Sample Size Calculation for Cross-Sectional Studies

When this approach applies

A common cross-sectional objective is to estimate the proportion of a defined population with a characteristic—for example, the prevalence of anaemia among a particular group. When an estimate of a single proportion is the primary objective, a frequently used initial formula is:

n₀ = Z² × p × (1 − p) / d²

Here Z is the chosen normal critical value (approximately 1.96 for 95% confidence), p is the anticipated population proportion expressed between 0 and 1, and d is the absolute precision in proportion units. Specify whether the precision is absolute or relative; confusing the two can produce a substantially incorrect sample size.

Worked example

Suppose previous literature suggests prevalence p = 0.30 and the study seeks 95% confidence with absolute precision d = 0.05. Then:

n₀ = 1.96² × 0.30 × 0.70 / 0.05² = 322.69, rounded up to 323 participants.

If an expected 10% of selected participants will not yield usable data, divide by 1 − 0.10 rather than simply adding 10%: 323 / 0.90 = 358.89, rounded up to 359 participants. These numbers are illustrative, not a recommended sample size for every cross-sectional study.

Adjustments and limitations

• For cluster sampling, consider an appropriate design effect based on the sampling strategy.
• A finite population correction may be relevant when sampling a substantial fraction of a known, finite population.
• For comparative or association objectives, a prevalence-only calculation may not yield sufficient power.
• The anticipated prevalence should come from comparable published evidence, not an arbitrary guess.
• The final protocol should document recruitment feasibility, sampling method and analysis plan.

What to write in your synopsis

Report the primary outcome, formula, source of expected prevalence, confidence level, absolute precision, any design effect, non-response allowance, and the final rounded sample size. Have the calculation reviewed for the particular design before ethics submission.

For transparent, design-specific calculations, use the existing ResearchTry Sample Size Calculator.

Frequently asked questions

What if no earlier prevalence estimate exists?

For a single proportion under simple random sampling, p = 0.50 maximizes p(1 − p); justify the choice and all other assumptions.

Does this formula work for comparing two groups?

Not by itself. Comparing proportions, means, or associations requires a design-specific sample size method and often a power assumption.

Further reading and primary guidance

• CDC Epi Info archived StatCalc guide: population survey sample size
• EQUATOR Network: STROBE observational study reporting guideline