5 intervention groups and ordinal outcomes

elisabethpaige — Tue, 31 May 2016 06:37:35 GMT

Hi...I am trying to write a grant for a clinical trial with 5 intervention groups and the following outcome:

positive behavior change

no change

negative behavior change.

what test would I use? Would 60 participants likely be enough?

Also, we are going to have a continuous scale with the same interventions

What test would I use.

Thank you very much!

I am going to be using the newest version of SAS.

Re: 5 intervention groups and ordinal outcomes

SteveDenham — Tue, 31 May 2016 13:12:03 GMT

If you are writing a grant for a clinical trial, you should have access to a biostatistician somewhere at your institution. These are the kinds of questions they are paid to answer.

Sample size questions depend on the analytical method and on what you consider to be the detectable effect size. For categorical variables, 60 participants = 12 per group. this would be adequate to detect a difference that would look like 10% in one group and 70% in another (two-tailed test, alpha=0.05, power=80%). If you wish to detect a smaller effect size, you would likely need many more participants per group.

Good luck, and sit down with a biostatistician as soon as you can.

Steve Denham

topic Re: 5 intervention groups and ordinal outcomes in Statistical Procedures

5 intervention groups and ordinal outcomes

Re: 5 intervention groups and ordinal outcomes