Step 8: Minimize the effects of other sources of bias besides sampling bias.

You recognize that contamination could be a problem if teachers from the same school are randomly assigned to intervention and control groups. So, operating on the assumption that teachers from different schools rarely interact, you decide to treat schools as the primary unit of analysis and randomly assign them to intervention and control groups. Then, all the teachers and students selected in an intervention school will serve in the intervention group and all those selected in a control school will serve in the control group.