mmilot
- Feb 9, 2022
- 2 min read

Are synthetic controls the future of research in the counseling industry?

For years Workreach Lab has been advocating for stronger studies that incorporate control groups when evaluating the effectiveness of counseling services (EAPs, mental health services, other counseling services/platforms), but due to their costs and complexity there tends to be limited interest by service providers.

Recent advances however now allow for control group studies with “synthetic controls” at a much lower cost and complexity level than RCTs. First used in policy and population health studies, synthetic controls are now being adapted for clinical research. Synthetic data are based on an approach that learns the complex statistical relationships in an original set of data, enabling the generation of new realistic samples of individual-level simulated data at larger volumes while preserving the full quality and performance.

For counseling service providers, synthetic controls lower the cost-barrier to better research and simplify data procurement for the rapid completion of impact studies that better showcase the value and effectiveness of services, contributing to the development of evidence-supported business cases.

Data from a synthetic control group behave similar to data coming from a real control group from which they are generated, including changes over time and associations between variables (e.g., such as mental health and work performance). With access to a large number of synthetic controls a comparison group can be built that is equivalent to the group of counseling users on demographics, occupational characteristics, presenting symptomology, depression level, and other variables.

A counterfactual becomes available to establish a causal association between intervention and outcomes by answering the question: what would have happened in a similar group of workers if they hadn’t accessed the counseling service? While synthetic controls are not perfect, they can provide much stronger evidence of impact than the more widely used nonexperimental approaches (e.g., pretest/posttest only), and studies can produce findings publishable in peer-reviewed journals.

Using a synthetic control approach developed at Workreach Lab to evaluate counseling services, we conducted a quasi-experimental control group study that determined the effectiveness of a workplace counseling service on measures of depression (PHQ), anxiety (GAD), and work performance (HPQ, WOS). With the synthetic controls, it was possible to determine the impact of the counseling all without needing to collect outcome data from an actual control group of workers. The causal (vs. correlational) impacts of counseling were discernable and statistically/clinically significant.