Consumer research categorical data analysis report

[#1]

This will undoubtedly be a silly one, but in the categorical report from Consumer Research, there are two warnings in place:  Base count warning 100 and base count minimum 30. I am trying to figure out what these actually represent, and how they differ from the usual chi-square 20% of cells with less than expected counts. These warning can be changed in the set preferences platform so am wondering if they are just guidelines that people may want to put in place that are specific to their application?

Thanks,

Greg

[#2]

Hi Greg,

I couldn't find anything in the documentation about this. It seems there is a gap. I would suggest that you contact JMP Technical Support (support@jmp.com) to see if they can help.

Regards,

Phil

[#3 · Accepted solution]

Hi @GregMcMahon,

Basically, these are just rules of thumb, and they are intended to help the user identify when sample sizes may be considered “small.” As noted in the Platform Preferences, both can be changed to whatever value the user wants.

I would say that a sample size of 100 is considered fairly reasonable in a "consumer research" related context but it is highly subject-matter dependent.

The sample size of 30 is often used as a boundary between what constitutes small and large samples, but ultimately defending that is up to the user and I would argue that it should be based on subject-matter knowledge (with some sound statistical thinking built into it for support).    Here's a discussion thread from Stats.StackExchange that discusses this: https://stats.stackexchange.com/questions/2541/what-references-should-be-cited-to-support-using-30-a....