There are differences in my intervention and control groups at the pretest. Which can I do?

you're welcome!

thanks a lot!

Suppose you are pretty sure the assignment of intervention and control groups is random, and these observed differences are not correlated to the assignment process. In that case, you may collect the data about the differences and adjust them by controlling the related variables in multiple regression analysis. However, if you doubt that the observed difference may cause selection bias in the assignment process, you can re-assign the cases with a rigorous random assignment. If the re-assignment is impossible, for example, the intervention has been implemented, you should collect the data about the differences and adjust the data by matching or balancing before the final estimatio