I am working with small sample size qualitative clinical data (use attach file)
In this context I have used a exact Fisher test as some expected frequency < 5.
More exactly I perform a perform the Freeman-Halton extension of the Fisher exact probability test for a 2x3 contingency table, (http://vassarstats.net/fisher2x3.html)
I know that for the chi-square, when chi-square is statistically significant, when you wish to determine where the significance is coming from, you could practice a residual analysis.
(e.g. here with Excel : http://www.normalesup.org/~carpenti/Donnees-notes/Khi2-Contingence/Residus-chi2.xlsx)
I wonder if it is also possible (and how) to do a residual analysis after the exact Fisher statistics?
The residuals are the same in both tests. Fisher's test is only evaluating the possible multinominal distributions given the (fixed) marginal sums, whereas the Chi² test is evaluating the normalized squared differences between the expected and the observed counts, where the expected counts are derived from the marginal sums under the independence assumption.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Sampling Design
- Sample Size