Sample Size and Statistical Power Calculator (Excel)
This Excel workbook works out how many respondents or participants a study needs. It also shows how much statistical power, or precision, a sample you can afford will give you. It has these sheets: Survey - Proportion: Cochran’s formula, adjusted for population size, cluster sampling and non-response. It also shows the Yamane formula for comparison, a table of sample sizes by margin of error, and how to share the sample across strata such as districts. Survey - Mean: sample size for estimating an average, such as income or test scores. Two Means: comparing two groups, such as treatment and control, with a power curve chart. Two Proportions: comparing two percentages, with an optional continuity correction and a quick lookup table. Paired & One-Sample: before–after studies, or one mean against a target. ANOVA: comparing three or more groups. Regression & Correlation: a multiple regression model, or a single correlation. How to use: Pick the sheet that matches your main research question. The Instructions sheet has a guide. Enter your assumptions in the blue cells. Take expected values from a baseline, a pilot or published studies, and set α (usually 0.05) and power (usually 80%). Read the yellow FINAL SAMPLE SIZE or TOTAL TO RECRUIT cell. It already allows for non-response and dropouts. For cluster surveys, enter the cluster size and ICC; the workbook then adds the design effect for you. To test what a fixed budget allows, use “Power for a given sample size” or “Margin of error for a fixed sample”. Red warnings appear if an input is missing or the effect is too small to detect. The study-design sheets search up to 1,001 participants per group. In your methodology, report every assumption and where it came from.