Statistics calculators
Free calculators for the tests and effect sizes most often reported in research papers. Each one shows the working, warns when an assumption is doubtful, and gives a sentence in APA style to paste into your results section. Every calculation runs in your browser and is checked against SciPy and statsmodels.
- t-test calculatorIndependent (Welch or Student), paired and one-sample t-tests from summary statistics or raw data, with Cohen’s d.
- Chi-square calculatorTest of independence for tables up to 6 × 6 and goodness of fit, with expected counts, residuals and Cramér’s V.
- Confidence interval calculatorIntervals for a mean, a proportion (Wilson, exact, Wald) and the difference between two means or proportions.
- Odds ratio calculatorOdds ratio, relative risk, risk difference and NNT from a 2 × 2 table, with confidence intervals and p-values.
- Relative risk calculatorRisk ratio with its confidence interval and p-value, relative risk reduction and the absolute risk difference.
- NNT calculatorNumber needed to treat or harm with Altman’s confidence interval, including intervals that pass through infinity.
- Fisher’s exact test calculatorExact two-sided and one-sided p-values for a 2 × 2 table, with the conditional odds ratio and its exact interval.
- Effect size calculatorCohen’s d, Hedges’ g and Glass’s Δ from means or a t value, and conversion between d, r and odds ratios.
- Cohen’s d calculatorCohen’s d from two means and SDs with its exact confidence interval, plus Hedges’ g and paired dz.
- Sensitivity and specificity calculatorSensitivity, specificity, predictive values and likelihood ratios of a diagnostic test, at any prevalence.
- Cohen’s kappa calculatorAgreement between two raters beyond chance, with weighted kappa for ordered categories.
Which test do I need?
| Your question | What you have | Use |
|---|---|---|
| Do two group means differ? | Means and SDs, or raw values | Independent t-test (Welch) |
| Did the same people change between two measurements? | Paired values or the mean and SD of the differences | Paired t-test |
| Are two categorical variables associated? | A table of counts | Chi-square test of independence |
| Same question, but with small counts in a 2 × 2 table | Four counts, an expected count below 5 | Fisher’s exact test |
| Do observed counts match expected proportions? | Counts per category | Chi-square goodness of fit |
| How precise is my mean or percentage? | Mean, SD and n, or a count out of n | Confidence interval |
| How strongly is an exposure linked to an outcome? | A 2 × 2 table (case-control or cohort) | Odds ratio |
| How much does a treatment change the risk? | Events and totals in two groups | Relative risk |
| How many patients must be treated for one to benefit? | Events and totals in two groups | Number needed to treat |
| How large is the difference between two groups? | Means and SDs, or a reported t value | Effect size (Cohen’s d, Hedges’ g) |
| How accurate is a diagnostic test? | Test results against a reference standard | Sensitivity, specificity, PPV and NPV |
| Do two raters agree beyond chance? | A table of both raters’ categories | Cohen’s kappa |
| How many participants does my study need? | Expected effect size, α and power | Power analysis |
A p-value alone says little about how big or important an effect is. Most journals now ask for an effect size with a confidence interval alongside each test, so the calculators report both.
Planning a study
- Power analysis calculator: Sample size for 80% power, or the power of a planned study.
- Sample size calculator: Survey sample size from a confidence level and margin of error.
- Forest plot maker: Pool study results in a meta-analysis and draw the forest plot.