Mean and Covariance Structure Analysis: Theoretical and Practical Improvements Export

@article{citeulike:557781, abstract = {The most widely used multivariate statistical models in the social and behavioral sciences involve linear structural relations among observed and latent variables. In practice, these variables are generally nonnormally distributed, and hence classical multivariate analysis, based on multinormal error-free variables having no simultaneous interrelations, is not adequate to deal with such data. Since structural relations among variables imply a structure for the multivariate product moments of...}, author = {Yuan, Ke H. and Bentler, Peter M.}, citeulike-article-id = {557781}, citeulike-linkout-0 = {http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.50.242}, citeulike-linkout-1 = {http://links.jstor.org/sici?sici=0162-1459%28199706%2992%3A438%3C767%3AMACSAT%3E2.0.CO%3B2-O}, comment = {This is a rather rare SEM article published in a mainstream statistical journal. It is thus written specifically for statisticians in mind using their jargon rather than of social scientists. Yuan and Bentler approach SEMs/mean and covariance structure models through a nonlinear regression context with response variables being the observed variables and their second moments, and show how the general nonlinear models theory can be applied here. They also derive scaling O\_p(n^{-1}) corrections for popular asymptotic chi2 fit indices, and show their superior performance in small samples. Cute paper, strongly suggested for SEM theory reading.}, journal = {Journal of the American Statistical Association}, keywords = {adf, asymptotics, bartlett\_correction, latent\_variables, mle, robust\_inference, sem, social\_science\_statistics, statistical\_theory}, number = {438}, pages = {767--774}, posted-at = {2006-03-21 04:30:35}, priority = {0}, title = {Mean and Covariance Structure Analysis: Theoretical and Practical Improvements}, url = {http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.50.242}, volume = {92}, year = {1997} }

TY - JOUR ID - citeulike:557781 L3 - citeulike-article-id:557781 N2 - The most widely used multivariate statistical models in the social and behavioral sciences involve linear structural relations among observed and latent variables. In practice, these variables are generally nonnormally distributed, and hence classical multivariate analysis, based on multinormal error-free variables having no simultaneous interrelations, is not adequate to deal with such data. Since structural relations among variables imply a structure for the multivariate product moments of... IS - 438 JF - Journal of the American Statistical Association EP - 774 TI - Mean and Covariance Structure Analysis: Theoretical and Practical Improvements VL - 92 SP - 767 KW - adf KW - asymptotics KW - bartlett_correction KW - latent_variables KW - mle KW - robust_inference KW - sem KW - social_science_statistics KW - statistical_theory N1 - This is a rather rare SEM article published in a mainstream statistical journal. It is thus written specifically for statisticians in mind using their jargon rather than of social scientists. Yuan and Bentler approach SEMs/mean and covariance structure models through a nonlinear regression context with response variables being the observed variables and their second moments, and show how the general nonlinear models theory can be applied here. They also derive scaling O_p(n^{-1}) corrections for popular asymptotic chi2 fit indices, and show their superior performance in small samples. Cute paper, strongly suggested for SEM theory reading. AU - Yuan, Ke AU - Bentler, Peter PY - 1997/// UR - http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.50.242 ER -