- Home
- Calculators
- Margin of error calculator
Margin of error calculator
How much precision a given base size buys, what it assumes, and why the subgroup number is the one to check.
A margin of error calculator shows how much precision a given base size buys. Enter the number of completes, the proportion you observed and a confidence level, and it returns the expected range around that result. The formula assumes simple random sampling. Opt-in panel samples are not random samples, so treat the figure as an indication of precision rather than a guarantee.
Inputs
Margin of error
±5.7
percentage points at 95% confidence, n = 300
- Confidence interval
- 44.3% to 55.7%
- Finite population correction
- not applied
- z score
- 1.960
A 50% result with n = 300 could be anywhere from 44% to 56% in the population. Differences smaller than the margin between two groups are not differences.
What this assumes. The formula is for a simple random sample drawn from the population. Opt-in panel research is not a random sample, so quote the figure as an indication of precision and say what it assumes, rather than presenting it as a guaranteed error bound.
How the margin of error calculator works
The last term is the finite population correction and is only applied when you enter a population size. For a large population it is effectively 1 and drops out.
Reading it in a B2B context
The number that matters is rarely the total. An IT decision maker study of 300 completes with a 40% enterprise subgroup has 120 enterprise completes and a margin of about ±9 points on that cut. Run the calculator on the subgroup size before promising a client that enterprise and mid market differ.
Margin of error assumes the sample is representative of the population you care about. It says nothing about whether respondents are who they claim to be. That is a verification question, and no amount of n fixes it.
Margin of error by base size at 95% confidence
| Completes | Margin of error |
|---|---|
| 50 | ±13.9 points |
| 100 | ±9.8 |
| 150 | ±8.0 |
| 200 | ±6.9 |
| 300 | ±5.7 |
| 500 | ±4.4 |
| 1,000 | ±3.1 |
Margin of error calculator: frequently asked questions
What is a good margin of error for a B2B survey?
Five points is the common target for headline numbers. Seven to ten points is normal and acceptable for subgroups, as long as it is reported. What matters is that differences you present as findings are larger than the margin on the groups being compared.
Is n=100 enough?
For a directional read on a single question, yes, at about ±10 points. For comparing two groups of 50 each, no, because each has a margin near ±14 points and almost nothing will separate.
Why does the calculator use 50% by default?
Because it gives the widest margin, so it is the conservative figure. Enter the observed proportion for a tighter, question specific margin.
Does margin of error account for bad respondents?
No. It assumes every respondent belongs in the sample. Respondents who do not hold the role they claim bias the estimate rather than widening it, and that is invisible in the margin. Verify first, then size.
Send the spec. Get real numbers back.
Audience, market, target n, expected interview length. Feasibility the same business day for standard audiences, with reachable counts, an estimated qualification rate we believe and a quoted cost per complete that stays fixed for the agreed brief.