Desperate for help!



Hi Guys,

Completely new here, and a little overwhelmed trying to sort some results out for my MSc.

I have a few research questions to answer and I have no idea which statistical tests I need to use to find out the answers! :( I had a questionnaire put out and from the dataset I need to answer the following:

Primary outcome: 1) Is there are correlation between the religious groups and their likelihood of participating in a clinical trial?
Secondary outcomes: 1) Is there any correlation between ethnicity and religious affiliation?
2) Is one's perceived health score likely to be related to religion and/ or age?
3) Is there a relationship between level of education and religious affiliation?

I know this is a HUGE request, but I am really desperate now as my supervisor wants a copy of my draft thesis this week (he neglected to tell me he is on leave for several weeks leading up to my hand in date!).



use a logistic regression with the response variable as "did not participate (0)" or "did participate (1)". Your explanatory variables should include categorical data (religion, ethnicity) and continuous data (age, years of education). The coefficients for your variables will tell you how much more likely they were to participate (in percent). For example, if the coefficient on years of education is 0.1 then they were 10% more likely to participate for every year of age. In R you'd do something like
mymodel = glm(y~religion+ethnicity+age+education,family="binomial")
coef(mymodel) #use this to get the coefficients out

For question 3 I think maybe you'd use a chi square table. Not sure. Or possibly a multinomial regression with education as the predictor variable.