This article was inspired by Weisz, Weiss, and Donenberg’s (1992) look at positive child psychotherapy effects in research suggesting “…a key task for researchers [is]….identifying those proper conditions under which effects of child therapy may be optimized.” The authors suggest controlled methodological experiment changes to enhance generalizeability in effectiveness trials in clinical/service delivery settings. It calls for including methodologies to transfer desirable aspects of efficacy research (e.g., greater independent variable control) into combined efficacy-effectiveness trials. The five broad changes to research/design methods/procedures are summarized in the last two paragraphs below.

Modified and/or extended research methods and procedures are proposed to bridge the current generalizeability gap between original “purist” laboratory efficacy research and the real-world’s “dirtier” effectiveness trials being replicated in clinics and service settings; this paper illuminates a thick line existing between the laboratory research and field-replicated research. In addition to a sample of five research methods/procedural factors that contribute to a lab-real world practice generalizeability gap/chasm (addressed by this paper), Clarke mentions other factors demanding research-design attention: therapist training; monitoring and protocol compliance; multiple therapist roles; participant recruitment methods; professional vs. nonprofessional therapists; measurement technology; participant-therapist-assessor masking to therapy condition; service setting; and participant assignment.

The five research design methods/procedures covered by this paper are summarized according to their real world application problem(a) and the author’s recommended modified research design solution(b). [1] Degree of therapy structure – (a) the problem is implementation: clinicians won’t show fidelity to regimented, manualized treatment scripts as they too often resist “rigid” or uni-modality interventions in favor of being “eclectic” with their clients on a session-by-session basis. (b) Original efficacy study researchers should provide stratified study designs of minimal protocol deviation vs. minor-moderate protocol deviation vs. non-manualized replication. [2] Integrated vs. isolated services – (a) original efficacy studies typically target one disorder, diagnosis, or problem domain; also, they typically provide one intervention under study. This “insularity” is not feasible in community treatment settings, school clinics, etc. (b) Design an integrated intervention regime of varying multiple research-based intervention “tracks” across a variety of researched disorders. [3] Usual care vs. no-treatment or placebo-attention controls – (a) the falsity of these experimental control conditions is apparent since they do not reflect what typically happens to patients who are not provided treatment. Also, efficacy studies may incur higher drop-out rates among enforced no-treatment and waiting-list patients (introducing potential bias). (b) What some studies are doing already: provide randomized “usual care” as the controls (i.e., defined as minimal treatment), which more closely (and ethically) mirrors real world conditions.

[4] Sample representativeness, sample homogeneity vs. real world population heterogeneity – (a) efficacy research subjects may be very different from their community services counterparts—making the results highly suspect. The psychological and social services world is characterized by frequent instances of co-morbidity and patients receiving services from multiple providers, etc. (b) Apply a two-tiered participant recruitment strategy in efficacy studies (Clarke’s “donut model”): (1) one homogeneous sample constitutes the donut hole; another sample (2) constitutes an unselected, heterogeneous, co-morbid donut ring—using a factorial research design. [5] Treatment parameters of dosage, modality, location, implementation – (a & b) possibly the most important issue, this one continues to seek answers to questions put by legislators and policy-makers: Does psychotherapy work? Are patients with ____ better treated in inpatient or outpatient facilities? And at what costs? For how long? In groups or individual therapy, or both? Delivered by professionals or paraprofessionals? Etc. Thus Clarke is a strong proponent of treatment process research, where sub-studies are embedded within larger efficacy studies so as to better inform and represent the users of research findings.

Citation

Clarke, G.N.(1995). Improving the Transition from Basic Efficacy Research to Effectiveness Studies: Methodological Issues and Procedures. Journal of Consulting and Clinical Psychology, 63(5): 718-725