From March 2002 NASW NEWS
Copyright ©2002, National Association of Social Workers, Inc.

Key to Helping May Lie Elsewhere

Therapy Technique May Not Matter Much

Most agree that more attention needs to be paid to therapeutic relationships.

By John V. O'Neill, MSW, NEWS Staff

There is abundant evidence that of the hundreds of approaches to psychotherapy, all work about equally well, says James Drisko, associate professor of social work at Smith College. Yet social work researchers keep trying to prove that one therapy technique is better than another, he said in a paper delivered at the Society for Social Work and Research conference in San Diego in January.

It would be better for the social work profession if time were spent informing the public that those who participate in any form of therapy do far better than those who don't, Drisko said. And the time of researchers could be better spent studying the factors common to all forms of therapy, like empathy and acceptance, and the agency and client contexts in which services are delivered — areas of research that fit the values of social work but are being done largely by psychologists.

Drisko said there are 25 years of increasingly improving studies from surveys and meta-analyses showing that relative to each other, "differences across therapies are not particularly significant or meaningful." Studies comparing one therapy to another sometimes show differences, but many of these have biases or methodological problems that make genuine comparisons difficult. When meta-analyses are done, those differences fall out, and outcomes are the same or similar, he said.

A meta-analysis is a quantitative method of review that combines data from original studies and analyzes methodological and study features to account for differences in findings across a set of studies.

As could be expected, there is disagreement about the value of meta-analyses indicating little difference between therapies and conflicting studies comparing one therapy to another which sometimes show significant differences. "Some read meta-analyses and say when you control for variables, the differences wash out. Some say they don't," said Wallace Gingerich, professor at Case Western Reserve University.

Drisko cited the work of Brigham Young University psychologist Michael Lambert who estimated that four factors account for outcome variance from psychotherapy, the how and why therapy works:

• The first and largest variance in outcome, 40 percent, can be attributed to factors outside the therapy itself, Lambert wrote. These include the client's context (neighborhood and family, peer, social, workplace and spiritual supports) and the client as the common factor (therapy facilitates naturally occurring healing aspects of clients' lives, like intelligence, motivation, trust and resilience). Key client factors identified in the psychology literature include: the number, symptoms and severity of problems identified; capacity to relate; and ability to identify a focal problem. Drisko added the policy and agency context, which includes questions like: "Are services accessible, culturally sensitive, "user friendly" and reasonably priced?" and "Are work conditions supportive?"
• Lambert estimates that 30 percent of outcome variance comes from the therapeutic relationship, which includes caring, warmth, empathy, acceptance, mutual affirmation and encouragement. "There is sort of a medical assumption that if you do the treatment better, the treatment causes the change, not the human factors, the common factors," said Drisko. "So we emphasize technique, and we make out the therapist as hero and healer. What's more important is how the therapist comes across to the client, how they [clients] interpret us connected to their context — that's the real deal. It's how we act for that one person that's pivotal to their success. That's the caring, warmth, empathy and basic things that also fit with NASW's Code of Ethics, like supporting the dignity of clients."
• Lambert attributes only 15 percent of outcome variance to therapy techniques unique to a specific treatment, such as systematic desensitization, biofeedback, the miracle question in solution-focused therapy, genograms in Bowenian family therapy, the psychodynamic therapists' focus on affect and self-understanding, strength therapists' focus on strengths, or behavioral therapists' focus on behavior. Technique matters, and theories are important to help understand things and organize thinking, but they seem to account for only 15 percent of outcome, said Drisko. "The catch is, it may be that my belief in a model and my ability to convey that persuasively to the client, and their belief in my faith, in turn, is what makes it work. The specific technique is less important than my allegiance to it and the comfort it gives me that this can make a change and that I can persuade you that this can make a change so you can take comfort in it."
• The other 15 percent of outcome variance Lambert attributes to the placebo effect, the hope and expectancy that comes from knowledge clients have that they are working in a ritualized format that is an accepted approach to successful change.

Assuming the common factors outside therapeutic technique account for the great majority of outcome variance, Drisko said, has several strong implications for social work:

• Rather than a particular type of therapy, the overall efficacy of psychotherapy, based on empirical findings, is the proper background for discussions of mental health policy and funding. The empirical data may be looked at several ways. Seventy-nine percent of people treated with psychotherapy will do better at a point four months out than people untreated. Meta-analyses show an "effect size" average of 0.8 for psychotherapy, a result stronger than for many service programs and many medical treatments. "In much of the research in social sciences, a 0.2 effect size is not uncommon; our world is [built] around small effects," said Drisko. "It isn't the common notion out there in the world that therapy works. We have become very critical of professionals, but empirical evidence shows that therapy works. Since social workers do more of this work than other professions combined, that's what we ought to be promoting."
• Research on how and why therapy works will provide a more complete and valid foundation for developing effective interventions. "We've become very insular about what technique is better, and may have lost the bigger picture — that it is the common factors and not the different kinds of therapy that are really important."
• Theory and technique are important, but social work places great value on relationships, and the profession needs to emphasize this component of psychotherapy.
• Attention to how therapy works is of great interest to social work practitioners, who will read and value research results.
• Attention to contextual factors, client factors like readiness to change and therapist variables deserve consistent attention in social work education, along with theory and technique.

"The kinds of conclusions and interpretations Drisko is making are certainly worthy of more exploration," said Gingerich. "If it's true, we ought to be paying attention to a lot more things than just treatment approach."

Areas Drisko mentioned that could benefit from more social work research include: organization of psychotherapy offices; agency and institutional problems like whether caring is shown by people answering the telephones when clients call; whether people are cared about in their families, workplace and communities; how severe problems are, for what duration, and how many problems there are; motivation of clients; relationship factors like empathy and warmth.

Kathleen Millstein, associate professor of social work at Simmons College, assigns her doctoral students an article from the March/April 1995 Family Therapy Networker by Scott Miller, Mark Hubble and Barry Duncan in which the three family therapists assert that "similarities rather than the differences between models account for most of the change that clients experience across therapies."

Further, they say, ". . . internecine quarrels about relative success have begun to give therapy a bad name, undercutting its very real and deeply helpful benefits."

"There is a huge emphasis on outcome research at the expense of research on process and therapeutic relationships and conditions that facilitate it," said Millstein.

Research shows that certain therapeutic techniques work best for certain types of problems, like phobia, or that medication and therapy works best for the severely depressed, said Millstein. But the research that "compares different types of interventions is not taking sufficient account of relationships. It's not an either-or."

She added: "We should pay more attention to relationships. Nobody in the world would disagree with that."

William Reid, professor of social work at the State University of New York at Albany, has another view: The opinion that most therapies are of equal or similar value is "a little bit out of date."

"Common factors are important and account for a lot of variation," Reid said. "Despite that, they are not as overwhelming as some people think."

Reid referred to a paper he published in 1997 in Social Work Research (21:5-18) in which he concluded from meta-analyses that when studies are searched for differences in effects by problem and population, different interventions had different results. Without taking into account these dimensions, differences may be "averaged out," with method A more effective than method B for some types of problems or populations, but method B more effective for others.

A total of 42 meta-analyses met his search criteria, of which 31 reported the existence of differential effects between competitive interventions.

"Given the weight of the evidence, it may make sense to consider differential effects or the lack thereof in respect to specified problem, population and intervention match-ups rather than to refer to a general tie-score effect. . . ," wrote Reid.

"Determining whether or not comparisons between intervention methods yield genuine differences in effectiveness will always be a daunting task, and one that will frequently yield null results," Reid wrote. "But as this review has suggested, differences have begun to emerge."

Evidence-based practice can miss relationship issues that are critical to recovery, said Millstein. There is a conservative movement in the medical arena to use only interventions that have controlled studies to support them. "If the client has problem A, use this technique; or problem B, use that technique. Controlled studies take away the heart of social work — the relationship piece. It doesn't work unless you have a good relationship."

Back to NASW NEWS Contents