chi squared

Shona

New Member
#1
I need to find the best type of analysis to compare whether people who say they would use a product are influenced by a characteristic. I have data that indicates yes or no for the buying part and yes or no for whether they think it has a certain characteristic. For each characteristic I have grouped in into pairs yy, yn, ny and nn and thought a chi square would do the trick by getting the result for each characteristic and comparing them - but there is a very high number of no-no for each characteristic and i'm not sure of the effect that would have. Any help on whether this matters or whether there is a better type of analysis to use would be gratefully received
Thanks
Shona
 

JohnM

TS Contributor
#2
Shona,

Yes, a chi-square test of independence "should" work here, but if one of the cells in the table has a VERY high frequency, it may not follow a chi-square distribution.

You can try Fisher's Exact Test for 2x2 tables. An interactive web page that calculates it is here.

John