0
$\begingroup$

I have a hypothesis such as "People consult internet before making a purchase decision" for a thesis research. In order to test it, I created a game where people had the option to consult "internet option" or not. I have a similar questions, but I have now realised that all my variables for this particular point are nominal; (the answers look 0: did not consult internet, 1: consulted internet)

However, I can't calculate a mean or do any of the tests like $t$-test, anova since it doesn't make sense.

I was wondering academically, my results would be enough to confirm (not cannot reject) that people consult internet if I have 80% of the same that consulted internet? Several different questions show similar results.

Improve this question
$\endgroup$

1 Answer 1

Reset to default
3
$\begingroup$

Your variables being all nominal is not a problem. If they are all Yes-No and coded 1-0 then they would be called binary or dichotomous. ("Nominal" is more social science jargon but does include other kinds of variable too.)

Far from the mean making no sense, means make complete sense here. If 80% say yes and 20% no the mean of 0s and 1s is 0.8, which is the proportion saying yes.

Chi-square tests and Fisher's exact test are some of the possibilities here.

Improve this answer
$\endgroup$
2
  • $\begingroup$ Thank you, What I meant about the mean is just that in market studies class (and from several other books) we are instructed that for qualitative variables (nominal and ordinal ones-even binary in my case) mean cannot be calculated- only mode and median are appropriate descriptives. But I will check out the Fischer's tests. $\endgroup$
    – user27113
    Commented Jun 20, 2013 at 15:17
  • $\begingroup$ That's nonsense for binary variables. $\endgroup$
    – Nick Cox
    Commented Jun 20, 2013 at 17:31

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.