I'm trying to conduct a simple study: to find out if there is a significant relationship between the time perspective of a person and his propensity to buy lottery.

So to keep it simple, my dependent variable would be binary (the person buys lottery or does not buy lottery) and the independent variable would have two factors: present-oriented and future-oriented.

Now, I think I have two hypotheses: 1) Present-oriented people buy more lottery than those not present-oriented. 2) Future-oriented people buy less lottery than those not future-oriented.

I've thought about doing two Chi-square tests for each hypothesis, but I'm wondering if there are some better alternatives. For instance, I would also be interested in considering the age of the interviewees and gender. Should I do a Logistic regression perhaps? or is it a good idea to do several Chi-tests?


  • 1
    $\begingroup$ Quick note: if present-oriented and future-oriented are dichotomous states ,such that everyone is in either one category or the other, then your two hypotheses are the same? 1. (present > future); 2. (future < present) $\endgroup$ May 22, 2015 at 5:25
  • $\begingroup$ How do you measure time perspective? I imagine that you have some kind of test. If so, then time perspective is non-binary because you have some kind of score for it and dichotomizing this kind of scores is usually a bad idea. $\endgroup$
    – Tim
    May 22, 2015 at 8:50
  • $\begingroup$ @JamesStanley I agree with you and have discussed this issue, but some others have said that people could be both. I still don't understand exactly how but they're more interested in the results in any case. Thanks for pointing that out. $\endgroup$
    – Egil137
    May 22, 2015 at 18:50
  • $\begingroup$ @Tim You're right, the result of this test is a score from 1-5. I'm mainly helping others design this study so I'll point that out. I imagine a logistic regression or some classification method would me more helpful then? $\endgroup$
    – Egil137
    May 22, 2015 at 18:56

1 Answer 1


If you are interested in testing the effect of time perspective on propensity to buy while controlling for potential confounders such as age or gender, then a Cochran-Mantel-Haenszel type test would be appropriate, where the analysis is stratified by these covariates.


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.