1
$\begingroup$

I have a question concerning the coding of Effect Sizes for a large (educational) Meta Analysis with mostly latent outcome variables.

Some studies provide "Intent to Treat" Data from which I calculate Effect Sizes (Hedges g) and some studies only provide (local average treatment effect) LATE or TOT (treatment on treated) Data. Some studies report both.

Is it okay to mix ITT and LATE Treatment-Effects-Data or should I stick to one type of treatment operationalization?

My meta analysis is concerned with educational interventions and since we seek to explain the variability in treatment success of these interventions afterwards through Intervention characteristics I usually would prefer to include the estimates of the subjects who really completed the intervention (rather than who were assigned to the treatment).

$\endgroup$
1
$\begingroup$

Including both should be fine but since you are planning to use meta-regression then it would seem a good idea to include another (binary) moderator: ITT versus LATE.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.