Binomial tests

[OskarBerg]

Hi all,
My colleague and I were arguing what statistical test is most appropriate to use in this simple situation:



We are in debate whether a Fisher exact test or multiple Chi-squared tests are more appropriate to see if Group 3 has significantly fewer married people than the other groups. What stata output would you use and how would you interpret it?
Let us know!

 

[Karabiner]

Binomial tests are for 1 sample. They compare a proportion from within a sample with a
specified value.
Here, you want to analyse the association between 2 variables ("group" and "marriage").
You can consider a global Chi² test for the 2x3 table. If it shows a statistically significant
result, you can make pairwise comparisons (2x2 subtables).

With kind regards

Karabiner

 

[Dason]

And just to check - you have access to the original counts and not just the proportions right?

 

[OskarBerg]

And just to check - you have access to the original counts and not just the proportions right?

Yes, we have the original counts. All groups have about the same sample size, married is a dummy variable