PowerMate alpha

Power analysis, without the fog.

You do not need to be a statistician to begin. You do need a clear question, a planned analysis, and honest ranges for uncertain assumptions.

Start with the planner Or try simulation

3-step starter run
  1. Open the starter planner (Welch path preloaded).
  2. Keep defaults once, click Calculate, and read the assumptions list.
  3. Change one assumption (effect or SD), recalculate, and compare N.

What power means

Power is the long-run chance that your planned test detects a particular effect if that effect is really present under the assumptions you entered. It is not the probability that your hypothesis is true.

Why N changes

Smaller effects, noisier outcomes, stricter alpha levels, unequal allocation, clustering, attrition, and multiple primary tests usually require more observations.

Which tool should I use?

Guided planner

Best when your primary analysis exactly matches one of the verified common designs. Example: two independent groups compared with Welch’s t test.

Use the planner

Simulation lab

Best when the study structure itself matters. Example: students nested in classrooms, where students in one classroom resemble one another.

Use simulation

Coverage map

Use this before forcing a complex study into a familiar calculator. It tells you what is supported, qualified, or waiting for a method-specific simulation.

Check support

A worked beginner example

  1. Question: Do two training programmes produce different average scores?
  2. Outcome: A continuous test score.
  3. Groups: Different people in each programme, so the groups are independent.
  4. Variability: Prior pilot results suggest SDs near 10 and 15, so equal variance is doubtful.
  5. Path: Choose “Welch unequal-variance t test” in the planner, enter the smallest important score difference, both SD estimates, power, alpha, and allocation.
  6. Interpretation: Treat the returned N as conditional on those assumptions; inspect the sensitivity chart before committing.
Three numbers people often confuseTarget effect is what you want the study to be able to detect. Observed effect is what a completed study estimates. Minimum detectable effect is the smallest effect a fixed design can detect at the chosen power. PowerMate keeps these concepts separate.