Does One Size Fit All?
At many junctures, we have emphasized the suffering and disruption caused by mental illness; we can now add the key fact that various forms of therapy can help with these profound problems. Meta-analyses indicate that both psychotherapy and drug therapy are effective treatments for many mental disorders. More focused studies provide a sim-ilarly encouraging message. Specific modes of therapy, and specific drugs, considered one by one, also provide help—so that people who receive treatment end up better off than those who do not.
However, many questions remain: Is everyone equally likely to benefit from these therapies? Or are some people more likely to benefit than others? And, crucially, how do the various forms of therapy compare with one another? Are some therapies faster, or more complete, or more successful than others? These are important questions to ask if we want our therapies to be effective, and also cost-efficient, in terms of time, emo-tional energy, and, of course, dollars.
Overall, therapy often improves the lives of adults who suffer from a range of psycho-logical, addictive, and health problems (see, for example, Barlow, 2004; Nathan & Gorman, 2002; Roth & Fonagy, 2004; Wampold et al., 1997). Therapy also helps children at various ages (Kazdin & Weisz, 2003; Weisz, Hawley, & Doss, 2004; Weisz, Doss, & Hawley, 2006).
Even so, therapy does not work for every individual, and, in fact, a certain proportion of patients—between 5 and 10%—actually get worse as a result of therapy (Lambert & Bergin, 1994; Lilienfeld, 2007; M. L. Smith, Glass, & Miller, 1980). What produces this deterioration? In many cases, the problem seems to be a bad therapist-patient relation-ship; in other cases, the problem may be outright incompetence in the therapist (Hadley & Strupp, 1976; M. L. Smith, Glass, & Miller, 1980). Still other cases of deteri-oration may have a subtler cause. Psychotherapy sometimes disrupts what is stable in the patient’s life yet provides no substitute (Hadley & Strupp, 1976; Lambert & Bergin, 1994). For example, the therapy may lead a patient to regard her marriage as unsatis-factory, but as she takes steps toward separation or divorce, she may become severely depressed at the prospect of being alone. Good psychotherapists are alert to such dan-gers and attempt to avert such deterioration whenever possible.
For many years, researchers thought that therapy was most likely to succeed with patients who were young, physically attractive, high in verbal ability, intelligent, and successful in other domains; however, these turn out not to be the key factors (Nathan, Stuart, & Dolan, 2000). Instead, therapy seems to have roughly equal benefits for people from a wide range of socioeconomic backgrounds (Petry, Tennen, & Afflect, 2000; Prochaska & Norcross, 2007). To put the point somewhat differently, what matters for the therapeutic benefit may not be who exactly the patient is, but (as we highlighted earlier) whether the patient feels a strong sense of alliance with the thera-pist (J. P. Barber et al., 2000; Martin, Garske, & Davis, 2000). Also crucial is whether the patient is, at the start, fully motivated to participate in the therapy and optimistic about the chances of recovery (Mussell et al., 2000). Improvement is also more likely with more therapy sessions rather than fewer (Hansen, Lambert, & Forman, 2002; Seligman, 1995). And some disorders—for example, phobias—are more responsive to psychotherapy than others, such as schizophrenia.
One point, however, is less clear and remains the focus of debate: Should people with subsyndromal conditions be treated? Some critics believe that they should and note that evidence suggests that these syndromes are a genuine concern. For example, one study surveyed over 2,000 individuals for the presence of major depression. Based on their signs and symptoms, these individuals were classified into three groups: nor-mal, diagnosable for major depression, or having subsyndromal depression—that is, they had some of the signs and symptoms of major depression but not enough to be diagnosed as having the disorder. On most measures, the people with subsyndromal and major depression were equally impaired (Judd, Paulus, Wells, & Rapaport, 1996; also Rapaport & Judd, 1998).
Others argue that treating all these individuals (and so offering similar therapies to those diagnosed with depression and those who are not-quite-depressed) would be a first step down a dangerous path, one in which people take pills or seek therapy to adjust their personalities just as they now seek nose jobs, liposuction, and face-lifts to adjust their body shapes. Indeed, this view might lead to a world in which nearly any eccentricity is regarded as problematic and a candidate for treatment, with the term normal reserved for the relatively few who are sufficiently bland to avoid labeling.Moreover, such indiscriminate diagnosis might lead people to use their subsyndromalthe difference in sex rather than to other factors (such as intellectual development, social class, and so on).
We discussed how investigators design studies and collect data. So we’ll start here with what investigators do once their data have been collected. Their first task is to organize these data in a meaningful way. Suppose the study used two groups of 50 boys and 50 girls, each observed on 10 separate occasions. This means that the investigators will end up with at least 1,000 separate numerical entries, 500 for the boys and 500 for the girls. Something has to be done to reduce this mass of numbers into some manageable form. This is usually accomplished by some process of averaging scores.
The next step involves statistical interpretation. Suppose the investigators find that the average score for physical aggression is greater for the boys than for the girls. (It probably will be.) How should this fact be interpreted? Should it be taken seri-ously, or might it just be a fluke, some sort of accident? For it is just about certain that the data contain variability: the children within each group will not perform identically to each other; furthermore, the same child may very well behave differently on one occasion than on another. Thus, the number of aggressive acts for the boys might be, say, 5.8 on average, but might vary from a low of 1.3 (the score from completely calm Calvin) to a high of 11.4 (the score from awfully aggressive Albert). The average num-ber of aggressive acts for the girls might be 3.9 (and so lower than the boys’ average), but this derives from a range of scores that include 0 (from serene Sarah) and 6.2 (from aggressive Agnes).
Is it possible that this difference between boys and girls is just a matter of chance, an accidental by-product of this variability? For example, what if boys and girls are, in fact, rather similar in their levels of aggression, but—just by chance—the study hap-pened to include four or five extremely aggressive boys and a comparable number of extremely unaggressive girls? After all, we know that our results would have been differ-ent if Albert had been absent on the day of our testing; the boys’ average, without his contribution, would have been lower. Likewise, Agnes’s twin sister was not included in our test group because of the random process through which we selected our research participants. If she had been included, and if she was as aggressive as her twin, then the girls’ average would have been higher. Is it possible that accidents like these are the real source of the apparent difference between the groups? If so, then another study, without these same accidents, might yield a different result. One of the main reasons for using statistical methods is to deal with questions of this sort, to help us draw use-ful conclusions about behavior despite the unavoidable variability, and, specifically, allowing us to ask in a systematic way whether our data pattern is reliable (and so would emerge in subsequent studies) or just the product of accidents.