Art and Science III: SlateStarCodex on Gottman
Scott Alexander, my favorite writer on the interweb, has been reviewing therapy books and got around to a particular specialty of mine, couples therapy. He reviews John Gottman’s landmark book, Seven Principles for Making Marriage Work. And, in the process of the review, he skewers Gottman. If you’re a couples therapist, you have to read the post. It’s a sober but hilarious takedown of the haughty claims so commonly made in our field.
As a couples therapist who is a religious SlateStarCodex reader, I felt compelled to write a response in the comments (which I’ve reworked into this blog post). I read Gottman’s books before I was into SlateStarCodex and rationality, and before I was into the research on what makes psychotherapy work. Scott’s post expresses many of the same sentiments I have had about the field recently.
John Gottman claims he can observe a couple interact for five minutes and with 94% accuracy predict whether they will divorce within five years. But Gottman’s prediction is actually a postdiction… and it’s a claim that’s quite similar to claims about the success of one treatment style over another – fast and loose with the data and uncritical in regards to methods. Even the study in his FAQ, which he claims is the one that’s truly predictive, admits it’s a post hoc analysis in the abstract (there appear to be two different studies he could be referring to but it’s true for both of them: one, two). It’s another frustrating instance of scientists wildly overstating their claims in public statements while being forced to make more reserved statements in the text of the actual paper that will be scrutinized by their peers.
The Heart and Soul of Change: Developing What Works in Therapy has a chapter on what works in Marriage and Family Therapy. It states, “A critical review of the differential efficacy data demonstrated few exceptions to the dodo verdict when allegiance is considered, comparisons are fair, and bona fide treatments are contrasted, eroding claims of differential efficacy and giving credence to the claim that all have won prizes.” (p.365) They also note that in research on couples therapy, Allegiance effects are a significant issue, as they are in all psychology studies comparing treatments. “Allegiance is the researcher’s belief in and commitment to a particular approach… [it] can exert a large influence on outcome in comparative studies.”
So Gottman’s particular principles are not listed as a definitive element of effective marriage therapy. It seems the Building Strong Families Program would also suggest that the elements described by Gottman don’t, as a general rule, necessarily generate a positive impact on couples.
This shouldn’t be a surprise. It is what we should expect.
Spending a lot of time focusing on The Great Psychotherapy Debate, I’m tempted to say that effective couples therapy follows the same principles outlined in that book. That is, therapy works not when we have specific ingredients for certain disorders (eg. exposure for PTSD, response prevention for OCD, challenging negative cognitions for Depression, reducing negative cycles for Marital Conflict) but because of an alliance in a relationship with a healing figure who agrees with you on your goals, provides an adaptive explanation for the problem that is accepted by the client(s), and clearly delineates where the client can put in effort towards making a positive change. See Wampold’s article on effective therapists for the briefest statement on what effective therapy should be.
I think this has left me with a better model (better at making predictions about outcomes) than I had in the past. I would bet on the claim that specific ingredients (eg. Tackling the Four Horsemen) or adherence to a treatment protocol (eg. Oppa Gottman Style) are not predictive of outcomes in couples therapy. So, in my view, it won’t be a particular model that we can recommend to clients. I would bet on the claim that the couples’ rating of the alliance (eg. “I feel heard, understood and respected by the therapist”), goal congruence (eg. “we talked about what is important to me”), and approach (eg. “the therapist’s explanation of the problem and way of working on it make sense to me”) will be a significantly better predictor of outcome. So, in my view, it means there will be particular therapists that you can recommend to clients.
Does this mean that anything works in individual/couples therapy? No, we do need a clearly defined model and method to explain and work on the problem. But other factors are much, much more predictive of outcome than what model/method we choose. Does the ambiguity there mean we are taking in gullible marks? Does it mean we’re just giving placebos? I don’t think that’s the right way of looking at it (and if it’s true it may “prove too much” about medicine in general). I think a better way of looking at it is “we are in the business of encouraging the process of healing and change. How can we improve our ability to foster that in our patients?”
To work well with couples you need to be confident in your ability to assess, interrupt and discuss the dynamics of the dyad in real time. I don’t think people who have just done ‘a little couples counseling on the side’ are the right people to refer couples to. In couples work, we have to navigate the tricky business of remaining neutral in conflicts. To do so requires viewing problems in the couple as invariably A. part of a dynamic of the system, where B. both members of the couple play a part in reinforcing the dynamic, and C. “bad” behavior is the result of historical trauma and/or presently hidden fears.
To improve my outcomes, I try to elicit, accept and enact feedback from my clients about what they would find helpful from me in the room. I base this on Scott Miller’s work on feedback in psychotherapy. I’m excited to attend my first Feedback Informed Treatment workshop next month. Going through Scott Miller’s blog, I noticed a reference to a recent paper by Norcross showing that patients prefer more directiveness than therapists would prefer to give. If we want our outcomes to get better, we need to take client’s feedback on their preferences into account.
I’m a more directive therapist to begin with but for some clients I sit back a lot and with others I lean forward. With couples, I find that in general it is necessary to take up a lot of space in the room. That means interrupting to point out something that’s being enacted. But it can also mean catching a couple in the middle of a conflict where they are both physiologically aroused. To effectively do this means noticing the arousal, interrupting and distracting by discussing a tangentially related topic for a few moments. Then point out the tactic, discuss physiological arousal and return to the conflict topic with a gentle startup (lol, Gottman). It’s definitely “a firmer hand” in that it’s directive, interrupting, and much more talking on the part of the therapist.
In the wake of Scott’s post, I’ve been talking about couples work with friends. They asked me what styles I employ and I was hesitant to reply (given what I said about comparing modalities in these posts) but when push came to shove, I identified Structural Family Therapy, Satir Family Systems, Gottman, and Emotionally Focused Therapy. I find it helpful to have many tools that I can deploy depending on what I find my clients in the room need at that moment.
Being a therapist places us loosely within what we could call the “helping” or “healing” professions. That vague box is a tricky place to be. I’ve come to realize it requires a continual effort to learn how to help and heal more effectively. Sorting through claims about what works and what doesn’t is part of that ongoing project for me.