3
$\begingroup$

If BMI data is broken into two groups, those with low BMI and those with high BMI, what statistical test would I use to see if there is a gender difference (male and female) between the two categories of low and high BMI?

$\endgroup$
2

2 Answers 2

5
$\begingroup$

If all you have is the data that has already been binned for you, then the appropriate test is a Chi squared test, on the two-by-two cross tab. This will test against the null hypothesis of no relationship between the rows and columns (in this case, sex and BMI category - it doesn't matter which is row or column).

Any stats package, and even Excel, can calculate the Chi squared test for you.

If you have the actual BMI values then you should avoid doing this categorisation, and instead make use of the much greater information in the continuous variable. So this answer presumes that you don't, you just have "low" or "high", so you have no alternative.

$\endgroup$
2
$\begingroup$

If you have large enough samples, you can go with z-test for comparing two proportions (e.g. proportions for male in the two groups).

Alternatively you can use the $\chi^2$-test of independence for a $2 \times 2$ contingency table.

$\endgroup$

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.