Predicting answers to yes/no questions based on stimulant use


I don't like signing up to forums just to ask questions, makes me feel like a bad person. Usually I just lurk about till I find an answer but it ain't worked this time. Oh well, I'll try to contribute in the future if I can.

Anyway the title might sound exciting but it's really not. I have responses to two modfied Percieved Functions Survey (by Annabel Boys if you're interested). The surveys focus on reasons why people use drugs, one includes alcohol and tobacco, one excludes them. The PFS simply gives statements like have you ever used X for the purposes of making friends/relaxing/playing sport etc.

I also have answers to a simple yes/no 'have you ever used stimulants for the purposes of work' question.

Now I'm testing to find out if previous stimulant use predicts more 'yes' answers in the PFS that excludes alcohol and tobacco. My results are already coded into SPSS and using cross tabs and frequencies it looks like I'm on the right track so I've run a non-parametric, 2-independant samples, Mann-Whitney test with stimulants as the factor. All looks well and it looks like it's significant.


I'm not too sure if it is. Should I have used a different test? In the test statistics it shows significance but how do I know if its significant for yes(as in the have used stims) or no(they haven't used stims).

Honestly I've been awake for a long time so I might just be going mad but a wee bit of advice would be great, or reassurance.

I think my retinas are burning.
Thanks, that was speedy.
I'll give it a go, out of interest, was I right about the mann-whitney test, it shows significance but not the direction of it?

If I ever meet you I'll buy you a cookie.


Cookie Scientist
The Mann-Whitney test is not really appropriate here as it is not designed to handle dichotomous dependent variables.