Open Door Review

=<<$2*70!$*!$*&$Y$&B2

H)%Y2(#_!=>!N>q!O0!8>q!Lq!G!Q49)*&6_!O>![+,--^>!=<<$2*70!$*!$*&$Y$&B2! '(A?-",-%#@:A&le [-^?!\E-g>! H)%Y2(#_!=>!N>_!G!F0&$_!5>!/>![+,,+^>!"#0!2<<$2*70>!;*!i>!8>!a)%7%)66!['&>^_! '(A?-",-%#@:A&3%B@,)"+(-):(&,-@,&E"#a& [11>!PPEd,^ C& a0Z!e)%M?!Nc:)%&!R*$Y0%6$(4!/%066>! >-001(3 ! The concept of the alliance is currently one of the most intensely researched subjects in the psychotherapy research literature. This new research synthesis examines the relation between alliance and the outcomes of individual Psychotherapy, including over 200 research reports based on 190 independent data sources, covering more than 14,000 treatments. Research involving 5 or more adult participants receiving genuine (as opposed to analogue) treatments, where the author(s) referred to one of the independent variables as “alliance,” “therapeutic alliance,” “helping alliance,” or “working alliance” were the inclusion criteria. Several potential moderators also were explored. In terms of the research synthesis presented in this report, it is important to emphasize that the authors know about the lack of a precise consensual definition of the alliance construct. As a consequence, the alliance and its relation to outcome and other therapy variables has been gleaned from studies which, in practice, define the alliance by the diverse instruments used to measure it. Within the 201 studies in this collection of data, over 30 different alliance measures were used—not counting different versions of the same instruments. Similar to previous reports, the four “core measures”: California Psycho- therapy Alliance Scale, (CALPAS), Helping Alliance Questionnaires (HAq), Vanderbilt Psycho- therapy Process Scale (VPPS), and Working Alliance Inventory (WAI) accounted for approximately 2/3 of the data. In research on the shared factor structure of the WAI, CALPAS and HAq, the concept of “confident collaborative relationship” was identified as the central common theme. Each of these four instruments has been in use for over 20 years and has demonstrated an acceptable level of internal consistency. Fifty-four of the research reports in this data set used less well validated instruments or assessment procedures; the relation of most of these measures to the core instruments, or to each other, are not well documented, and sometimes nonexistent. As noted, the diversity in the “de facto” definition of the alliance that has emerged via the use of a variety of assessment measures has become an important source of variability across studies. For identifying studies published between 1973 and 2000, the authors relied on data from previous analyses (Horvath & Symonds, 1991 & Horvath & Bedi, 2002) but the effect sizes (ES) where recalculated (using more up-to-date methods) for all but 10 of the oldest unpublished studies which were no longer available. To locate data from the years 2000 to 2009, first electronic databases were searched (PsycINFO/EBSCO) using the same keywords as the Horvath and Bedi (2002) analysis. Next the bibliography of studies included in the analysis was cross-referenced. The criteria for inclusion in this report were: (1) the study author referred to the therapy process variable as “ alliance” (including variants of the term); (2) the research was based on clinical as opposed to analogue data; (3) five or more adult patients participated in the study, and; (4) the data reported were such that we could extract or estimate a value indicating the relation between alliance and outcome. In contrast to previous meta-analyses, the literature search was extended to material available in Italian, German, or French, as well as English. The number of studies in the current study is roughly double the size of the data available for the previous (Horvath & Bedi, 2002) meta-analysis. The analyses of this research synthesis were done using the assumptions of a random model and numerical estimates were calculated using restricted maximum likelihood (random effects) model. The aggregate effect size (ES), for the 190 independent alliance/outcome relations was r _ .275. The 95% confidence interval of this averaged ES ranged from .249 to .301. The magnitude of the relationship

CLK

.01230/1.40/5&&'67894/0/571.8/5&&/6648./1.40&

Made with FlippingBook Ebook Creator