EPIC welcomes opportunities to collaborate with program managers to design and test innovative ideas. We have found the following conditions are favorable for application of the “test, learn, adapt” strategy:

Exposure Control

Program managers can control exposure to the program innovation.


Testing innovations requires being able to clearly separate facilities, businesses, communities, or individuals into treatment (innovation) and control (status quo) groups.

Sample Size

Tens or hundreds of facilities, businesses, or people can be exposed to the innovation.


Larger samples offer sufficient statistical power to detect an innovation’s impact, should one exist.

Observable Outcomes

Behavioral or environmental outcomes can be observed in a timely manner.


Indicators of final outcomes, or important intermediate outcomes, allow us to measure policy-relevant impacts. To lower the cost of data collection, we often use already-available administrative data.

Elucidate How and Why

Opportunities exist to go beyond the question, “Does it work?”


The best project designs do more than simply answer, “Does the innovation have an impact on average?” They can also answer questions like, “In which circumstances does it work best?” and “Through what mechanisms does it work?” The answers help identify where to best target an innovation and which program levers will maximize impacts.

Catalyst for Broader Innovation

The targeted behavior change is common to many programs (e.g., compliance; self-reporting; technology adoption), the proposed innovation is increasingly popular or widely debated, or the proposed test can shed light on competing theories about what drives environmentally-relevant human behaviors.


The evidence we develop can have broad applications to analogous contexts and programs.