Hi,

Categorical confounding variable:

http://www.cdc.gov/nchs/tutorials/dietary/Advanced/EstimatePrevalence/Info3.htm

http://sphweb.bumc.bu.edu/otlt/MPH-Modules/BS/BS704_Multivariable/BS704_Multivariable7.html

Ordinal dependent variables in Regression Analysis:

http://www.norusis.com/pdf/ASPC_v13.pdf

http://www.upa.pdx.edu/IOA/newsom/da2/ho_ordinal.pdf

Thanks Senthilvel. I am wondering if I am overcomplicating matters and just need to adjust the results somehow

I don't think you are overcomplicating things. A regression method is *the* tool to adjust for covariables.

I would suggest to increase the "complicateness" in your case even further by pointing out that your response is not a metric ("intervall scaled") variable. It might be inappropriate to treate "ratings" as a metric variable, especially when there are only relatively few different values and when it is hard to see that the difference betwen, say, a rating of 1 and 2 means somehow quantitatively the same like the difference in a rating of 4 and 5. The correct model would be an "ordered logistic regression", but again with all the covariables that should be considered.

I think it is better applying Logistic regression analysis .