In my dissertation , i have come across two situation
1) In my 3*2 table , some of cell value are less than 5 , so the question is can chi-square with Yate correction be applied?
2) In my 2*2 table , if the cell value is less than 5 then can we apply Fischer Exact test ?
I had asked Q2. to earlier on research gate , many scholars had suggested it can be used provided zero is not present in completely in rows or columns.
I require an answer which can justify during viva voice
Kindly if any attachment available so that i can justify in front of the examiner
Thankyou
Justification to your Q-2 could be explored here- http://udel.edu/~mcdonald/statfishers.html
Answer to Q-1: Am sure that if your expected cell frequencies are below 10, you probably should be using the Yates correction, but if less than 5 am not sure. You can explore it here- http://www.statisticshowto.com/what-is-the-yates-correction/
All D Best
Yates' correction prevents overestimation of statistical significance for small data.It can be used for cells below 5. The problem is that it may overcorrect. I would have a close look at the distribution and describe it. On the basis of that description I would interprete the results. In a 2X2 table I read the column header as a condition. The distribution in a column is the transformation of the vertical marginal distribution under the condition indicated in the header. Look at column percentages which make the reading of different columns comparable.
Try Fischer's exact test, which is specifically designed for when you have less than 5 cases in some cells. i am not enough of an expert to advise you extensively, but spss has a whole description on which test to use under which conditions.
3*2 table if > 20% the expected cell value are less than 5 -yates correction to be used.
for 2*2 table you can use fischer exact test
- Badges
- Science topic
- Similar topics
- Analytical Chemistry
- Column