Every feature from all five editions for 15 days.
Standard edition from US$ 155 a year · 30-day money-back guarantee.
Excel’s built-in CHISQ.TEST returns a p-value for a table whose expected frequencies you have already worked out, and stops. No mosaic plot to see where the association lies. No risk difference or odds ratio with a score confidence interval. No distinction between independent and related tables, and no McNemar test for before-and-after designs. When the data are categorical — pass/fail, treated/untreated, exposed/unexposed — you need tests designed for proportions, not means, and knowing that two variables are associated is rarely enough.
Where is the association strongest? How large is the effect — as an absolute difference, a relative risk or an odds ratio? Is the confidence interval precise enough to act on? And when the same subjects are measured twice, which test applies?
See the table and the conditional proportions before any test. The contingency table shows the counts for every combination of the two variables, with the marginal totals. Grouped and stacked frequency plots show the proportion of each category within each group, so a difference between the groups is visible before it is tested.
Pearson χ² and likelihood ratio G² tests tell you whether two categorical variables are associated. The mosaic plot shows where. Each tile’s area represents the proportion, and colouring by Pearson residual highlights the cells that depart most from what independence would predict.
A significant χ² test does not tell you how large the effect is. For a 2×2 table of independent observations, Fisher’s exact test tests for independence and the score Z test for a difference between the two proportions. Proportion difference (risk difference), proportion ratio (risk ratio) and odds ratio each answer a different question — absolute difference, relative risk or odds. Each has score-based or exact confidence intervals that perform well even with small samples or proportions near 0 or 1.
When the same subjects are measured twice, the observations are paired and the standard χ² test does not apply. Before and after treatment, or two raters classifying the same cases, are the usual designs. The McNemar-Mosteller exact test for marginal homogeneity handles the dependent structure correctly. So do the proportion difference with a Tango score CI and the odds ratio with an exact or Wilson score CI.
See contingency table results in detail — independence tests, mosaic plots, proportion tests and effect sizes with CIs.
Contingency table analysis is one part of a complete statistical analysis toolkit. The Standard edition also includes ANOVA and ANCOVA, simple and multiple regression, logistic regression, PCA and factor analysis, descriptive statistics, hypothesis testing and correlation. See everything in the Standard edition →
Learn how the study design determines the procedure: chi-square, Fisher exact or McNemar? Then see Cohen’s kappa and weighted kappa when the question is agreement rather than association.
Try it on your own data first. The 15-day trial is every feature from all five editions — install it and start straight away.
Standard edition: US$ 155 per year or US$ 395 for a perpetual licence. Every purchase carries a 30-day money-back guarantee. Need a quote for purchasing? Add the licence to the cart and save it as a PDF quote.
