Psychotherapy Supervision in the 21st Century: Some Pressing Needs and Impressing Possibilities

Abstract

As a professional and educational service, psychotherapy supervision looms large in importance. Yet if psychotherapy supervision is to most viably advance in the century ahead, a number of pressing measurement, research, and training/practice needs cry out to be better addressed. Ten such needs are identified by drawing on recent major research reviews and other substantive supervision publications. The author concludes that better addressing these needs would expand and fortify the empirical base of psychotherapy supervision and also enhance its training and practice base.

Psychotherapy supervision has long been regarded as a key means by which therapist trainees learn to become effective psychotherapists.^1–7 Psychotherapy supervision typically is an integral part of all mental health preparation programs, be they in psychiatry, clinical or counseling psychology, or psychiatric nursing.⁸ Surveys suggest that many mental health professionals provide psychotherapy supervision services and devote a fair portion of their professional time to doing so.⁹

Over the last two decades, much growth and evolution has occurred in the theory, research, and practice of psychotherapy supervision.¹⁰ Such growth and evolution, for example, can be seen in the rise of developmental models about supervisors and supervisees,^11–16 an increase in research into varied aspects of the supervisory experience,^17–19 and efforts to concretize and render ever more specific and meaningful the supervisory encounter.^20,21 That growth and evolution is good, reflecting progress in a much valued and, in some respects, much underrated professional activity: clinical supervision.

With such advances noted, what can or should be expected of psychotherapy supervision as we move into the 21st century? What needs must be better addressed if psychotherapy supervision is to most fruitfully advance as a professional service? What possibilities, if explored more vigorously, could help psychotherapy supervision further advance and emerge with a more solid empirical and practical base? This review, although not exhaustive, will attempt to identify some of the important needs and possibilities that merit more pressing attention from supervision theorists, researchers, and practitioners alike in the years ahead.

In the last decade, a number of substantive supervision research reviews, articles, chapters, and books emerged, and many of these converged with regard to important needs and possibilities in psychotherapy supervision. Ten key needs and possibilities culled from the recent literature are accented here.

Needs and Possibilities for Psychotherapy Supervision

1. The need for valid, reliable supervision measures. Research is only as good as the measurement tools and procedures that are used for assessment and evaluation. In recent years, several supervision publications have emphasized one point vigorously: more valid, reliable, supervision-specific measures are needed to advance research efforts.^{1,17,18,22–24} Many measures now in existence—for example, the Supervisory Levels Questionnaire–Revised²⁵ or the Supervisory Working Alliance Inventory²⁶—have been criticized for not having sufficient psychometric soundness. In some cases, sample sizes have been too small, a measure's scales have not been orthogonal, or the most appropriate statistical procedure has not been used in data analysis.¹⁷ This is not to say that progress has not occurred here; it clearly has. (See, for example, the Role Conflict and Role Ambiguity Inventory,²⁷ the substantially revised short form of the Barrett-Lennard Relationship Inventory,²⁸ and the Psychotherapy Supervisory Inventory.²⁹) Still, some of the primary problems that plague supervision research are these: 1) too many supervision measures have been borrowed from psychotherapy research and have not necessarily been developed with the supervisory endeavor in mind, and 2) too many supervision measures have been “one-time wonders,” created for the study at hand and never used again. Lambert and Ogles,¹⁸ in their review on the effectiveness of psychotherapy training and supervision, had the following to say on this point:

The field is still characterized by a plethora of homemade devices that are used only once and discarded. In addition, few researchers bother to publish reliability or validity data on their scales. . . . Advances in knowledge can be expected to increase with advances in criterion measurement. Research in this area is highly recommended and promises to affect theory as well as practice. (p. 441)

Thus, one of the most pressing needs for psychotherapy supervision in the next century remains the development and establishment of reliable, valid criterion measures to guide supervision research.

2. The need to research supervision outcome.Research that examines the effects of psychotherapy supervision on patient improvement or lack of improvement has been minimal. In the past 5 years alone, three major reviews^18,19,30 have arrived at that conclusion. Some modest, indirect support for a training/supervision–therapy outcome link can be found,³¹ but more than a modest, indirect link needs to be established between the two. As Stein and Lambert³² put it: “Given the enormous national investment of physical and human resources in graduate programs, it is quite remarkable that more compelling evidence is not available that demonstrates that graduate training directly relates to enhanced therapy outcome” (p. 194).

Increasing research attention has been brought to bear on the subject of psychotherapy outcome over the last generation; that attention has borne much fruit, yielding new insights into the workings of psychotherapy and providing a more solid empirical foundation for its practice.³³ That same sort of attention must now be brought to bear on psychotherapy supervision outcome research. Thus, a most pressing need for the next century remains as follows: establishing a strong, direct link between supervision and therapy outcome—and, assuming that is found, establishing what in supervision enhances patient outcome and how that is done.

Establishing such a link, however, may be easier said than done. Wampold and Holloway³⁴ have provided a useful discussion on this issue, proposing a causal model of supervision outcome phenomena. In discussing the different elements of their model, they make a most interesting point: “Connection between supervision process and distal outcomes [e.g., patient change] will need to be established by studies focusing on various pieces of the mediated causal process . . . rather than by trying to establish relations among distal elements” (p. 23). If their suggestion is followed, establishing a supervision–therapy outcome link may be more easily accomplished.

3. The need for more rigor in supervision research. Advances in supervision research have occurred over the last two decades,^9,35 but the need for more rigor in future experimental efforts remains paramount. In the past couple of years, two major research reviews by Ellis and Ladany et al.^17,36 have emphasized the importance of this need as never before. Because of the significance of those two reviews, they deserve special attention from supervision researchers and theorists. In the first review,³⁶ 144 research studies of clinical supervision appearing from 1981 to 1993 were evaluated with regard to methodology (e.g., threats to validity) and statistical variables (e.g., inflated Type I error). In their second review,¹⁷ Ellis and Ladany sought to replicate and extend what they had done earlier “by performing a more circumscribed methodological review” (p. 450). Their findings and conclusions are both sobering and startling:

91% of the investigators did not attempt to control systematically Type I or Type II error rates. The majority of the investigations of supervision were simultaneously unlikely to detect true effects and very likely to find spurious significant results.³⁶ (p. 43)

Three threats to construct validity were consistently found: monomethod bias (79%), confounding of the construct with limited levels of the construct (69%), and inadequate preoperational explication of the constructs (69%).³⁶ (p. 41)

Studies were found to have inconsequential hypotheses (83%), ambiguous hypotheses (80%), and diffuse statistical hypotheses and tests (99%).³⁶ (p. 41)

More than 70% of the studies reviewed used ex post facto designs, and hypothesis validity was severely compromised in nearly every study.³⁶ (p. 44)

The overall quality of [clinical supervision] research during the past 15 years was substandard. This cannot be taken lightly. . . . Few conclusions were justifiable given the lack of replicated results and the conceptual and methodological problems besetting the studies.¹⁷ (p. 492)

Readers should approach the empirical [supervision] literature with skepticism.¹⁷ (p. 496)

As those statements make clear, matters of research design and experimental control have been key problems in supervision research over the last 15 years. Ellis and Ladany make a number of useful recommendations to which supervision researchers should attend in order to best deal with those problems—for example, formulate unambiguous hypotheses; perform a priori statistical power analyses; incorporate manipulation-treatment checks; and test assumptions underlying statistical procedures. They even present a detailed description of a “good” supervision study. Their study and set of recommendations provide a useful blueprint to guide supervision research, and, as they emphasize, their words have some relevance for both qualitative and quantitative research investigations. (See Worthen and McNeill³⁷ for a recent example of a well-conducted, rigorous qualitative study.)

Again, those two reviews, because of their comprehensiveness, thoroughness, and painstaking attention to detail, represent significant contributions to the supervision literature. Calls for more rigor in supervision research are not new.³⁸ But it becomes clear that those calls have yet to be heeded. Supervision research since 1980 generally has not been sufficiently rigorous; if it is to be most informed and informative, then one of the most pressing empirical needs confronting us now is to infuse future experimental efforts with the proper rigor.

4. The need for the development of supervision manuals. Psychotherapy treatment manuals have increasingly emerged as viable means to research psychotherapy outcome and train therapists in particular theory-specific skills. Indeed, over a decade ago, Luborsky and DeRubeis³⁹ referred to the treatment manual as a “small revolution” in psychotherapy. Since then, treatment manuals have only become more plentiful, and their promise has been recognized by many.^18,40–45

Although treatment manuals are not without drawbacks, they still have served a useful purpose in advancing the field of psychotherapy. Unfortunately, comparable manuals for training supervisors have been slow to develop. Only one, published a couple of years ago, exists.⁴⁶ That is a beginning, but 10 times that many psychotherapy treatment manuals existed 10 years ago.³³ Because such manualized attention in supervision is lacking, the field's advancement has been delayed.

A good supervision manual could be useful in at least three ways: 1) in defining and concretizing the supervision experience, thereby rendering it more researchable; 2) in facilitating the training of supervisors in particular supervision skills; and 3) in facilitating the training of supervisors in therapy-specific supervision approaches, such as cognitive approaches. Thus, such manuals can have ready benefit for supervision training and practice as well as research. The development of manuals to train supervisors must receive more serious attention from the supervision community, and that clearly is a pressing need that has yet to be reckoned with.

5. The need for a multi-method, multi-rater, behavioral, longitudinal focus in supervision research. As constructive critique of supervision research has continued to mount, two problematic patterns have begun to emerge: 1) in many studies, investigators inquire only about the perceptions of either supervisor, supervisee, or both; and 2) many if not most studies are cross-sectional in nature.

Before a need is identified here, it must first be said that the field requires further cross-sectional studies, and the study of supervisor and supervisee perceptions of supervision must continue. But a nice complement to such studies, as many reviews and commentaries appearing over the past decade have emphasized,^{16,18,19,23,47–49} would be these:

1. A greater focus on the behavior of supervision, examining what actually happens in supervision, what supervisors and supervisees actually do, what actual changes occur in supervisee performance, and what actual changes occur in patient outcome.

2. The use of multiple indices to measure supervision process and outcome.

3. The use of multiple raters to provide data about the supervisory experience.

4. Longitudinal studies of psychotherapy supervision, allowing investigators to examine the process of growth and development of therapist trainees over time.

With observable behavioral data, a number of viable targets could be identified for study—for instance, changes in supervisee planning and conceptualization, in-session cognitions, and intentions.¹⁶ From a developmental, longitudinal perspective, it would be interesting to study how trainees' in-session cognitions or intentions might change over the course of a year's supervision with the same supervisor; that type of work thus far has not been done. As for using multiple indices for measurement, several can be readily identified: rating scales, questionnaires, patient outcome data, process measures.²³ Those all can be viable means for researching supervision and avoiding the monomethod threat. Multiple raters could include supervisor, supervisee, and patient, as well as trained observers who evaluate audiotaped or videotaped sessions.^16,23

Multiple indices and raters of process and outcome allow for a fuller, more comprehensive picture of the supervisory endeavor to emerge. A longitudinal focus would bring a valuable perspective to the dismantling and understanding of the supervisory experience across time; such a focus is absent in the supervision literature, which is all the more surprising when one considers how popular a developmental view of supervision has become since 1980.^47,50 Thus, these are clear, compelling needs that must be better addressed in the years ahead.

6. The need for follow-up and replication studies. In two recent major research reviews,^17,18 one very prominent problem was identified: the supervision research literature is characterized by a lack of follow-up and replication studies. As Lambert and Ogles¹⁸ noted: “Few follow-up studies have been reported in the literature, and even when included as part of the research design, follow-up studies are marred by uncontrolled variables” (p. 427). To that, Ellis and Ladany¹⁷ added the following: “Without replications, there is no way of establishing the veracity of theories or previous findings. That is, to help rule out rival explanations (i.e., to ascertain whether a particular pattern of results was due to Type I or Type II errors, or other rival explanations), the results over several replications or replication and extension studies need to be virtually identical to those of the original study” (p. 493). In their review of 104 clinical supervision studies,¹⁷ Ellis and Ladany found only four replication or replication-extension investigations.

It is surprising to see that so little has been done with regard to follow-up, replication, or replication-extension. The need for such work is self-evident and poses a real challenge that supervision researchers would do well to address sooner rather than later.

7. The need to study moderating variables. What variables moderate the supervision process? That question was raised in a well-done empirical study conducted almost 10 years ago by Tracey et al.⁵¹ They investigated the effects of supervision content (e.g., noncrisis versus crisis client), and supervisee reactance level (e.g., low versus high) on the supervision process itself. Up to that point, much theory and research had supported the idea that beginning supervisees need more structure whereas more advanced supervisees need less. Tracey et al. challenged that, hypothesizing that both supervision content and supervisee reactance level would affect desire for structure; their hypotheses were supported. When presented with a suicidal patient, both beginning and advanced supervisees wanted a more structured supervision approach. When presented with a noncrisis patient, highly reactant advanced supervisees opted for more unstructured supervision; beginning and low-reactant advanced supervisees did not. Because of its significance, this study by Tracey et al. has been often cited in the supervision literature since its publication.

But what can be drawn from that study with regard to psychotherapy supervision research in the 21st century? Need remains for important moderating variables to be better incorporated into supervision research. As yet, that has not been done to any substantive degree. How do such variables as level of ego development, conceptual level, and dispositional affect (e.g., negative affectivity) affect supervisees' ability to use supervision and perform effectively in therapy?¹⁶ How does a variable such as self-criticality affect supervisors in their growth as supervisors and in their ability to perform effectively in supervision?⁵² Those are viable research questions, viable moderating variables, that, thus far, have been only minimally studied or not studied at all. That must change. By better incorporating certain moderating variables into supervision research, a more informed perspective about supervisee needs can be gained and a more informed perspective can be brought to bear on supervision practice.

8. The need to study diversity in supervision. Appreciation of diversity and respect for gender, ethnicity, and lifestyle differences has emerged as a major force in the United States. With that has come increasing attention to how diversity affects psychotherapy, and some definite advances have been made in regard to psychotherapy research, theory, and training and diversity issues.³³ But the effects of diversity in supervision have been minimally investigated.

Neufeldt et al.,¹⁹ in their recent review of research on supervisor variables, indicated that 1) no research on how supervisor gender affects either therapist behavior or patient outcome has been conducted, and 2) only one study about supervisor ethnicity and its training affects has been conducted. Ellis and Ladany,¹⁷ in considering supervisee and client variables in supervision, reported on seven studies of supervisee–supervisor gender matching, finding all to be flawed in some significant way, and concluding that “inferences pertaining to gender effects in supervision seem inappropriate” (p. 469). They found only three studies about supervisee ethnicity. Stoltenberg and McNeill,²² in summarizing needs in developmental supervision research, asserted the following: “Although issues of gender, multicultural, and gay and lesbian supervision have been discussed in the literature . . . few empirical investigations have been conducted to examine their interaction with developmental models” (pp. 198–199). When comparing these comments and conclusions with those made in earlier reviews,^23,38,48 one finds that not much has changed in the last 10 to 15 years.

The foregoing leads to one simple but inescapable conclusion: incorporation of matters of diversity—gender, ethnicity, and lifestyle—into clinical supervision research is sorely needed. As Lopez asserts, “Culture matters . . . in clinical supervision”⁵³ (p. 586); the same could be said for gender and lifestyle.^22,23,54 But research that informs us how they matter is missing.

9. The need for training in how to supervise. Another pressing need for psychotherapy supervision is for a meeting to take place between logic and practice with regard to supervision training. Logic suggests that training in how to supervise would be beneficial for would-be supervisors. Yet in practice, the norm is that those providing supervision services have received minimal to no training in how to provide supervision.^1,24 I have previously commented on this situation as follows³⁵:

Something does not compute. We would never dream of turning [unsupervised] untrained therapists loose on needy patients, so why would we turn untrained supervisors loose on those untrained therapists who help those needy patients? Just as becoming a therapist is a labor-intensive endeavor for which training and supervision are needed, so too can the same be said about becoming a supervisor. (p. 604)

That opinion is not unique; many others have also emphasized the importance of and need for supervision training.^1,24,55 Such training has the potential to better prepare supervisors, render them more skilled from the outset, enable them to handle supervisory problems and issues more effectively, better facilitate the growth and development of the therapist trainees they supervise, and better facilitate the growth and development of the patients their trainees treat. Thus, it is imperative that logic and practice meet, in such a way that training in the art and science of psychotherapy supervision is made more available to would-be supervisors and is considered a necessity, not a dispensable luxury.

10. The need for psychotherapy supervision standards. In the last few years, mention has been made of the need to establish standards for the education and practice of psychotherapy supervision. Rodenhauser⁵⁶ notes that the date of the last set of American Psychiatric Association standards for supervisors was 1957. In an earlier publication, he asserted the importance of “attention, at the level of a national organization such as the American Association of Directors of Psychiatric Residency Training, to the need for instructional models and standards for training psychotherapy supervisors”⁵⁵ (p. 88). The American Psychological Association Division of Psychotherapy last addressed any such standards in 1971.⁵⁷ Thus, it has been at least a generation since the subject of psychotherapy supervisor standards has received any attention at all. Yet the potential value of supervisor standards seems clear: they can provide a consensual guide to what critical areas must be addressed in psychotherapy supervision training and the basic standards of behavior to which psychotherapy supervisors must adhere in practice.

If supervisor standards were important enough to define in 1957 and 1971, they are no less important now. Other professional groups recently have seen fit to establish such standards, and their efforts merit study.^58–61 At a minimum, it could be beneficial to reexamine those earlier APA efforts and update them as needed. Standards are by no means a cure-all for the ills that beset us, but they are one means of bringing some specificity and concreteness to the education of psychotherapy supervisors and the practice of psychotherapy supervision. Specificity and concreteness are now lacking, and the need for that to be corrected is pressing.

Conclusion

The role of psychotherapy supervision is critical to the process of educating and “making” psychotherapists. Supervision has a long, rich history, is well within the tradition of apprenticeship training, and is central to what we as teachers and educators do. Because of the central, critical place of psychotherapy supervision in psychotherapy education, the salience of these 10 needs becomes all the more clear.

Some of the needs identified here will be far more difficult to address than others. For example, recent methodological discussions make it clear that supervision is a complex process that is difficult to research.³⁴ Still, if supervision is to most fruitfully advance, is to ever have a solid empirical foundation, then efforts must be made not only to research it but to research it well—attending to such matters as rigor, reliable and valid measurement, replication and extension, and diversity in those efforts. Furthermore, if the training/practice base of supervision is to be best strengthened, then attention must be given to developing supervision standards and to training supervisors in how to supervise. The 10 needs identified here are indeed pressing—calling for substantive, immediate attention. If these needs are better addressed, it will open up some impressive possibilities for psychotherapy supervision in the next century. Working to realize those possibilities is the charge now before us.