Begin by identifying the primary research question and the analysis that will answer it. A comparison of two groups, a multiple-regression model, a prevalence estimate, and a multilevel study do not require the same information or calculation.
Start with the inferential objective
For hypothesis-testing studies, an a priori power analysis usually requires a target significance level, desired power, expected effect size, and the structure of the planned model. Justify the expected effect through closely related literature, a pilot, or a substantively meaningful threshold.
Respect the design structure
Clustering, repeated measures, unequal allocation, multiple predictors, rare outcomes, and anticipated covariate adjustment can alter the required sample.
Document the decision
Report the analysis used, every numerical assumption, the software or formula, and any attrition allowance.
Cite this articleOpen ready-to-copy reference entriesAPA · MLA · Chicago · Harvard
Alova, C. A. R. (2026, August 17). Sample size is a design decision—not just a formula. A&A Statistical and Research Consultancy Center. https://aa-research-consultancy.ailaaniaalova.chatgpt.site/resources/sample-size
Alova, Chard Aye R. “Sample Size Is a Design Decision—Not Just a Formula.” A&A Statistical and Research Consultancy Center, 17 Aug. 2026, https://aa-research-consultancy.ailaaniaalova.chatgpt.site/resources/sample-size.
Alova, Chard Aye R. “Sample Size Is a Design Decision—Not Just a Formula.” A&A Statistical and Research Consultancy Center. August 17, 2026. https://aa-research-consultancy.ailaaniaalova.chatgpt.site/resources/sample-size.
Alova, C.A.R. (2026) ‘Sample Size Is a Design Decision—Not Just a Formula’, A&A Statistical and Research Consultancy Center, 17 August. Available at: https://aa-research-consultancy.ailaaniaalova.chatgpt.site/resources/sample-size (Accessed: 26 September 2026).
Academic reference styles normally omit professional credentials and job titles; these remain displayed in the article byline.