Which statistical test do I need?
Answer guided questions, or upload your data and let the tool detect variable types and normality. You'll get a recommended test, plain-English reasoning, and step-by-step instructions, designed to be used alongside JASP, SPSS, R, or Python.
Answer the questions
Step by step · pick the mode you preferHow many participants do I need?
Answer three questions, get an a priori sample size and a methods sentence ready to paste into your write-up.
What kind of test are you running?
Pick the option that best matches your study.
How big do you expect the effect to be?
Effect size is how strong the difference or relationship is. Without prior research to go on, medium (0.5) is the standard recommended default.
Cohen's d = 0.50 · Medium effect is the default when no prior estimate is available (Cohen, 1988).
Confidence and power
These control how certain you want to be. Most journals expect alpha = 0.05 and power = 0.80. Leave these unless your supervisor says otherwise.
Participants
,
per group
Power
80%
achieved
Effect size
0.50
Cohen's d
Ready to copy into your methods section
,
Based on Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Lawrence Erlbaum Associates. For educational use, verify with your supervisor or G*Power.
Data science decisions made easier
Three small flowcharts for the common questions: which chart should I use, which model should I try first, which metric should I report. Each result gives you a Python snippet ready to paste, plus links to the docs so you can learn beyond the recommendation.