Hello all!
I have a 2x2 contingency table, in which one of the 4 cells has a zero (0) frequency.
Is it acceptable to compare these data with Fisher Exact Test?
Can I use the statistical significant p-values (lower than 0.05) even if the odds ratios and confidence intervals cannot be computed?
In my research, zero value means that a specific allele has not been detected in one of my two groups.
Is there any valid source that I can use to document this functionality?
I would appreciate any help on this issue :)
Yes, you can use Fisher's exact test, assuming that the zero isn't a "structural zero".
For odds ratio, it's common to use the Haldane-Anscombe correction, which is simply to add 0.5 to all cells.
So for a 2 x 2 table with the values: 12, 15, 8, 0, the odds ratio can be calculated as
(12.5 / 15.5) / (8.5 / 0.5)
= 0.047
Some software packages will implement this correction by default when there are cells with low frequencies.
Thank you for your answer Sal Mangiafico :) I will use it as you recommended!
Certainly the observation of no alleles for that cell is important information.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS