theYinkaSanya

In 2023, I worked on a report that interviewed 4,400 respondents across Nigeria. When we published, a few people pointed out that the sample size was too small for an adult population of about 100 million at that time, which might seem like a fair concern.

EFINA had released a report earlier, with over 25,000 respondents. Their sample was far larger than ours (Intelpoint’s); the findings were the same.

I want to say “no be statistics we go chop,” but this is it: two reports with very different sample sizes can say exactly the same thing.

If you calculate a sample size for Nigeria’s entire adult population using the right confidence level, margin of error, and population variability, among other variables, you land at fewer than 500 people. Using SurveyMonkey’s Sample Size Calculator or any other calculator, you get the same figure.

So if Intelpoint had interviewed fewer than 500, we’d have been statistically right. Both EFINA’s and Intelpoint’s samples were well above that.

But why do they both say the same thing: “the law of diminishing returns”, and it’s practically the same interpretation as the economic meaning of that phrase.

A larger sample gives you a better representation of the population you’re studying. But there’s a point where adding more respondents stops changing the findings in any meaningful way. At that point, it’s practically a case of expending more effort and resources for almost nothing extra.

This then begs the question whether it’s worth the extra effort.

Consider a conversation that comes up on X every time a report about Nigerians’ earnings is released. Aside from the 2023 report, I’ve worked on several reports where remuneration was one of the insights, and they all point to the same thing: only a small fraction of Nigerians earn at least a million naira monthly.

Reports from PiggyVest, PaidHR, and even the Lagos State Government, among others, say the same. Their sample sizes range from a few thousand to tens of thousands, and the conclusion doesn’t move.

Yes, people who earn a few million naira a month definitely know others who do too. That doesn’t mean a significant share of the population sits in the same bracket.

Determinants

The type of research also determines the sample size. Qualitative research uses a much smaller sample size than quantitative, even when both are examining the same sector.

Is the research exploratory, descriptive, or causal? The response to this also plays a role.

For quantitative, population composition is also a key factor. A study examining both Nigeria and Kenya using the same methodology can use a larger sample size for Kenya despite a lower population size. If the methodology has a target of 300 respondents from each geographical area: Nigeria, with 36 states and the Federal Capital Territory, will get a sample size of 11,100, while Kenya, with 47 counties, gets 14,100.

Aside from the geographical area, the homogeneity or heterogeneity of the population is also a key factor. Others include budget and the goal of the study, etc.