Test Purpose Assumptions When to Use

Tests whether the proportion of Binary data, independent

Small sample sizes for
Binomial Test successes in a binary sample observations, random
binary outcomes
differs from a known proportion sampling
Categorical data,
Chi-Square
Tests if observed categorical data independent Testing if data follows a
Goodness of Fit
fits a specific distribution observations, expected specified distribution
Test
frequency ≥ 5
Binary or categorical
One-Sample Runs Tests for randomness in a Checking the randomness
data, independent
Test sequence of data of a sequence
observations
Compares paired categorical data
Paired categorical data, Comparing paired
McNemar Test for changes (e.g., before and
dichotomous outcomes proportions in 2x2 tables
after)
Compares paired data to test Paired data, ordinal or When assumptions of the
Sign Test whether the median difference continuous, random paired t-test are violated
between pairs is zero sampling (non-normal data)
Paired data, ordinal or
Wilcoxon Signed- Compares paired data to assess Non-parametric alternative
continuous, random
Ranks Test differences in the distribution to the paired t-test
sampling
Permutation Test Tests whether there is a Non-parametric test for
Paired data, random
for Paired difference between two paired paired samples with no
sampling
Replicates samples assumption of distribution
Tests for association between Categorical data,
Small samples, low
Fisher’s Exact Test two categorical variables in independent
expected cell counts
small samples observations
Categorical data,
Large samples to test for
Chi-Square Test of Tests if two categorical variables independent
association between two
Independence are associated observations, expected
categorical variables
frequency ≥ 5
Independent samples, Non-parametric test
Tests if two or more independent
Median Test ordinal or continuous comparing medians of two
groups have the same median
data or more groups
Compares two independent Independent samples,
Mann-Whitney U Non-parametric alternative
samples to assess differences in ordinal or continuous
Test to the independent t-test
their distributions data, similar shapes
Compares the empirical
Kolmogorov- distribution of a sample to a Continuous data, Testing goodness of fit for
Smirnov Test reference distribution or independent distributions, or comparing
compares two sample observations two distributions
distributions
Non-parametric test for
Tests for differences in Binary responses,
Cochran’s Q Test comparing three or more
proportions across related groups matched data
matched groups
Tests for differences between Non-parametric alternative
Ordinal or continuous
Friedman Test groups in a repeated measures to repeated measures
data, related groups
design ANOVA
Breaks down significant chi- Categorical data, Understanding the nature
Partitioning after
square results to understand significant chi-square of significant chi-square
Chi-Square Test
specific differences result associations
Compares more than two Independent samples,
Kruskal-Wallis Non-parametric alternative
independent groups to assess ordinal or continuous
Test to one-way ANOVA
differences in their distributions data