This is a revised submission of a request for the competing continuation of a K05 Senior Scientist Award.
The aims of the proposed five-year research program are primarily theoretical, namely to develop new psychometric methods for drug abuse research;apply these newly developed methods to important drug abuse data sets;and do science education on these topics. This proposal describes a research program on the development and application of new multivariate psychometric and statistical techniques that hold special promise for improving the quality of psychosocial, social, behavioral, epidemiological, and functional imaging health-related research on drug use and abuse. Standard multivariate methods may not be applicable to nonexperimental drug abuse data, and to a lesser extent experimental data, because these may contain selectivity, outlier cases, arbitrary distributions, errors in measurement, hierarchical sampling plans, small samples, dependent observations, and related problems. This proposal aims to develop new methods that can enable researchers to reach reliable scientific conclusions with such problematic data. Special attention is given to structural equation models that may involve latent variables in mediation and moderator relations, and the development of methods for using item response theory measurement models with latent variable structural relations. New methods that are developed will be evaluated for applicability to real data with simulation research and will be applied to existing data sets. No data will be gathered under this grant, but applications to important substantive issues in drug abuse will be sought in association with NIDA/UCLA program project grant P01 DA01070, via collaboration with nationally active and UCLA-based substance abuse researchers working on their own projects, and by cooperation with NIDA and other health-related agencies. Methods that are developed under this grant also will be widely disseminated.
|Bentler, Peter M; Huang, Wenjing (2014) On Components, Latent Variables, PLS and Simple Methods: Reactions to Rigdon's Rethinking of PLS. Long Range Plann 47:138-145|
|An, Xinming; Bentler, Peter M (2013) Nesting Monte Carlo EM for high-dimensional item factor analysis. J Stat Comput Simul 83:25-36|
|Tong, Xiaoxiao; Bentler, Peter M (2013) EVALUATION OF A NEW MEAN SCALED AND MOMENT ADJUSTED TEST STATISTIC FOR SEM. Struct Equ Modeling 20:148-156|
|An, Xinming; Yang, Qing; Bentler, Peter M (2013) A latent factor linear mixed model for high-dimensional longitudinal data analysis. Stat Med :|
|Wu, Jianmin; Bentler, Peter M (2013) Limited Information Estimation in Binary Factor Analysis: A Review and Extension. Comput Stat Data Anal 57:392-403|
|Bentler, Peter M; Yuan, Ke-Hai (2011) Positive Definiteness via Off-Diagonal Scaling of a Symmetric Indefinite Matrix. Psychometrika 76:119-123|
|Li, Libo; Bentler, Peter M (2011) Quantified choice of root-mean-square errors of approximation for evaluation and power analysis of small differences between structural equation models. Psychol Methods 16:116-26|
|Jennrich, Robert I; Bentler, Peter M (2011) Exploratory Bi-factor Analysis. Psychometrika 76:537-549|
|Bentler, Peter M; Liang, Jiajuan; Tang, Man-Lai et al. (2011) Constrained Maximum Likelihood Estimation for Two-level Mean and Covariance Structure Models. Educ Psychol Meas 71:325-345|
|An, Xinming; Bentler, Peter M (2011) Extended mixture factor analysis model with covariates for mixed binary and continuous responses. Stat Med 30:2634-47|
Showing the most recent 10 out of 18 publications