Contingency tables: what if it doesn't go to plan?
Hey guys,
I want to use a contingency table in JASP to check if two categorical factors in two of my experiments are connected: group and relationship status. Relationship status is not fixed - I didn't know beforehand how many people in total or in each group would be in a relationship and how many would be single.
When it comes to group, I have two situations:
In experiment A, I preregistered and aimed for 27 people in each of my two groups. However, Covid happened and we had to stop prematurely after 50 participants in total. Would you consider group or grand total fixed in this scenario? My gut would tell me to go with independent multinomial sampling here but I'm not entirely sure.
In experiment B, we had a time limit for testing and tried to test as many people as possible. However, during testing we aimed to keep the group sizes somewhat even but were not totally strict about this. We actually ended up with 34 people in each of the two groups but this is due to people having to be excluded due to unrelated reasons. So here again, I'm wondering: Would you consider group fixed and therefore suggest independent multinomial sampling? Or would you consider this to be a case for Poisson sampling?
Cheers,
Max
Comments
Hi Max
I'd consider group fixed in both scenarios.
Cheers,
E.J.