Thank you Keenan great video. I was trying to figure out why I didn't have the nice data when I put >2 rows/columns in contingency data. To better assess where the statistical significance arrises from - Can I separately analyse 2 x 2 using fishers etc; using different combinations of different groups i'm analysing
Hello! The contingency format is super rough looking in terms of visuals sadly. I break my "no pie graphs rule" when the data is only 2x2, making two, but then I'll perform the Fisher's separately. If you deviate from 2x2 format of data, we typically don't use Fisher's although this is mainly because it's tough by hand. Since that's less of a problem with computers, you can, but I've seen Chi-Square applied more often when it goes beyond 2x2. You can run as many separate Fisher's as you need if you have the data, but there may be some kind of multiple-hypothesis correction to apply separately in that case. Hope this helps!
Thank you Keenan great video. I was trying to figure out why I didn't have the nice data when I put >2 rows/columns in contingency data. To better assess where the statistical significance arrises from - Can I separately analyse 2 x 2 using fishers etc; using different combinations of different groups i'm analysing
Hello! The contingency format is super rough looking in terms of visuals sadly. I break my "no pie graphs rule" when the data is only 2x2, making two, but then I'll perform the Fisher's separately. If you deviate from 2x2 format of data, we typically don't use Fisher's although this is mainly because it's tough by hand. Since that's less of a problem with computers, you can, but I've seen Chi-Square applied more often when it goes beyond 2x2. You can run as many separate Fisher's as you need if you have the data, but there may be some kind of multiple-hypothesis correction to apply separately in that case. Hope this helps!