and Abuja, and publishes a "national" headline. Kano, Borno and Enugu are barely in the data, so the headline is really a Lagos headline.

NigeriaPolls is built to avoid all three, and to say plainly which kind of result each page shows.


2. Who gets asked: the sampling frame

Our national fieldwork starts from a random-digit-dial phone sampling frame of over 11,000 Nigerian mobile numbers, spread across all 36 states and the FCT. The numbers are generated across live mobile number ranges rather than collected from a list of people who volunteered, so every working mobile line in the frame has a chance of being called. A measured check on 1 October 2026 found 85.5% of sampled numbers live.

Random selection from a defined frame is what lets us put a margin of error on a result and defend it. A frame built from volunteers cannot do that.

Quotas and weights are set against the most recent official benchmarks: National Population Commission projections, INEC voter registration figures and National Bureau of Statistics household data. Those cover state, urban and rural split, gender and age group.


3. How people are reached

Phone interviewing is the core mode, because a phone call reaches people who are not online and does not need a smartphone or literacy in the survey language. Interviews are offered in English, Hausa, Yoruba and Igbo where the question needs it. Our field partner network handles in-country fieldwork logistics.

Votes cast on nigeriapolls.com are a different thing. They are open web votes from visitors who chose to take part, and they are shown as exactly that. We never present an open web vote as a national result.


4. Weighting: making the sample match the country

No sample mirrors the population on its own, so results are weighted. We use raking, an iterative method that adjusts the sample to known population targets on several variables at once.

VariableBenchmark sourceWhy it matters
StateNPC census projectionsPolitical and cultural differences are state-level
Urban or ruralNBS household surveysUrban bias is a leading cause of polling error in Nigeria
GenderCensus and voter registrationWomen are under-represented in phone surveys
Age groupCensus projectionsYounger people answer more often than older people

Weights are capped so that no single respondent can count for far more than another. When a group is too thin to weight responsibly, the methodology note says so.


5. Quality control

Each response passes automated checks before it enters any analysis:

CheckWhat it catches
Speed checkAnswers given faster than the question can be read
Straight-liningThe same answer repeated down a whole question block
Duplicate detectionThe same phone number or device submitting more than once
Pattern detectionBot-like alternating answer sequences
Location checkResponses that do not fit the claimed state

Flagged responses are reviewed and excluded before publication. Seed, test and demonstration rows never count toward any figure shown on the site.


6. Reading a margin of error

A margin of error describes sampling noise on a random sample. With a random sample of 1,000 adults the margin is about plus or minus 3 percentage points at 95% confidence. With 2,000 it is about 2.2 points. Sub-groups carry larger margins, so a state result from a national sample is far less precise than the national headline. Our margin of error guide walks through the arithmetic.


7. What every NigeriaPolls poll page shows

  • The exact question wording
  • Field dates, start and end
  • The response basis: who was asked and how
  • The number of verified participants
  • Whether the result is a probability sample or an open web vote
  • The margin of error, only where the design supports one

The publishing system enforces this. A poll cannot go public if the sample size on its page disagrees with the verified responses behind it, and the site build fails if that check fails. That rule exists so that a number on NigeriaPolls always traces to rows you could count.


8. Nigeria Pulse and polls are different tools

Nigeria Pulse tracks daily conversation across Nigerian platforms. It shows what people are discussing and what mood dominates. It is a conversation signal, not a poll, and we label it that way. Polls ask a defined group the same question and count the answers. We use each for what it does well and never present one as the other.


9. How to evaluate any poll in Nigeria

QuestionRed flagGreen flag
Who conducted it?Unknown organisationNamed firm with a published methodology
What was the sample size?Not statedStated, with margin of error where valid
How were people selected?"Online respondents", "social media poll"Random sample from a defined frame
When was it fielded?"Recently"Specific dates
Was it weighted?Not mentionedYes, with named benchmarks
Can I see the questions?NoPublished word for word

If a poll fails more than two of these, treat it as entertainment, not evidence.


Conclusion

Careful polling is slower and costs more than a social media vote. It is also the only route to a number you can defend. NigeriaPolls publishes the sampling frame, the question, the dates and the count so that readers can check the work for themselves.

For questions about method, write to [email protected]. The full approach is on the methodology page.

Related: Understanding Margin of Error | Choosing a Survey Sample Size | Our Methodology

Tags

#methodology#sampling#transparency#polling#statistics

Cite this article (CC BY 4.0)

NigeriaPolls Research. (14 April 2026). "How NigeriaPolls Conducts Rigorous Public Opinion Research." NigeriaPolls. CC BY 4.0. https://nigeriapolls.com/blog/how-we-conduct-rigorous-polls

Free to share, remix, and republish with attribution. See terms.