The four responses in the December issue of Mental Retardation to my earlier article (Sturmey, 2005a) are interesting and welcome. I offer the following comments in reply.
Beail (2005) commented that my conclusion that behavioral treatments are to be preferred was incorrect because the outcome literature on behavioral treatments is limited. Specifically, he noted that Didden, Duker, and Korzelis' (1997) finding that the participants in the majority of studies in their meta-analysis were, on average, 16 years of age; usually people with severe and profound mental retardation; and commonly exhibited self-injury or stereotypy; few had mental health problems; the range of intervention methods included 25% of investigators using aversive methods; and many took place in institutional settings. Further, many studies had methodological limitations, such as lack of functional analyses, no use of least intrusive alternatives, limited reporting of generalization data, and employment of insensitive measures in some meta-analyses.
Beail's (2005) representation of Didden et al. (1997) is accurate, but incomplete. First, their meta-analysis was not a meta-analysis of behavioral treatments but of all available treatments. What is notably absent is any controlled studies of psychotherapy, cognitive, and other therapies, a finding replicated by Prout and Nowak-Drabik (2003) and other reviewers of the literature (Sturmey, 2004). There have been several randomized controlled trials of packages of cognitive and behavioral methods published since these reviews (e.g., Taylor, Novacco, Gilmer, Robertson, and Thorne, 2005). Hence, inferences on the effectiveness of psychotherapy have to be made on the basis of comparing this large, but imperfect literature on behavioral approaches with complete absence of controlled trials for psychodynamic and psychoanalytic psychotherapy and many other therapies (J. Jacobson, Foxx, & Mulick, 2005) and with the outcome literatures for well-evaluated, but ineffective, therapies, such as sensory integration training (Vargas & Camilla, 1999) and auditory integration training (Medford et al., 2000). Further, Didden et al.'s meta-analysis is out of date and, obviously, excludes recent studies on the efficacy of behavioral interventions.
There have been fewer behavioral studies of mental health disorders than studies of challenging behavior in people with mental retardation, but I contest Beail's (2005) statement that “there is very little evidence for the effectiveness of behavioral treatments of mental health.” There are multiple studies on behavioral treatment of phobic disorders (e.g., Matson, 1981), several studies on psychotic disorders (e.g., Mace, Webb, Sharkey, Mattson, & Rosen, 1988), and few studies on depression (Sturmey, 2005a.) A more systematic review of the literature might reveal more. Further, many of the participants in the studies analyzed by Didden et al. (1997) may have had psychiatric diagnoses, as is common in people with aggression and self-injury, but because these studies were behavioral. The researchers underplayed or did not report psychiatric current diagnoses. Thus, many participants in behavioral studies of aggression were likely to also be diagnosed with a range of mental health disorders and treated with psychotropic medications. Finally, Beail's contention that “The evidence base may not generalize or transfer to community-based interventions” (p. 444) is a simple empirical question that current and future researchers can readily answer. The publication of a number of such studies over the last 10 years answers this question (e.g., Christina & Poling, 1997; Mace et al., 1988; I. Taylor & O'Reilly, 1997).
J. Taylor (2005) stated that outcome research is concerned with efficacy and internal validity, but effectiveness required demonstration of external validity. He also claimed that (a) there is “a wealth of evidence, particularly cognitive–behavioral approaches” (p. 450), (b) that self-management is incorrectly labeled as a behavioral intervention when it is truly a cognitive intervention, and (c) “cognitive–behavioral treatment is a more beneficial approach. … self-actualizing in nature … [and promotes] portable internalized control in order to facilitate transfer across situations [that is] is helpful” (italics added p. 451).
Demonstrating that research findings generalize to clinical practice is indeed important. However, the studies that J. Taylor (2005) cited in support of his hypothesis presented no data to evaluate this possibility. Lynch (2004) reviewed reports of clinical experience, but reported no data, and Prout and Nowak-Drabik (2003) conducted a meta-analysis of published research, which contains no data on clinical practice. Likewise, the citation of Willner's (2005) review as evidence of “a wealth of evidence, particularly cognitive–behavioral approaches” (p. 450) seems inaccurate. In his review of research on cognitive therapy, Willner discussed two groups of procedures: self-management and cognitive therapy, but did not include data on routine clinical practice. Hence, Taylor's contention that there are data from clinical practice that support the use of cognitive–behavioral therapy awaits investigation. Further, although there is accumulating evidence of the effectiveness of cognitive–behavior therapy packages, such as anger management (J. Taylor & Novacco, 2005), there are currently no data to indicate the relative contribution of behavioral and cognitive components of this package. Indeed, the absence of evidence for cognitive therapy, the evidence of the efficacy of behavioral components in cognitive–behavioral packages (Sturmey, 2004, in press, a), and data from components analyses showing that behavioral components may be the effective elements in such packages (e.g., N. Jacobson et al., 1996). These strands of evidence suggest that it is the behavioral components of such packages that are the likely effective agent.
J. Taylor (2005) claimed self-management as a cognitive therapy. Yet Skinner (1953, pp. 227–241) provided one of the earliest behavioral accounts of self-management that rejects the controlling, initiating self as the cause of observed behavior. In his account, a controlling response, such as writing a note to oneself, changes the future probability of a controlled response, such as buying bread at the grocery store. The relevant independent variables are those that control the controlling response. Depending on a person's learning history, he or she may eventually emit private verbal behavior, such as saying to oneself “Don't forget the bread,” to control his or her behavior. Skinner recommended that psychotherapists assist their clients in discovering their own functional assessment and changing their own behavior based on these methods of self-control (pp. 350–358). This behavior analytic approach to self-management has been useful with people who have mental retardation, including those with mild and moderate mental retardation in community settings, both conceptually (I. Taylor & O'Reilly, 1997) and practically (Christina & Poling, 1997). Hence, self-management is a long-established behavioral procedure that has been conceptualized in radical behavioral terms for over half a century.
Important differences exist between behavioral and cognitive accounts of self-management. Radical behaviorists do not deny the existence of private events, such as thinking and feeling, but do not give them any special status as a cause of behavior (Skinner, 1953.). Rather, behavior analysts construe them as private behavior to be explained like any other behavior by a functional analysis of the controlling environmental variables (cf. I. Taylor & O'Reilly, 1997.) In cognitive models, cognitive therapy somehow permanently changes the controlling, initiating, and portable (but nonobservable) self, which in turn causes behavior to change: hence, the claim that self-management strategies are self-actualizing and produce generalization. However, to date, there is no data-based report of generalization from cognitive–behavior therapy conducted either in therapists' offices or the institutional settings where some evaluations of cognitive therapy have taken place. If generalization is observed, it will be an interesting phenomenon that can be subjected to a functional analysis. If it is not observed, behavior analysis has both a conceptual system and a technology that can be used to promote generalization (Stokes & Baer, 1977).
J. Taylor (2005) correctly noted that the question of “what works for whom and for what” (p. 451) has not been directly addressed. Indeed, the small number of researchers using group designs generally have not done so. The absence of these studies still does not detract from findings from group studies of behavioral interventions showing their effectiveness. In any case, functional analysis provides a generic framework for designing individually based interventions that may indicate what does and does not work for individuals (Sturmey, 1996, in press, b) and leads to treatments with larger effect sizes (Didden et al., 1997).
Hurley (2005) and King (2005) both appealed to clinical and personal experience as testimony for the efficacy of psychotherapy; King is explicitly antiscientific. Their personal and professional experiences and opinions command respect but also caution. Expert experience and opinion can be a valuable source of hypotheses that are worthy of future empirical investigation, but impassioned voices have been wrong. Many people fervently believe in extra sensory perception (ESP), creationism, mind-reading, and aliens, but their passion is not evidence. Experience and passionate advocacy suggest that secretin, facilitated communication, auditory integration therapy, and sensory integration therapy were highly effective, but in controlled studies researchers failed to confirm these passionate beliefs (J. Jacobson et al., 2005; Sturmey, 2005b). Hurley and King may be right, but empirical research is needed to confirm their experience and impressions.
Hurley (2005) also claimed a variety of behavioral interventions as “cognitive–behavioral interventions” (e.g., relaxation training) because “[clients] can use interpersonal relationship format, follow advice, accept feedback, and learn new ways of thinking and behaving” (p. 446). Relaxation training is more parsimoniously described as antecedent control by therapist instruction of the behavior of clients' muscles, maintained by therapist differential reinforcement of relaxed behavior. Reference to the self-performing “self-therapy” is unnecessary (Skinner, 1953). Modifying the behavior of muscles is not cognitive therapy.
Finally, Hurley (2005) asked, “What would Sturmey do” for a variety of client problems that she described. I would treat them like every other behavior change procedure. I would conduct a functional assessment, teach the client self-regulation if possible, and/or modify the environmental variables that control the behaviors of interest to produce clinically significant behavior change in many settings and respond in ways that are as effective and acceptable as possible (Skinner, 1953; Sturmey, 1996; in press,b).
Author: Peter Sturmey, PhD (firstname.lastname@example.org), Department of Psychology, Queens College, CUNY, 65-30 Kissena Blvd., Flushing, NY 11367