Which Test?

Draft: This page has not been reviewed by a statistician yet. Use it as a starting point and check important choices with a statistician.

Chi-square test

Compares the proportion of patients with a yes/no outcome between independent groups.

Clinical example

Whether the proportion of patients with a postoperative wound infection differs between two types of dressing.

When to use it

  • The outcome is yes or no, one per patient.
  • You compare two or more independent groups.
  • You don't need to adjust for other variables, and about 5 or more patients are expected in every cell of the table.

Assumptions to check

Independent observations

Each patient appears in the table once.

Expected counts of about 5 or more

Look at the expected counts in every cell of the group-by-outcome table. If some are below 5, use Fisher's exact test.

What to report

  • Number and percentage with the outcome in each group
  • Risk difference (or risk ratio) with 95% CI
  • The exact p-value, not just "p < 0.05"

Common mistakes

Report the effect, not just the p-value

Give the estimate (difference, ratio) with its 95% confidence interval and the exact p-value.

An odds ratio is not a risk ratio

When the outcome is common (more than about 10% of patients), the odds ratio is further from 1 than the risk ratio. Log-binomial regression gives a risk or prevalence ratio instead. Don't describe an odds ratio of 2 as "twice the risk".

Alternatives

If your situation is…Use
Fewer than 5 expected in some cellFisher's exact test
You need to adjust for other variablesLogistic regression

In your software

R
# R
tab <- table(trial$dressing, trial$infection)
chisq.test(tab)
SPSS
CROSSTABS
  /TABLES=dressing BY infection
  /STATISTICS=CHISQ RISK
  /CELLS=COUNT EXPECTED ROW.

Sources