Paper presented at the American Evaluation Association 2010 conference

Slides:



Advertisements
Similar presentations
MCUAAAR: Methods & Measurement Core Workshop: Structural Equation Models for Longitudinal Analysis of Health Disparities Data April 11th, :00 to.
Advertisements

Latent Growth Modeling Chongming Yang Research Support Center FHSS College.
1 Multilevel Mediation Overview -Mediation -Multilevel data as a nuisance and an opportunity -Mediation in Multilevel Models -
Behavioral Intention and Partner Type on Condom Use Among Men in Drug Treatment Yong S. Song, PhD, University of California, San Francisco Donald Calsyn,
Growth Curve Model Using SEM
15 de Abril de A Meta-Analysis is a review in which bias has been reduced by the systematic identification, appraisal, synthesis and statistical.
1 WELL-BEING AND ADJUSTMENT OF SPONSORED AGING IMMIGRANTS Shireen Surood, PhD Supervisor, Research & Evaluation Information & Evaluation Services Addiction.
“Ghost Chasing”: Demystifying Latent Variables and SEM
Problem Identification
Change Score Analysis David A. Kenny December 15, 2013.
Structural Equation Modeling Intro to SEM Psy 524 Ainsworth.
Mediation: Sensitivity Analysis David A. Kenny. 2 You Should Know Assumptions Detailed Example Solutions to Assumption Violation.
Factor Analysis Psy 524 Ainsworth.
Hypothesis Construction Claude Oscar Monet: The Blue House in Zaandam, 1871.
Multiple Sample Models James G. Anderson, Ph.D. Purdue University.
Nursing Care Makes A Difference The Application of Omaha Documentation System on Clients with Mental Illness.
Statistics and Quantitative Analysis U4320 Segment 12: Extension of Multiple Regression Analysis Prof. Sharyn O’Halloran.
Correlational Research Chapter Fifteen Bring Schraw et al.
Growth Mixture Modeling of Longitudinal Data David Huang, Dr.P.H., M.P.H. UCLA, Integrated Substance Abuse Program.
Research PHE 498. Define Research Research can be considered as systematic inquiry: A process that needs to be followed systematically to derive conclusions.
Biostatistics Case Studies 2008 Peter D. Christenson Biostatistician Session 5: Choices for Longitudinal Data Analysis.
2008 FAEIS Annual Longitudinal Assessment With a Comparison to the 2007 Survey Results The purpose of the FAEIS annual evaluation is to develop longitudinal.
Measurement Models: Exploratory and Confirmatory Factor Analysis James G. Anderson, Ph.D. Purdue University.
Program Evaluation Dr. Ruth Buzi Mrs. Nettie Johnson Baylor College of Medicine Teen Health Clinic.
CJT 765: Structural Equation Modeling Class 12: Wrap Up: Latent Growth Models, Pitfalls, Critique and Future Directions for SEM.
Controlling for Baseline
Roghayeh parsaee  These approaches assume that the study sample arises from a homogeneous population  focus is on relationships among variables 
Personally Important Posttraumatic Growth as a Predictor of Self-Esteem in Adolescents Leah McDiarmid, Kanako Taku Ph.D., & Aundreah Walenski Presented.
Public Finance and Public Policy Jonathan Gruber Third Edition Copyright © 2010 Worth Publishers 1 of 24 Copyright © 2010 Worth Publishers.
CJT 765: Structural Equation Modeling Final Lecture: Multiple-Group Models, a Word about Latent Growth Models, Pitfalls, Critique and Future Directions.
Stephen Nkansah-Amankra, PhD, MPH, MA 1, Abdoulaye Diedhiou, MD, PHD, H.L.K. Agbanu, MPhil, Curtis Harrod, MPH, Ashish Dhawan, MD, MSPH 1 University of.
Data analysis using regression modeling: visual display and setup of simple and complex statistical models 1 CIRA Center for Interdisciplinary Research.
Crystal Reinhart, PhD & Beth Welbes, MSPH Center for Prevention Research and Development, University of Illinois at Urbana-Champaign Social Norms Theory.
Paper presented at the American Evaluation Association 2010 conference PLEASE do not distribute and cite only with permission; work is under journal review.
Dynamic Factor Models: A Tool for Analyzing Longitudinal Public Health and Medical Databases Presentation for the 135 th Annual Program Meeting of the.
Statistics & Evidence-Based Practice
Tatiana Jan Blair Holly.
Scott Elliot, SEG Measurement Gerry Bogatz, MarketingWorks
Knowledge, attitude, and practices and influencing factors related to seasonal influenza vaccination among health-care workers in Qingdao, China, 2013–14:
Structural Equation Modeling using MPlus
IMPACT OF A PEER-GROUP INTERVENTION ON URBAN HEALTH WORKERS IN MALAWI
CJT 765: Structural Equation Modeling
Emil Coman, Ph.D. 1 & Peg Weeks, Ph.D. 2
Linear Regression CSC 600: Data Mining Class 13.
Learning Objectives For models with dichotomous intendant variables, you will learn: Basic terminology from ANOVA framework How to identify main effects,
This research was supported by NIAAA K01AA
Statistical Data Analysis
Statistical Models for the Analysis of Single-Case Intervention Data
HLM with Educational Large-Scale Assessment Data: Restrictions on Inferences due to Limited Sample Sizes Sabine Meinck International Association.
Shudong Wang NWEA Liru Zhang Delaware Department of Education
Interactive Models: Two Quantitative Variables
Class 8 Spring 2017 Ang &Huan (2006)
Propensity Score Matching Makes Program Evaluation Easy
Introduction to Statistics
Inferences and Conclusions from Data
Shoo T, Kamala B, Rosecrans K, Miller K, Al-Alawy H, Rwezahura P
Empirical Tools of Public Finance
Measuring Change in Two-Wave Studies
Hypothesis Construction
Why use marginal model when I can use a multi-level model?
JAMA Ophthalmology Journal Club Slides: Longitudinal Associations Between Visual Impairment and Cognitive Functioning Zheng DD, Swenor BK, Christ SL, West.
The Process of Advertising Research
Day 2 Applications of Growth Curve Models June 28 & 29, 2018
Special Topic: Longitudinal Mediation Analysis
Group Experimental Design
Rachael Bedford Mplus: Longitudinal Analysis Workshop 23/06/2015
Autoregressive and Growth Curve Models
Types of questions TVEM can answer
Structural Equation Modeling
PrEP Use and STIs are not Associated Longitudinally in a Cohort Study of YMSM/TGW in Chicago Ethan Morgan, PhD Institute for Sexual and Gender Minority.
Presentation transcript:

Paper presented at the American Evaluation Association 2010 conference Evaluating potential impact of intervention in community settings when no comparison data is available: mixture latent growth modeling for exploring differential change in female condom use Paper presented at the American Evaluation Association 2010 conference Maryann Abbot1, MA; Emil Coman2A Ph.D., & Peg Weeks3 Ph.D. 1: Project Manager; 2: Senior Research Scientist-Evaluator,& 3: Executive Director Institute for Community Research, Hartford, CT USA A: acknowledged indirect effect from Dave Kenny, his SEM mentor. DISCLAIMER: All error found here were totally mediated by 2nd author!

Statement of problem Evaluation of intervention ‘effects’ in one-group pre-post test design (no controls) is challenging Structural equation modeling (SEM) or covariance structure analysis (CSA) with latent grouping variable (mixture ‘regression’1, 2) is an option The question of causality cannot be answered definitely however BUT we can describe ‘what happened’ and try to understand why Jo, B., & Muthén, B. (2000). Intervention studies with noncompliance: Complier average causal effect estimation in growth mixture modeling. In N. Duan & S. Reise (Eds.), Multilevel Modeling: Methodological advances, issues, and applications (pp. 112-139). Mahwah, NJ: Lawrence Erlbaum Associates. Tarpey, T., & Petkova, E. (2010). Latent regression analysis. Statistical Modelling, 10(2), 133.

‘what happened/does it seem to work’? The challenge Lack of a comparable group change data Few studies ongoing on Female Condom (FC) use Host of historical/context effects Primarily new marketing campaign recently launched for FC 2 Other HIV prevention local interventions, not necessarily focused on FC however We want answer to the question: ‘what happened/does it seem to work’? i.e. has the intervention changed FC use and attitudes in any of the participants; who are those who have not changed, what seem to make the difference?

The Hartford FC study Intervention conducted in Hartford, CT aimed at increasing awareness and use of the female condom (FC) as a women-initiated HIV and sexually transmitted infections (STI) prevention method Prospective, longitudinal cohort study of 461 women (baseline, 1 & 10 months) We provided a brief knowledge/skills demonstration to all women at baseline and wave 2: included demonstrating proper FC use (using their hands as a model to show proper insertion and removal) explaining common problems with FC insertion/use and possible solutions answered questions

The Hartford FC study Designed to to reduce the primary initial barriers to FC adoption, namely, the lack of information, demonstration, community availability, and troubleshooting support for FC use Baseline (wave 1: n = 461), wave 2 (n = 392), & wave 3 (n = 245); common sample across time points of 228 participants 55% African/Black American, 30% Puerto Rican/Other Latino, 14% White/Other 71%, 22%, and 20% at waves 1, 2, and 3 respectively have not used FC in the past 30 days

SEM reminder Models are specified by replacing correlations with causal paths (one-directional), or deleting them (forcing them equal to zero). For mixture analyses - specify a latent categorical variable: can be both cause and effect of other variables, or of certain parameters (like slopes of growth curves). MODEL: %OVERALL% i s | fctimes1@0 fctimes2@1 fctimes3@2 ; Fit for mixture models is relative, based on AIC, BIC, entropy.

Methods Mixture growth modeling1 method of identifying latent classes (unobserved groupings) in the sample analyzed Purpose: to identify latent classes of potential and actual FC users who exhibit different growth trajectories in terms of their past 30 days FC (female condom) use Similar to say Segawa et al. (2005) but with no control group Li, F., Duncan, T. E., Duncan, S. C., & Acock, A. (2001). Latent growth modeling of longitudinal data: A finite growth mixture modeling approach. Structural Equation Modeling: A Multidisciplinary Journal, 8(4), 493-530. Segawa, E., Ngwe, J., Li, Y., & Flay, B. (2005). Aban Aya Coinvestigators (2005). Evaluation of the effects of the Aban Aya Youth Project in reducing violence among African American Adolescent males using latent class growth mixture modeling techniques. Evaluation review, 29(2), 128-148.

Growth Mixture Latent Classes Intercept Slope Times FC use 1 2 1 Latent Classes Intercept Slope ?

Overall results FC use - percentages

Overall results FC use - percentages

SEM results Based on classification entropy and AIC and BIC fit indices, the best solution was deemed a three latent classes grouping for the no-covariate times used FC GMM mixture model It divided up the sample into 109 fast increasing use participants, 93 moderate increase Ps, and 26 consistent (stable) Ps. The intervention seems to have been very successful for 48% of the sample, somewhat successful for another 41%, while not impacting the rest of 11% participants.

More results They had slopes of .399 (p<.001), .103 (p=.021), and .008 (p=.831). The intercepts for the groups were -1.145, 1.375, and 2.122. The groups differ significantly in terms of Times used FC at waves 2 and 3 FC stage of use at all three waves FC beliefs at all three waves FC knowledge at wave 3

Growth Mixed Modeling results

Mixture GLM results ‘Stage of FC use 1 = never thought about using; 2 = thought about or got but didn’t use ; 3 = tried but decided not to use again ; 4 = use FC occasionally 5 = use FC regularly or as primary

Mixture GLM results FC beliefs 4= strongly favorable ; 3= favorable; 2= unfavorable ; 1= strongly unfavorable

Mixture GLM results FC knowledge Average knowledge score 0 = incorrect; 1 = correct

Additional problems Mixture modeling with vs. without covariates have advantages/disadvantages; covariates may separation between classes1. Growth models have alternative specifications possible: correlation between slope and intercept growth parameters zero/not. Slopes set to common 0,2,3, etc may need to follow real time lags (baseline, 1 month, 10 month interview) “Recovered’ classes may not always represent meaningful groups 1. Lubke, G. (2010). Latent Variable Mixture Models. In Hancock, G., & Mueller, R. The Reviewer's Guide to Quantitative Methods in the Social Sciences: Routledge.(p. 209-219).

Conclusions There seems to be in our sample of female participants three distinct groups of actual and potential FC users. Their FC use, FC beliefs, FC knowledge, and stage of FC use differ, so we might conclude these are distinct groups that need to targeted by differential (tailored) intervention strategies for FC adoption/increased use. Additional explorations should be done on other demographic but more importantly relationship variables.

Thank you! http://evaluationhelp.wordpress.com www.icrweb.com ---