# Rao-Scott chi-square correction

#### a_zensun

##### New Member
I have a multiple-response variable, and believe I need to use a Rao-Scott correction to the chi-square test. I am using Excel and only have the tools/information to do a first-order correction.

I want to make sure I am doing it correctly...

chi-squared (corrected) = chi-squared/correction factor

correction factor = 1-(total count of multiple responses/(total number of subjects*number of multiple response variables))

Is the number of multiple response variables only the responses used in a subject's multiple response, or is it all of the variables in the question whether or not a subject used that variable as part of a multiple-response. (i.e. possible answers are work, play, eat, sleep. one subject responds with "work, play" another subject responds with "work, eat", and another subject responds with "sleep". Is the "number of multiple response variables" "4" because it is the total number of possible responses, or is it "3" because only 3 of the 4 responses were used as part of a multiple-response?).

Any help would be greatly appreciated.

Thanks!