4.1.3 Impact Evaluation: Non-Randomized Group Designs

When it is not realistic to randomly assign program participants to either a treatment group or a control group there is an alternative. A quasi-experimental research design matches participants and non-participants. A comparison group can be formed by matching study participants, or groups of participants, on a set of measures of the program outcome(s) or measures associated with the program outcome(s). The idea is to have two groups with as many similar characteristics as possible.

Propensity scoring methods, or statistical assessments of participant similarities, are preferable when groups are matched with multiple pre-intervention measures. However, the type of matching algorithm used to implement the propensity scoring should be selected based on simulation studies, previous research that demonstrates the algorithm validity, and evaluation goals.

What elements would you like to review?
These elements are from the SEP Guidance Checklist

Click here to view the original SEP excerpt.