Statistics Calculators

19 tools, all running locally in your browser. Nothing is uploaded.　中文版 →

The hard part here is never the computation — it is the choice.

If you do not know which test applies, start with the statistical test chooser: walk through data type, number of groups and pairing, and it tells you where you land and why.

Three mistakes dominate, each with a tool that addresses it:

1. Not correcting for multiple comparisons. Run 20 tests and one comes back p < 0.05 by luck alone. FDR / Benjamini-Hochberg handles this; in omics data it is not optional.
2. Using a t-test on data that is not normal. With small samples the normality test itself has little power, which is exactly when non-parametric tests are the safer choice.
3. Reading correlation as effect. Correlation coefficients quantify co-variation — Pearson captures linear, Spearman monotone — and R² = 0.9 implies nothing about importance. High R² is especially cheap when n is small.

One thing belongs before the experiment: sample size estimation. Power computed after the fact tells you nothing; a sample size fixed in advance does.

The omics-facing tools are listed separately: log₂FC, enrichment testing, RNA-seq normalisation and ΔΔCt. The principles above still apply — only the context changes.

Statistics