2
$\begingroup$

I have two groups of patients undergoing different treatments (active versus placebo), followed up at 3 different time points (time 1, time 2, time 3). The outcomes are categorical (improvement vs no improvement) so I know I should not use repeated-measures ANOVA. What is the appropriate test to use? I use SPSS for analysis.

$\endgroup$

2 Answers 2

2
$\begingroup$

A generalized linear mixed model is appropriate, binomial distribution and logit link. Your response is essentially binary (1 or 0) improved or did not. I would use a fixed effects of treatment (placebo or not) , time period and their interaction. Random effect of patient. This can be used to assess the probability of improvement or not, accounting for repeat measures on a patient.

I use r and sas, but below is a helpful link.

link for how to do glmm's in spss

https://www.ibm.com/support/knowledgecenter/en/SSLVMB_24.0.0/components/glmm/glmm_cablesurvey_howto.html

you want to select binary logistic instead of multinomial

Goodluck!

$\endgroup$
-2
$\begingroup$

We can use chi-square tes. We use this test to see if there is any relationship between two categorical variables. In SPSS chi-square tes is called chisq

$\endgroup$
1
  • $\begingroup$ Perhaps I couldn't make my point clear in the previous question. I used mixed model ANOVA for comparison of continuous variables at three time points. Now I want to make comparison of proportions improved between the two groups of treatment at time 2 and time 3. Should I make this comparison separately at time 2 and time 3 using chi-square test? $\endgroup$
    – amjad
    Nov 3, 2016 at 18:14

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.