Sample Size and Expected Frequencies.
In Chi-square tests, there is a standard rule of thumb called the “ 5 cell rule” which means in at least 80% of the categories being compared, each group must have an expected count of 5 or more. If either the sample size is small or the counts are less than 5, then a Fisher’s exact test is used.
The expected cell frequency should not be less than one in any category and to ensure this, we need to aim at a sample size that’s at least 5 times the number of categories we are looking at.
Ref:
McHugh ML. The chi-square test of independence. Biochem Med (Zagreb). 2013;23(2):143-149. doi:10.11613/bm.2013.018
Categorical data
- Chi-square tests are designed for categorical, not continuous, data.
Chi square test studies data that are counted and sorted to categories and not continuous data. For example, to check if attending a class effects on how students perform on a test, we would not use test scores like 70/100, instead we would group them as “pass” or “fail”. The Chi-square will then compare the 2 groups mentioned below
Pass
Fail
Attended
25
6
Skipped
8
15
Ref:
Author: Ambereen
Comments
Post a Comment