Design #4: Randomized Pre-Test and Post-Test Design

An experimental design typically enables the evaluator to establish by randomization of individuals (or sites), two (or more) groups not significantly different at baseline for all (or almost all) independent or dependent predictors of impact. There are several methods to plan and conduct an experimental study. A common approach is to evaluate an existing program with the usual intervention (X1) provided to C group participants and comparing this group to an E group exposed to a hypothetically more effective program with additional intervention procedures (X1 + X2 + X3 + X4). Standardized baseline (O1) and follow-up assessments (O2 + O3) are conducted for all C and E group participants during specific observation periods. If a large number of participants (400-1,000) are needed in the evaluation, individuals can be randomly assigned (R) at each program site to the E or C group to control for inter-site selection biases.

Random assignment may be conducted daily if participants enter the program on the same day, or participants may sign up every week when they enter the program over a defined period of time. This design should produce excellent control over the three major biases to internal validity, assuming no major implementation problems. Confirmation that the randomization process has been established with equivalent groups at each site and overall is essential. Case studies 5, 6, 7 in this chapter (and case studies in Chapter 6) are examples of Design #4. Case study 7 is also an excellent example of a group randomized design in a low-income country: Nepal. The Nepal case study and many comparable examples demonstrate the complexity of a GRCT. It also confirms that this design can be successfully implemented not only in high-income, but also in low-income countries.