Hi,

It is a cross-sectional study and no repeated measures. If it is a controlled trial then the effect of intervention requires at least one follow-up to note changes from baseline and to compare across groups.

The outcome measure is binary.

There are three arms (groups).

ANOVA is not applicable here. as the dependent variable is not on a continuous scale (interval).

Chi-square test would be applicable with the three arms in rows and Outcome in columns as Yes or No.

Hi Professor Anand,

Thanks very much for your response. Indeed, I should have taken into account that the dependent variable is binary.

I was also wondering whether the matching has any effect on the method that should be used for the sample size calculation?

Hi,

It is always advisable to take matched groups as there is homogeneity in baseline variables and the effect of any intervention can be understood in that scenario. Heterogeneity in baseline variable comparison leads to confusion about whether the changes are due to intervention effects or some other.

To calculate sample size in your case go for sample size for proportions. If you give the allocation ratio as 1: 1 for two proportions, the third group automatically gets the same number of subjects.

Try software G*Power which is free to download and comprehensive.