REAL-WORLD CHALLENGES TO RANDOMIZATION AND THEIR SOLUTIONS

Kenya Heard, Elisabeth O’Toole, Rohit Naimpally, Lindsey Bressler. J-PAL North America, April 2017. pdf copy here

INTRODUCTION
Randomized evaluations, also called randomized controlled trials (RCTs), have received increasing attention from practitioners, policymakers, and researchers due to their high credibility in estimating the causal impacts of programs and policies. In a randomized evaluation, a random selection of individuals from a sample pool is offered a program or service, while the remainder of the pool does not receive an offer to participate in the program or service. Random assignment ensures that, with a large enough sample size, the two groups (treatment and control) are similar on average before the start of the program. Since members of the groups do not differ systematically at the outset of the experiment, any difference that subsequently arises between the groups can be attributed to the intervention rather than to other factors.

Researchers, practitioners, and policymakers face many real-world challenges while designing and implementing randomized evaluations. Fortunately, several of these challenges can be addressed by designing a randomized evaluation that accommodates existing programs and addresses implementation challenges.

Program design challenges: Certain features of a program may present challenges to using a randomized evaluation design. This document showcases four of these program features and demonstrates how to alter the design of an evaluation to accommodate them.
• Resources exist to extend the program to everyone in the study area
• Program has strict eligibility criteria
• Program is an entitlement
• Sample size is small

Implementation challenges: There are a few challenges that may threaten a randomized evaluation when a program or policy is being implemented. This document features two implementation challenges and demonstrates how to design a randomized evaluation that mitigates threats and eliminates difficulties in the implementation phase of an evaluation.
• It is difficult for service providers to adhere to random assignment due to logistical or political reasons
• The control group finds out about the treatment, benefits from the treatment, or is harmed by the treatment

 

TABLE OF CONTENTS
INTRODUCTION ……………………………………………………………………….. 3
TABLE OF CONTENTS……………………………………………………………………. 4
PROGRAM DESIGN CHALLENGES ……………………………………………………………. 5
Challenge #1: Resources exist to extend the program to everyone in the study area…………… 5
Challenge #2: Program has strict eligibility criteria …………………………………… 9
Challenge #3: Program is an entitlement…………………………………………………12
Challenge #4: Sample size is small …………………………………………………….16
IMPLEMENTATION CHALLENGES……………………………………………………………..20
Challenge #5: It is difficult for service providers to adhere to random assignment due to logistical
or political reasons …………………………………………………………………20
Challenge #6: Control group finds out about the treatment, benefits from the treatment,
or is harmed by the treatment………………………………………………………….23
SUMMARY TABLE ……………………………………………………………………….27
GLOSSARY ……………………………………………………………………………28
REFERENCES ………………………………………………………………………….29

/p>

Comments?

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: