Chi‐squared and Fisher–Irwin tests of two‐by‐two tables with small sample recommendations

By I. Campbell. Statistics in Medicine, 2007

879 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/50a0149d079f47d51c9b2fa353ab47f36aa815d4

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Statistics and Probability

Citations in the corpus

The invisible addiction: Cell-phone activities and addiction among male and female college students (2014, methodology): “Therefore, we collapsed some categories in order to increase the cell frequencies following Campbell (2007) recommendations on the appropriate statistical test that mostly specify at least 5 as the minimum expected number.”
The Tale of Cochran's Rule: My Contingency Table has so Many Expected Values Smaller than 5, What Am I to Do? (2018, methodology): “Campbell (2007) discussed 2 × 2 tables in more detail and suggested that a modification of the Karl Pearson X2 statistics proposed by his son Egon Pearson (1947), that is, using (N − 1)/N ∗ X2, is more accurate in case of conditioning on only one of themargins, but this involves changing the test…”
The analysis of 2 × 2 contingency tables—yet again: Author's reply (2011, methodology): “Campbell’s adaption [3] is as follows: Use the ‘N −1’ chi-squared test unless the smallest expected count is less than 1, else, use Fisher–Irwin’s exact test.”
Arrhythmia risk stratification of patients after myocardial infarction using personalized heart models (2016, methodology): “The characteristics were compared between the two groups using two-tailed rank-sum test 60 or Fisher’s exact test 61 , as appropriate.”
Usability: Lessons Learned … and Yet to Be Learned (2014, neutral): “For example, there have been recent articles on new developments for computing binomial conﬁdence intervals (Agresti & Coull, 1998; Sauro & Lewis, 2005) and a better formula for chi-square tests (Campbell, 2007) when sample sizes are small (for details, see Sauro & Lewis, 2012).”