Sample Size Formula
For a population proportion at 95% confidence: n = (Z² × p × (1-p)) / e²
Where Z = 1.96 (for 95% confidence), p = estimated proportion (use 0.5 if unknown, most conservative), e = margin of error.
For most consumer research with an unknown population proportion, this produces the commonly cited figure of 384 respondents for ±5% margin of error at 95% confidence.
Sample Size Reference Table
| Margin of Error | Confidence Level | Required Sample Size |
|---|---|---|
| ±10% | 95% | 96 |
| ±5% | 95% | 384 |
| ±3% | 95% | 1,067 |
| ±5% | 99% | 664 |
Sample Size for Sub-Group Analysis
If you need to analyze results by sub-segment (e.g., by age group, region, or customer type), each sub-group needs its own adequate sample — typically a minimum of 100 respondents per sub-group for directional reliability, 200+ for robust statistical comparison between groups.
Qualitative Research Sample Sizes
Qualitative research (interviews, focus groups) does not use statistical sample size formulas. Instead, researchers aim for thematic saturation — the point at which additional interviews stop revealing new themes. This typically occurs between 8-15 in-depth interviews or 3-5 focus groups for a homogeneous population, more for highly diverse populations.
Frequently Asked Questions
Does sample size need to scale with population size?
Not proportionally. A sample of 384 provides the same statistical confidence whether the population is 50,000 or 50 million — this counterintuitive fact surprises many newcomers to research methodology.
What sample size do market research reports use?
Syndicated market research reports vary widely — primary research components typically survey 200-1,000 industry stakeholders or buyers, supplemented by secondary research and expert interviews.
