Hi,
I am new to statistical analysis and wanted to make sure my approach was sound. I have two questions. For my final project I am looking at several dichotomous variables. My two outcome variables, x and y, are interesting in their own right, but I also want to look at outcome of 'x and y,' and 'x not y,' so i created two new dichotomous variables, x_plus_y, and x_not_y.
To start, i set up several chi2 calculations to look for statistically significant associations between the different dichotomous predictor and outcome variables:
tab predictorvar outcomevar , row chi2
Q : is the tab command the right approach for the dichotomous comparisons, or do you suggest something additional?
For my predictor variable, it can be described as a dichotomous variable (presence or absence of heart attack) or a nominal categorial variable. I also want to look for associations by subtype using the categorical form of the variable. In Lab 5 we used tabodds outcomevar trendvar, where trendvar=caccat. The difference is, where caccat was an ordinal categorical variable, mine is nominal.
Q : Can I use my nominal categorical variable, in the same tabodds outcomevar trendvar command? if not, how can i look at my outcomevar of interest among the different subtypes of MI in my catvar? Is this a job for the logistic command, and if so, how would I set it up? Or do I have to do tab predictorvar outcomevar, row chi2 for each subtype separately as a dichotomous variable?
Thanks!