1
$\begingroup$

I am testing relationship between gender and perceived cause of suicide in a group. So I wanted to do a crosstab in SPSS Gender(Male/ Female) vs Family Neglect (Yes/ No). However, the number of males in higher than the females.

  1. Am I still able to proceed with Fisher Exact Test/Chi Square?
  2. If not, do I adjust the proportions or use other tests?
$\endgroup$
1
  • $\begingroup$ Yes, it's appropriate to run a chi-squared analysis, since it will determine whether gender is independent from (not associated with) family neglect. Fisher exact will provide a more robust outcome based on multiple realizations (all possible table configurations). With logistic regression, you could simultaneously generate odds of suicide and test for association. $\endgroup$
    – wjktrs
    Jan 7 at 6:32

1 Answer 1

3
$\begingroup$

The fact that there are more males than females does not preclude a chi-square test. This is actually pretty typical for chi-square. Whether you need an exact test depends on whether expected cell sizes are small, although people vary in how small they think it is.

But, why chi-square here? Wouldn't a regression be better? Depending on how the "perceived cause of suicide" is worded, you might want (regular) logistic or multinomial logistic.

And ... If you haven't collected data already, have you considered a "choose all that apply"? Suicide rarely has one and only one cause.

$\endgroup$
2
  • $\begingroup$ Thank you very much. I was toying around that idea as well but am just unsure how to run logistic regression analysis. =( $\endgroup$
    – Dev
    Jan 7 at 4:21
  • $\begingroup$ OK, it's good to know what you don't know! What to do next depends on the type of project this is and how important it is and whether you are going to be doing statistics in the future. Maybe do chi-square, maybe start learning logistic, or maybe hire an expert. $\endgroup$
    – Peter Flom
    Jan 7 at 11:26

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.