Statistical calculators and
R output interpreters

Two means, paired or one-sample. Summary stats or raw values both work.

Turn a t, z, chi-square, F, or r statistic into its p-value, or go the other way from alpha.

Paste a column of numbers for the mean, median, SD, quartiles, skewness, and a live histogram and boxplot.

Paste a list or a frequency table for the mean, median and mode together, with the range, midrange and a read on when the median beats the mean.

Paste a list or a frequency table for the sample and population standard deviation, variance, standard error and CV, with every step shown.

Paste data for the full percentile table, 1st to 99th plus any custom percentile, or the percentile rank of a value, in type 7 or 6.

Paste data for the interquartile range, quartiles, five-number summary, the 1.5 IQR fences and every flagged outlier, on a labeled boxplot.

Paste one group or several named groups for the five-number summary, quartiles, IQR and outliers, drawn as side-by-side box plots.

Paste two columns for Pearson r, Spearman rho or Kendall tau-b with a CI, p-value and a live scatter plot.

Paste columns for the full Pearson or Spearman matrix, colour coded, with the p-value and the n behind every cell.

Paste x and y for the fitted line, the slope and intercept with SEs, t and p, R-squared, and a scatter with the fit.

Paste x and y, pick a new x, and see the prediction interval for one new observation beside the confidence interval for the mean.

Paste raw group data for the full ANOVA table (SS, df, F, p), effect sizes, a group-means plot, an assumption check and the R code.

Your ANOVA is significant, so which groups differ? Tukey HSD, Bonferroni pairwise t-tests and Dunn with BH, plus a letters display.

Enter a 2x2 table for the odds ratio, risk ratio, NNT and attributable risk, each with a confidence interval and the right p-value.

Enter a 2x2 table for the exact one- and two-sided p, the odds ratio with an exact CI, and an optional mid-p, with the null distribution drawn.

Conversion counts for two variants, with sample-size planning and a Bayesian read.

Is the difference real or noise? Two conversion rates, two averages, or a rate versus a target, with a plain-English verdict.

Counts in a table: independence, goodness-of-fit, homogeneity. Residuals show which cell drives the result.

Eight interval types for means, proportions, and variances, drawn to scale.

The plus-or-minus on a poll or survey, the sample size for a target margin, or the margin on an average.

Paste raw data and bootstrap a CI for the mean, median, SD, or any quantile.

Paste a column of p-values and see what survives Bonferroni, Holm, or the BH false-discovery rate.

Shows whether two treatments are close enough to call equivalent, not just “not significantly different”.

Convert a z-score to a percentile and back, with the shaded curve to match.

Area below, above, between or outside a value on any normal curve, plus inverse cutoffs, with a shaded plot.

The 68-95-99.7 rule for any mean and SD: the three bands and their ranges, which band a value is in, or the range for a coverage.

The chance of k successes in n trials, exact or cumulative, a range, or the inverse count, with a shaded bar chart.

The chance of k events at an average rate lambda, exact or cumulative, a range, or the inverse count, with a shaded bar chart.

A one- or two-proportion z-test the classroom way: hypotheses, the z statistic, a critical value, and the p-value, with the continuity-correction note.

Look up t critical values by df and alpha, or go backwards from t to p. Printable.

Standard normal areas: z to probability in any direction, and back. Printable.

Look up chi-square critical values by df and alpha, or go from a statistic to its p-value. Printable.

Look up F critical values by df1, df2 and alpha, or go from an F statistic to p. Printable.

Look up the chance of k successes in n trials: exact and cumulative cells for n = 1 to 20, with the p > 0.5 mirror handled for you. Printable.

How strong a correlation has to be, at your sample size, to count as real. Critical r by df and alpha, with a verdict and an exact p-value. Printable.

How strongly the data favor H1 over H0, with a plot showing how the answer moves as the prior changes.

Base rate in, posterior out. Explains why a 95%-accurate test can still be usually wrong.

The distribution for a proportion: density, tail areas, quantiles and credible intervals for any alpha and beta, with a shaded curve.

Whether you can stop the test: P(B beats A), the expected loss of shipping each variant in conversion points, and a credible interval for the lift.

Update a prior with data in closed form: Beta-Binomial, Normal-Normal or Gamma-Poisson. Prior, likelihood and posterior on one plot, the update arithmetic, and a credible interval.

Paste summary(lm(...)). Every coefficient, the R², and the F-test explained line by line.

Paste summary(glm(...)). Coefficients become odds or rate ratios; deviance and AIC get a plain reading.

Paste summary(lmer(...)). Variance components, the ICC with its formula, and where the p-values went.

Paste summary(coxph(...)). Hazard ratios in plain English, the concordance, and the assumption the printout never checks.

Paste summary() from brms or rstanarm. Credible intervals in plain English, Rhat and ESS decoded, and an honest verdict on whether the fit converged.

Paste aov() output. Recomputes each F from the sums of squares and adds effect sizes.

Reads the four plot(lm) panels for you: what a healthy one looks like and what yours is saying.

Checks whether your predictors are too correlated for the coefficients to mean anything.

Four cells in; precision, recall, F1, MCC and kappa out, with advice on which to trust when classes are imbalanced.

Answer plain questions about your goal and data; it walks the decision tree to the right test.

Shapiro-Wilk, Anderson-Darling and a Q-Q plot, plus a straight answer on whether normality even matters for your case.

Wilcoxon, Mann-Whitney, Kruskal-Wallis and Friedman, for when means and normality are off the table.

Draw your causal diagram; it finds the minimum set of variables to adjust for and flags colliders.

Start here. Routes you to the right calculator for what you are comparing, or sizes a survey estimate from your margin of error.

How many participants a one-sample, two-group, or paired t test needs, from Cohen’s d or from your means and SDs.

How many users an A/B test needs, or a one-proportion test against a target, from your rates or from Cohen’s h.

How many per group a one-way ANOVA needs, from Cohen’s f, from eta-squared, or from the group means you expect.

Solve for any one of sample size, power, effect size, or alpha across the common tests.

Events, sample size, or power for a log-rank test, with accrual, follow-up, and dropout built in.

Translate between Cohen's d, r, odds ratios, and eta-squared without hunting for the formula.

Drag effect size, n, and alpha; watch the two error regions trade against each other.

ADF, KPSS and Phillips-Perron on your series, and what to do when they disagree.

Both correlograms with the significance bands, and an honest read on the ARIMA order.

Grubbs, ESD, Hampel and IQR on the same data, side by side.

Paste scores and labels for the ROC curve, AUC with a DeLong CI, and the threshold that fits your costs.

Paste item responses for alpha, its confidence interval, and a per-item read on what to keep or drop.

Paste a confusion matrix or two rating columns for kappa, weighted kappa, a 95% CI and the agreement band.

Paste a rater table for all six intraclass correlation forms with CIs, and a picker for which one to report.

Pool a handful of studies under both the fixed-effect and random-effects models, with a forest plot and the heterogeneity numbers in plain English.

Turns your problem code into a minimal, runnable example people will actually answer.