Most readers would agree that the phrase "evidence-based practices" describes a concept that is sweeping the field of health care, including mental health care. Indeed, who among us would not agree that before submitting to a health care procedure, we have a right to know the success rate, side effects, and other important issues related to that procedure? Randomized trials are considered the state-of-the-art research method when it comes to building the scientific evidence base on which treatment decisions can be founded.

In the past decade we started on a course of comparing program A with program B, using random assignment to program. We constructed fidelity scales to ensure that the program structures were in fact different. We assumed that the differences in program structures created by the program innovators were the differences that really mattered, when in fact the programs may have been very similar (positively, negatively, or both) in what really matters—that is, the process occurring between service recipients and practitioners.

I suggest that we refocus our research agenda on studying processes, which may cut across differently named program models. Research in mental health and other fields indicates that such studies might include processes such as collaborative goal setting, skills training, developing a person-centered plan, building the relationship between practitioner and service recipient, providing environmental accommodations, and coaching. Investigations of this nature can be made through randomized studies comparing different ways to implement these processes as well as through naturalistic observational studies that more fully incorporate in the study design and data collection the perspectives of the persons being served.

The study of evidence-based processes does not relegate program structures to insignificance. I would suspect that certain structural ingredients, such as small caseloads or consumer-providers, allow critical processes to occur more effectively. In process study, however, program components can be examined with respect to how they promote or hinder certain processes. The evidence-based process, rather than the evidence-based practice, is examined as the primary source of outcome variance.

Continuing study of program structure alone in order to differentiate models and their unique effects on people's outcomes is not where the real action may be. Controlled clinical trials of programs have not controlled the processes going on within those programs; in fact, they have virtually ignored the processes. It is time to emphasize the human interactive process as it occurs within differing programs as a fundamental target of scientific research.