Research Methodology
Recently, Dr. Tasca and several graduate students have developed multilevel statistical models to address the complexities of data analysis for group treatment research. A list of recent publications related to the Research Methodology theme appears below.
- Tasca, G. A., Presniak, M. D., Demidenko, N., Balfour, L., Krysanski, V., Trinneer, A., & Bissada, H. (2011). Testing a maintenance model for eating disorders in a tertiary care treatment seeking sample: A structural equation modeling approach. Comprehensive Psychiatry. Published online, February, 2011.
Fairburn et al (Fairburn, CG, Cooper, Z, Shafran, R. Behav Res Ther 2003;41:509-528) proposed additional maintenance mechanisms (ie, interpersonal difficulties, mood intolerance, low self-esteem, and perfectionism) for some individuals with eating disorders in addition to core eating disorder psychopathology (ie, overevaluation of eating, weight, and shape and their control). This is the first study to both elaborate and test this maintenance model as a structural model. Adults seeking treatment of an eating disorder (N = 1451) at a specialized tertiary care center were included in this cross-sectional study. In the first part of the study, diagnostically heterogeneous participants (n = 406) were randomly selected to test a structural model based on the maintenance model. In the second part of the study, remaining participants (n = 1045) were grouped according to eating disorder diagnosis to test for invariance of the structural paths of the final model across diagnoses. Overall, the structural model with core and additional mechanisms fit the data well and, with 1 exception, represented maintenance processes for each of the diagnostic groups. Treatment models based on both core and additional maintenance factors for those seeking therapy at a specialized tertiary care center may result in improved treatment outcomes for these patients with eating disorders.
- Tasca, G. A., Ramsay, T., Corace, K., Illing, V., Bone, M., Balfour, L., & Bissada, H. (2010). Modeling longitudinal data from a rolling therapy group program with membership turnover: Does group culture affect individual alliance? Group Dynamics, 14, 151-162.
Many community- and hospital-based group treatment programs have an open enrolment, that is, a rolling admissions structure, in which a group member who drops out or successfully completes therapy is replaced by another individual. Although practically efficient and perhaps clinically useful, the interdependence of these group participants' data may result in incorrect inferences drawn from the data analyses if this interdependence is not accounted for. We present an analytic strategy that uses time varying covariates in multilevel models to illustrate a methodology to address these data analysis problems. Participants were adults with eating disorders (N _ 229) who attended an average of 12 weeks of a rolling admissions group-based day hospital program during an 8-year period, and who completed a group therapy alliance measure weekly. Individual alliance to the group increased from week to week, and this growth remained significant even after controlling for the time varying level of other group members' alliance to the group. Further, the level of an individual's alliance score during any given week was positively related to the group's alliance during that week. The multilevel time varying covariate models presented here add to a very small but emerging set of analytic strategies available for researchers to address some of the hurdles to correctly analyze data from rolling admissions group-based treatment programs. Results from this study provide evidence that a group's culture is passed on and affects an individual's alliance to the group despite changes in group membership.
- Tasca, G.A., Ramsay, T., Corace, K., Illing, V., Bone, M., Balfour, L., & Bissada, H. (2010). Modeling longitudinal data from a rolling therapy group program with membership turnover: Does group culture affect individual alliance? Group Dynamics, 14, 151-162.
Title: Modeling Longitudinal Data from a Rolling Therapy Group Program with Membership Turnover: Does Group Culture Affect Individual Alliance?
Authors: Tasca, G.A., Ramsay, T., Corace, K., Illing, V., Bone, M., Balfour, L., & Bissada, H.
Journal Title: Group Dynamics: Theory, Research, and Practice.
Publication Date: 2010
Abstract: Many community and hospital-based group treatment programs have an open enrolment, i.e., a rolling admissions structure, in which a group member who drops out or successfully completes therapy is replaced by another individual. Although practically efficient and perhaps clinically useful, the interdependence of these group participants’ data may result in incorrect inferences drawn from the data analyses if this interdependence is not accounted for. We present an analytic strategy that uses time varying covariates in multilevel models to illustrate a methodology to address these data analysis problems. Participants were eating disordered adults (N = 229) who attended an average of 12 weeks of a rolling admissions group-based day hospital program during an 8 year period, and who completed a group therapy alliance measure weekly. Individual alliance to the group increased from week to week, and this growth remained significant even after controlling for the time varying level of other group members’ alliance to the group. Further, the level of an individual’s alliance score during any given week was positively related to the group’s alliance during that week. The multilevel time varying covariate models presented here add to a very small but emerging set of analytic strategies available for researchers to address some of the hurdles to correctly analyse data from rolling admissions group-based treatment programs. Results from this study provide evidence that a group’s culture is passed on and affects an individual’s alliance to the group despite changes in group membership. - Tasca, G.A., Illing, V.A., Ogrodniczuk, J., & Joyce, A. (2009). Assessing and adjusting for dependent observations in group treatment research using multilevel models. Group Dynamics, 13, 151-162.
Title: Assessing and adjusting for dependent observations in group treatment research using multilevel models.
Authors: Tasca, Giorgio A.; Illing, Vanessa; Ogrodniczuk, John S.; Joyce, Anthony S.
Journal Title: Group Dynamics: Theory, Research, and Practice. Vol 13(3), Sep 2009, 151-162.
Publication Date: 2009
Abstract: Group treatment data are nested by design, that is, clients nested in groups. Dependence associated with the nesting of group intervention data can inflate Type I error rates, which poses unique challenges to group treatment researchers. This article evaluates the extent and variability of dependence in data taken from 3 previously published randomized clinical trials of group psychotherapy. Three methods of assessing dependence by calculating intraclass correlation coefficients (?) were examined. Results showed great variability in ?s across studies, across methods of calculating ?, and across outcome variables. The distribution of ?s suggested that the amount of dependence in the data was moderate. Two methods of addressing dependence in grouped treatment data through multilevel modeling were used. These methods resulted in minimally compromised statistical power compared with results from uncorrected data. These 2 methods may allow researchers to reliably assess their group treatments. Group intervention researchers are encouraged to consider their assumptions and conceptualizations of treatment change, and to choose corresponding methods of assessing for and addressing. (PsycINFO Database Record (c) 2009 APA, all rights reserved). - Tasca, G.A., Illing, V.A., Joyce, A., & Ogrodniczuk, J. (2009). Multilevel models for nested change data: A guide for group treatment researchers. Psychotherapy Research, 19, 453-461.
Article Title: Multilevel models for nested change data: A guide for group treatment researchers.
Authors: Tasca, G.A., Illing, V.A., Joyce, A., & Ogrodniczuk, J.
Journal Title: Psychotherapy Research, 19, 453-461.
Date of Publication: 2009
Abstract:
Researchers have known for years about the negative impact on Type I error rates caused by dependencies in hierarchically nested and longitudinal data. Despite this, group treatment researchers do not consistently use methods such as multilevel models (MLMs) to assess dependence and appropriately analyse their nested data. The goals of this paper are: to review some of the study design issues with regard to hierarchically nested and longitudinal data, to discuss MLMs for assessing and handling dependence in data, and to present a guide for developing a three-level growth MLM that is appropriate for group treatment data, design, and research questions. We present an example from group treatment research to illustrate these issues and methods. - Tasca, G.A. & Gallop, R. (2009). Multilevel modeling of longitudinal data for psychotherapy researchers: I. The basics.
Psychotherapy Research, 19, 429-437.
Article Title: Multilevel modeling of longitudinal data for psychotherapy researchers: I. The basics
Authors: Tasca GA;Gallop R;
Journal: Psychother Res Volume 19, 429-437.
Date of Publication:2009
Abstract:
Psychotherapy researchers are often interested in change or development over time (i.e., pre- to posttreatment to follow-up or the development of process variables across multiple sessions). Traditional methods of assessing change and development are often unsatisfactory because of violations of statistical assumptions and because they do not model individual change. Modern longitudinal data analysis methods, including multilevel models (MLMs), provide an opportunity to model dynamic fluctuations in individual data across time. The objective of this article is to focus on the fundamentals of MLMs for longitudinal data analysis in psychotherapy research. To do so, the authors illustrate basic equations of MLMs and a strategy for developing increasingly complex models. They also present data from a psychotherapy research as an example of the application of MLMs. Finally, they offer some caveats and advice for conducting and presenting MLMs. - Gallop, R., & Tasca, G.A. (2009). Multilevel modeling of longitudinal data for psychotherapy researchers: II. The complexities. Psychotherapy Research, 19, 438-452.
Title: Multilevel modeling of longitudinal data for psychotherapy researchers: II. The complexities.
Authors: Gallop, R.; Tasca, G.A.
Journal Title: Psychotherapy Research, 19, 438-452.
Publication Date: 2009
Abstract:
We previously reviewed the basic elements and steps to building multilevel models (MLM) for longitudinal data that is typically found in psychotherapy research. The objective of this paper is to focus on complexities associated with the MLM for longitudinal data analysis in psychotherapy research, which may result in proper use or misuse of the modeling structure. To do so, we illustrate complex scenarios and discuss issues in the implementation and interpretation of the MLM. Issues we address are (a) impact of missing data in the MLM, (b) determination of the complexity of the covariance structure and its implication on model interpretation, (c) issues with centering, (d) model diagnostics for MLM, (e) model formation including implementation dependent on the treatment of time and distribution of the outcome, and (f) model estimation. We also present data from psychotherapy research settings as examples of these complex situations. Finally we offer some caveats and advice for recognizing these complexities and proper procession to ensure accurate implementation of the MLM and the interpretation of the results. - Tasca, G.A., Balfour, L., Ritchie, K., & Bissada, H. (2006). Developmental changes in group climate in two types of group therapy for Binge Eating Disorder: A growth curve analysis. Psychotherapy Research, 16, 499-514
Article Title: Developmental changes in group climate in two types of group therapy for binge-eating disorder: A growth curve analysis
Authors: Tasca GA; Balfour L; Ritchie K; Bissada H;
Journal: Psychotherapy Research Volume 16 , Issue4
Date of Publication: 2006
Abstract:
The development of group climate across 16 sessions of group psychodynamic-interpersonal psychotherapy (GPIP) and group cognitive-behavioral therapy (GCBT) for 65 female treatment completers with binge-eating disorder (BED) was assessed. Engaged scale growth for GPIP patients varied across sessions and was best represented by a cubic growth curve. This suggested that GPIP progressed in definable phases that reflected a rupture and repair sequence of engaged group climate. For patients receiving GCBT, engaged, avoiding, and conflict scale growth was gradual and consistent (i.e., linear), indicating an increase in positive group climate across sessions. This likely reflected patients taking greater responsibility for treatment as suggested by the CBT model. Linear growth in engaged climate mediated the relationship between attachment anxiety and outcome in GPIP. A consistent increase in engaged group climate through the rupture and repair phase may be a necessary condition for successful treatment of BED patients with high attachment anxiety who receive GPIP. © 2006 Society for Psychotherapy Research.