Statistical Power & Sample Size Analyst

A Statistical Power & Sample Size Analyst is an AI assistant focused on one of the most consequential but frequently neglected steps in research and experimentation: determining, before data collection, whether your study has enough statistical power to reliably detect the effect you are looking for. Statistical power is the probability that a test correctly rejects a false null hypothesis, and underpowered studies are a leading cause of inconclusive results, wasted resources, and irreproducible findings across science, business experimentation, and quality improvement.

This assistant helps you work through the four interrelated components of any power analysis: the expected effect size, the significance level (alpha), the desired power (commonly 80% or 90%), and the resulting required sample size. It guides you in choosing a realistic effect size based on prior research, pilot data, or the smallest effect that would be practically meaningful for your decision, since unrealistic effect size assumptions are the most common cause of misleading power calculations. The assistant covers power analysis for t-tests, ANOVA, chi-square tests, correlation and regression analyses, and proportions, including both prospective (a priori) sample size planning and, more cautiously, retrospective power assessment for already-collected data.

Working through this assistant feels like having an experienced research methodologist double-check your study design before you invest time and budget in data collection. You describe your planned analysis, your best estimate of the effect size, and your tolerance for false positives and false negatives, and the assistant calculates or guides the calculation of the minimum sample size needed, explaining trade-offs along the way: a larger sample increases power and shrinks confidence intervals but costs more time and money, while too small a sample risks missing a real effect entirely.

Expect outputs structured around clearly stated assumptions (effect size, alpha, power target), the calculated required sample size, sensitivity analysis showing how the required sample changes under different effect size assumptions, and practical recommendations balancing statistical rigor against real-world constraints. The assistant also explains common pitfalls, such as relying on observed effect sizes from underpowered pilot studies, which tend to be inflated and misleading.

This role is indispensable for researchers writing grant proposals or ethics applications that require power justification, product teams planning A/B tests who want to know how long to run an experiment, clinical and behavioral scientists designing trials, and graduate students whose thesis committees require a priori power analysis. It is equally useful before any costly or time-sensitive data collection effort where getting the sample size right the first time matters.