Tell me more ×
Cross Validated is a question and answer site for statisticians, data analysts, data miners and data visualization experts. It's 100% free, no registration required.
up vote 1 down vote favorite

I have to test one predictor associate with 10 responses, the predictor is nominal, and response are continuous, so what kind of test can be used to test the association between them?

share|improve this question
2  
You need to edit your question to provide us with more details. A specific example with some data would help. Perhaps you could use linear regression of your response on the predictor. You could then exponentiate the resulting beta coefficient and look at the resulting relative risk plus or minus its 95% confidence intervals. You could build ten separate models (i.e., one for each response variable) using this technique. Keep in mind that you would need to correct for multiple testing. – Alexander May 13 '12 at 22:57
It's not clear what you're asking but if you're looking to characterize heterogeneity across groups in a the mean vector of a multivariate (10-dimensional in this case) response, this sounds like a problem for MANOVA. – Macro May 14 '12 at 3:41
yes, using MANOVA is an option. if I have one response, we can do linear regression to see the association of the predictor with response by see the beta coefficient, right? the more significant the coefficient, the stronger association it has, however, what if the response is 10 dimensional and I assume each dimension is independent? So do 10 linear regression for each response? – user974270 May 14 '12 at 10:56
"the more significant the coefficient, the stronger association it has" - this is a common fallacy. Large sample sizes can make very small effects very significant. The larger the effect size, the stronger the association is. Also, if you have a single response you can do linear regression (which is equivalent to ANOVA). 10 regressions for each response is effectively doing ANOVA for each dimension separately, which may be bad if there are large dependencies, but under the assumption of independence, that sounds fine. – Macro May 14 '12 at 11:07
thanks for correcting me. So, I want to rank two predictors A and B regarding which one is more associate with 10 response. Then you say do 10 ANOVA separately, so I for each predictor, I get 10 F test value for A and B, how do I comare the association between A and B to see which one is more associated with 10 response? using the summed F test statistics? – user974270 May 14 '12 at 12:27
show 1 more comment

Know someone who can answer? Share a link to this question via email, Google+, Twitter, or Facebook.

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Browse other questions tagged or ask your own question.