The Department of Labor’s Behavioral Interventions (DOL-BI) portfolio explores how behavioral science can improve the performance and outcomes of DOL programs. As part of this work, CEO contracted with Mathematica Policy Research and ideas42 to design, implement, and rigorously test behavioral trials in select Labor programs. In addition to specific trials, the project team developed a Practitioner’s Playbook in 2017 that provides an overview for applying behavioral insights to labor programs. Visit CEO’s Behavioral Interventions page to learn more about CEO’s DOL-BI portfolio, including access to reports of past trials, information on current work, and additional tools for applying behavioral insights.
Explore the Practitioner’s Playbook
This playbook is designed to give program administrators and managers at the U.S. Department of Labor (DOL) and other social programs an overview of how they can use insights from behavioral science to improve the effectiveness of their programs and services. Explore each step to learn how to identify behavioral problems and use strategies informed by behavioral science.
The first step in the behavioral design process is to define your problem clearly and concretely. Managers encounter many types of problems in running their programs. Some problems are more likely to benefit from applications of behavioral science—we refer to these problems as having “behavioral” components—whereas others will be more responsive to traditional solutions.
Below is an explanation of some of the types of problems that managers frequently encounter in labor and other social programs—problems that are likely to have a behavioral component.
Low take-up. Fewer people than expected participate in a program that would benefit them. Some DOL programs may be underused by their target populations. Many programs that have clear benefits may still suffer from low participation rates. In some cases, this may be due to ineffective outreach or education about the program’s benefits. But sometimes the take-up problem persists even with strong marketing.
Poor follow-through. People do not take all the steps needed to benefit from a program. People may intend to take a certain action, but fail to do so. For example, they may intend to enlist workforce staff help to begin their job search soon after losing their jobs, but find it hard to get started.
False beliefs. People misunderstand aspects of a program or base their choices, decisions, and actions on incorrect assumptions. People may have misperceptions about DOL or other social programs that cause them to behave in unexpected ways. For example, people may not understand the eligibility rules for a program, and consequently do not apply when they could benefit from the support.
High attrition. More people start a program than finish it. People may start a program, but fail to complete it. For example, they may be required to attend a series of sessions to complete a program they opted to participate in, but they only attend the first or a few sessions and then drop out.
Download the Playbook to learn more about working with stakeholders to understand the problem you want to solve. It is important to do this before making changes to your program to ensure you target the right problem.
Once you have a fuller understanding of the problem you are trying to solve, the next step is to determine if it is caused by behavioral bottlenecks or by structural factors.
Develop a behavioral map
Use your knowledge of the context, the users, and the program to map how your target population engages with the program. This can help you identify points at which the users are likely either to make decisions or face roadblocks that can lead them away from the desired outcomes.
To get you started, we’ve provided an example of a behavioral map in Appendix A that we developed for a trial designed to help employees increase their retirement savings. Below, we describe the types of behavioral problems that are most common and likely to apply in Labor programs. We also discuss “fingerprints” that signal a particular bottleneck might be at play.
Common behavioral bottlenecks
Psychologists have discovered many biases or psychological limitations that could limit people’s engagement with Labor and other social programs. One or more of these behavioral bottlenecks may be contributing to the problem you’ve identified and are trying to solve.
Four bottlenecks—limited attention, forgetting, optimism bias, and procrastination—often contribute to problems observed within Labor and other social programs. These four bottlenecks may or may not be contributing to your targeted problem, however. Figuring out which bottlenecks are at play is like being a detective on a crime scene.
Download the Playbook to find lists of common “fingerprints” to help you determine which bottlenecks may be contributing to your problem.
Once you have identified the behavioral bottlenecks contributing to your targeted problem, it is time to design potential solutions, also referred to as “behavioral interventions.” (Throughout this section we refer to these as solutions, but we consider designs to be potential solutions unless a rigorous evaluation has found evidence of their effectiveness.)
Your designs should follow the diagnoses and address as many of the bottlenecks you discovered as possible.
Designing a behavioral solution typically involves four steps:
- Develop specifications. What do you want your solution to do? What bottlenecks do you want to solve? What constraints do you have to work around?
- Consider known solutions. Review what has worked to address similar bottlenecks before, both in similar and unrelated contexts.
- Fan out an converge. Generate as many creative ideas as possible without prejudging them, then narrow to the most promising and feasible options.
- Iterate and adapt. Create prototypes, then refine them based on feedback, pilot-testing, and monitoring during early implementation.
Download the Playbook for more strategies for program design and operations.
Once you have designed your potential solution, or “behavioral intervention,” it is ready to go in the field.
Communicate goals to implementing partners. Whether the behavioral intervention is being evaluated or not, it is important to communicate with program staff about why it is being implemented and what you hope to learn from the process.
Pilot your solutions so you can tweak design. Ideally, implementation should be rolled out gradually, so you can begin to assess how your solution fits into the program’s overall workflow. If necessary, you should be prepared to make adjustments to the design and to limit or drop some elements of the solution altogether.
Test variations to learn more about trade-offs. You also can consider comparing multiple interventions by implementing them simultaneously, to assess which has the most potential or lowest burden, before deciding which to adopt on a wider scale. This allows you to determine which interventions are the most cost-effective for the program, and minimizes your risk if one or more elements of the intervention do not work.
Get feedback often. It is also important to get feedback from the participants, staff, and program administrators on how the intervention is affecting their program experiences and/or workflow.
Download the Playbook for more tips on implementation.
Behavioral interventions are frequently tested using a variety of methods. This is because we are still learning which behavioral strategies work when and for whom, as well as how to best design behavioral solutions and use behavioral strategies to design more effective programs. The behavioral science field is relatively young, and more evidence is needed to expand our knowledge base.
To determine if your behavioral solution worked, you will often want to employ strategies that yield “causal” evidence, such as an independent evaluation. Behavioral interventions are often tested in this manner to help us, the behavioral designers, confront our own biases. Many behavioral strategies, even those based on a solid foundation in the behavioral sciences, do not work or do not work as anticipated in a new context. Testing allows us to check our diagnosis and design processes and learn from any errors we may have made.
Download the Playbook for more information about testing your solutions.
After designing, implementing, and testing your intervention to assess its effectiveness in solving the targeted problem, it is valuable to take a step back and reflect on what you learned. Did your intervention work? Which components worked well and which didn’t? Did the intervention’s effect(s) vary for members of different subgroups? If you ran two or more interventions, which one was most effective and why?
Behavioral interventions may be particularly useful when they are understood not as one-time initiatives but as springboards for ongoing efforts to improve program effectiveness. Behavioral science can work as a tool when you are inventing and refining new program elements, or as a way to customize specific interventions for a particular subgroup. Interventions that are effective in one area of a program may reveal insights that can be used effectively in other areas.
Download the Playbook for more on next steps and additional resources.