1
$\begingroup$

For a rather large sample (>800 participants) I have answers from a questionnaire, containing multiple Yes/No-Questions. What I want to find out is, if the proportions of yes & no's differ between each question.

For that I thought about using multiple one-sample z-tests. Meaning, I would calculate the proportions of yes's for each question and then use 6 z-tests to test if the proportions for Question1 differ from Question2s proportions, if Q1 differs from Q3 .... and so forth. To account for multiple comparisons, I plan to use Bonferroni Correction.

Does my approach make sense, or can someone think of a better way of doing this?

$\endgroup$

1 Answer 1

1
$\begingroup$

A z-test requires the values to be numeric. Yours are not. So you need a different test to compare categorical variables. A test of proportions, where you would input the frequencies of successes (yes), in R you would achieve this with prop.test(). Also, Bonferroni correction is unecessary, Holm method is a better alternative (also called Bonferroni-Holm).

$\endgroup$
1
  • $\begingroup$ Thank you for your input. I think z-tests can be used for testing proportions of categories. But the prop.test() function seems to implement a non-parametric test, which I prefer. r-bloggers.com/… $\endgroup$
    – Bobipuegi
    Commented Oct 8, 2018 at 9:06

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.