Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

Supported by

Meta-analysis for an interaction effect

Hello,

Is it possible to perform in JASP a mini meta-analysis (3-4 studies) for an interaction effect between a dichotomous and a continuous variable (tested via hierarchical regression)? The ES in this case is partial R-squared.

Thanks in advance,
Orly

Comments

  • I've asked around but will do some with more emphasis now. Sorry for the tardiness

  • Not sure what you mean exactly by 'hierarchical regression', but the interaction between an dichotomous and a continuous variable in a meta-regression involves 4 parameters that have to be estimated (intercept, coefficients for the dichotomous and continuous predictor, and a coefficient for the interaction between the latter two). If you'd do that on 4 studies, that results in a perfect fit and no degrees of freedom left to test any hypotheses, so it's just a transformation of the data. So the short answer is: No, you can't (most certainly not with only 3 studies); unless you just want to transform your data. (Meta-regression is just like normal regression, except for known error variance.)

Sign In or Register to comment.