Power Analysis & Sample Size Specialist

One of the most consequential decisions in any study is determining how many participants, observations, or experimental units are needed. Too few, and your study is underpowered — unable to detect a real effect even if it exists. Too many, and you waste resources, time, and in clinical contexts, expose unnecessary participants to experimental conditions. This AI assistant specializes in statistical power analysis and sample size determination across virtually every study design.

The assistant begins by asking you about your research design, the primary statistical test you plan to use, and your assumptions about the expected effect size. It then walks you through the core parameters — alpha (Type I error rate), beta (Type II error rate), desired power, and effect size — explaining the relationship between them and what each choice means for your study's credibility and feasibility.

For common designs, including independent and paired t-tests, chi-square tests, one-way and factorial ANOVA, simple and multiple regression, correlation tests, and survival analyses, the assistant provides step-by-step calculations and explains the formulas used. It also handles more complex scenarios such as multilevel models, repeated measures designs, and non-inferiority trials.

A critical feature is the assistant's emphasis on effect size estimation. Rather than defaulting to arbitrary conventions, it guides you toward evidence-based effect size choices drawn from pilot data, meta-analyses, or domain literature, and explains the consequences of optimistic versus conservative assumptions. This assistant is ideal for researchers preparing grant applications, writing ethics submissions, designing pilot studies, or reviewing the adequacy of published research.