Every method, every formula source

Equality of variances assuming normal data.

Requires

Numeric outcome, grouping column, normal residuals.

Reports

Chi-square, df, p.

Source

Bartlett, M. S. (1937). Proc. R. Soc. A, 160, 268–282.

Model a binary outcome from one or more predictors.

Requires

Binary outcome and one or more numeric predictors.

Reports

Coefficients, SE, z, p, odds ratios with 95% CI, McFadden and Nagelkerke R², AIC, likelihood-ratio test, accuracy.

Source

Hosmer, D. W., Lemeshow, S. & Sturdivant, R. X. (2013). Applied Logistic Regression (3rd ed.). Wiley.

Compare observed category counts against expected proportions.

Requires

One categorical column.

Reports

Chi-square, df, p, expected counts.

Source

Pearson, K. (1900). Phil. Mag., 50, 157–175.

Test association between two categorical variables.

Requires

Two categorical columns; expected counts ≥ 5 preferred.

Reports

Chi-square, df, p, expected counts, Cramér's V, phi.

Source

Pearson, K. (1900). Phil. Mag., 50, 157–175.

Compare three or more matched binary measurements.

Requires

Three or more paired binary columns.

Reports

Q, df, p.

Source

Cochran, W. G. (1950). Biometrika, 37, 256–266.

Agreement between two raters on categorical judgements.

Requires

Two categorical columns with the same levels.

Reports

Kappa with 95% CI, observed and expected agreement.

Source

Cohen, J. (1960). Educational and Psychological Measurement, 20, 37–46.

All pairwise correlations across numeric variables with multiplicity control.

Requires

Three or more numeric columns.

Reports

Matrix of r, per-pair p, corrected p, heat map.

Source

Cohen, J. (1988). Statistical Power Analysis for the Behavioral Sciences (2nd ed.). Erlbaum.

Internal consistency of a set of scale items.

Requires

Three or more numeric item columns measured on the same scale.

Reports

Alpha, number of items, total variance.

Source

Cronbach, L. J. (1951). Psychometrika, 16, 297–334.

Summarise central tendency, dispersion and shape for every numeric variable.

Requires

At least one numeric column.

Reports

n, missing, mean, 95% CI, median, mode, SD, SE, variance, min/max, IQR, percentiles, skewness, kurtosis, outlier count.

Source

Tukey, J. W. (1977). Exploratory Data Analysis. Addison-Wesley.

Standardised magnitude of an observed difference or association.

Requires

Reported automatically alongside each inferential test.

Reports

Cohen's d, Hedges' g, dz, eta², omega², epsilon², Cramér's V, phi, rank-biserial r, odds ratio.

Source

Lakens, D. (2013). Frontiers in Psychology, 4, 863.

Exact association test for 2×2 tables with small counts.

Requires

Two binary columns.

Reports

Exact p, odds ratio with 95% CI, risk ratio.

Source

Fisher, R. A. (1922). J. R. Statist. Soc., 85, 87–94.

Count and proportion of each level for categorical variables.

Requires

At least one categorical column.

Reports

Counts, percentages, cumulative percentages, mode.

Source

Agresti, A. (2013). Categorical Data Analysis (3rd ed.). Wiley.

Compare means of two independent groups (pooled variance).

Requires

Numeric outcome, two-level grouping column, homogeneous variances.

Reports

t, df, p, mean difference, 95% CI, Hedges' g.

Source

Student (1908). Biometrika, 6, 1–25.

Normality test based on skewness and kurtosis; suited to larger samples.

Requires

Numeric column, n ≥ 30 recommended.

Reports

JB statistic, df, p.

Source

Jarque, C. M. & Bera, A. K. (1980). Economics Letters, 6, 255–259.

Rank concordance, preferred for small samples with many ties.

Requires

Two ordinal or numeric columns.

Reports

tau-b, z, p.

Source

Kendall, M. G. (1938). A new measure of rank correlation. Biometrika, 30, 81–93.

Non-parametric comparison of three or more independent groups.

Requires

Numeric/ordinal outcome, grouping column with 3+ levels.

Reports

H (tie-corrected), df, p, epsilon², post-hoc pairwise Mann–Whitney with correction.

Source

Kruskal, W. H. & Wallis, W. A. (1952). JASA, 47, 583–621.

Test equality of variances across groups (Brown–Forsythe variant).

Requires

Numeric outcome and a grouping column with 2+ levels.

Reports

W, df1, df2, p.

Source

Brown, M. B. & Forsythe, A. B. (1974). JASA, 69, 364–367.

Model a continuous outcome from one or more predictors.

Requires

Numeric outcome and one or more numeric predictors.

Reports

Coefficients with SE, t, p, 95% CI, R², adjusted R², F, RMSE, VIF, Durbin–Watson, residual plots.

Source

Draper, N. R. & Smith, H. (1998). Applied Regression Analysis (3rd ed.). Wiley.

Compare distributions of two independent groups without normality.

Requires

Numeric or ordinal outcome, two-level grouping column.

Reports

U, z (tie-corrected), p, mean ranks, rank-biserial r.

Source

Mann, H. B. & Whitney, D. R. (1947). Ann. Math. Statist., 18, 50–60.

Change in paired binary measurements.

Requires

Two paired binary columns.

Reports

Continuity-corrected chi-square, df, p, discordant pairs.

Source

McNemar, Q. (1947). Psychometrika, 12, 153–157.

Control family-wise error or false discovery rate across a run.

Requires

Two or more p-values in the same analysis family.

Reports

Bonferroni, Holm–Bonferroni, and Benjamini–Hochberg adjusted p-values.

Source

Benjamini, Y. & Hochberg, Y. (1995). J. R. Statist. Soc. B, 57, 289–300.

Compare a sample mean with a reference value.

Requires

One numeric column.

Reports

t, df, p, mean difference, 95% CI, Cohen's d.

Source

Student (1908). The probable error of a mean. Biometrika, 6, 1–25.

Compare means across three or more independent groups.

Requires

Numeric outcome, grouping column with 3+ levels, homogeneous variances.

Reports

F, df, p, sums of squares, eta², omega², post-hoc pairwise tests.

Source

Fisher, R. A. (1925). Statistical Methods for Research Workers. Oliver & Boyd.

Compare two repeated measurements on the same units.

Requires

Two numeric columns of equal length, paired row-wise.

Reports

t, df, p, mean difference, 95% CI, Cohen's dz.

Source

Student (1908). Biometrika, 6, 1–25.

Association between two variables controlling for a third.

Requires

Three numeric columns.

Reports

Partial r, t, df, p.

Source

Fisher, R. A. (1924). The distribution of the partial correlation coefficient. Metron, 3, 329–332.

Linear association between two continuous variables.

Requires

Two numeric columns, approximately linear relationship.

Reports

r, 95% Fisher-z CI, t, df, p, r².

Source

Pearson, K. (1896). Mathematical contributions to the theory of evolution. Phil. Trans. R. Soc. A, 187, 253–318.

Test whether a sample plausibly comes from a normal distribution.

Requires

Numeric column, 3 ≤ n ≤ 5000.

Reports

W, p, Q–Q plot.

Source

Royston, P. (1995). Remark AS R94. Applied Statistics, 44, 547–551.

Distribution-free test of median difference for paired data.

Requires

Two paired numeric columns.

Reports

Positive/negative counts, exact binomial p.

Source

Dixon, W. J. & Mood, A. M. (1946). JASA, 41, 557–566.

Monotonic association robust to outliers and non-normality.

Requires

Two numeric or ordinal columns.

Reports

rho, t, df, p.

Source

Spearman, C. (1904). The proof and measurement of association between two things. Am. J. Psychol., 15, 72–101.

Robust alternative to one-way ANOVA when variances differ.

Requires

Numeric outcome, grouping column with 3+ levels.

Reports

F*, df1, df2, p.

Source

Welch, B. L. (1951). Biometrika, 38, 330–336.

Two-group mean comparison without assuming equal variances.

Requires

Numeric outcome and a two-level grouping column.

Reports

t, Welch df, p, mean difference, 95% CI, Hedges' g.

Source

Welch, B. L. (1947). Biometrika, 34, 28–35.

Non-parametric comparison of two paired measurements.

Requires

Two paired numeric columns.

Reports

W, z, p, rank-biserial r.

Source

Wilcoxon, F. (1945). Biometrics Bulletin, 1, 80–83.