Checking the adequacy of the sample size when the data set is qualitative

#1
I am trying the check whether my simple size is adequate. I have activity patterns for a set of people (An activity pattern is something like (1,2,9,3,4)). These activity patterns are qualitative data and due to that reason I cannot calculate a Standard Deviation using below equaltion.

(Z-score)2 x SD x (1-SD)/ME2 = Sample Size

I know there are some thumb rules to get the sample size but I am wandering whether there is a more mathematical way to get it. I would be grateful if anyone can suggest me a method. I checked FAQs , a text book, other threads, and googled it but could not find a proper answer.

Thank you for your attention.
 

spunky

Doesn't actually exist
#2
Sample size determination is related to the power of a statistical test. You need to decide what test to run first to decide on the sample size.

For instance, even if it's not explicitly stated, the formula you present above is for proportions (the SD x (1-SD) gives it away).
 

noetsi

Fortran must die
#3
I am not sure what you mean by qualitative data. I assume dummy variables in a quantitative model like regression. In that case you can stop with what spunky said.

If you mean the type of qualitative data that involves coding like content analysis in that case statistical power does not apply. There are informal rules when the data is adequate or not, largely judgmental in nature.
 

spunky

Doesn't actually exist
#4
I am trying the check whether my simple size is adequate. I have activity patterns for a set of people (An activity pattern is something like (1,2,9,3,4)). These activity patterns are qualitative data and due to that reason I cannot calculate a Standard Deviation using below equaltion.

(Z-score)2 x SD x (1-SD)/ME2 = Sample Size

I know there are some thumb rules to get the sample size but I am wandering whether there is a more mathematical way to get it. I would be grateful if anyone can suggest me a method. I checked FAQs , a text book, other threads, and googled it but could not find a proper answer.

Thank you for your attention.
I am not sure what you mean by qualitative data. I assume dummy variables in a quantitative model like regression. In that case you can stop with what spunky said.

If you mean the type of qualitative data that involves coding like content analysis in that case statistical power does not apply. There are informal rules when the data is adequate or not, largely judgmental in nature.
I bet by "qualitative" data OP means categories. Like the famous Smoking VS non-Smoking and Cancer VS no-Cancer, 2x2 contingency table.
 

noetsi

Fortran must die
#5
I bet by "qualitative" data OP means categories. Like the famous Smoking VS non-Smoking and Cancer VS no-Cancer, 2x2 contingency table.
Me to. But I actually do content analysis so I wondered. Another case where words mean different things in different professions. Dummy variables really are not qualitative data. That would be text analysis.