Author Description

<i>LISREL 8</i> offers an impressive array of facilities for data analysis including indirect and total effects and their standard errors; direct specification of mean parameters; a Ridge Option for handling covariance and correlation matrices that are not positive-definitive; and modification indices for all iterative estimation methods. Advancements of special interest include: <br> <br> * the new command language -- <i>SIMPLIS</i> -- that is easy to use and minimizes the possibility of mistakes in the problem set up; <br> * high quality path diagrams that can be produced on the screen and printed; <br> * many new measures that judge the fit of the model; <br> * linear and nonlinear constraints that may be imposed on any set of model parameters; <br> * general covariance structures that allow formulation of models that cannot be specified as <i>LISREL</i> models; and <br> * interval restrictions that may be imposed on any single parameter. <br> <br> Advanced users will gain new insights into the use of asymptotically distribution-free (ADF) analysis, as well as power calculations for tests of model augmentation, automatic model modification, and other program features. <br> <br> Included with <i>LISREL 8</i> is <i>PRELIS 2</i>, a preprocessor for <i>LISREL</i> which greatly improves and accelerates analysis of binary, categorical, ordinal, censored, continuous, and/or incomplete data. New features of <i>PRELIS 2</i> include: <br> <br> * missing values on a variable that may be inputed by matching on other variables; <br> * new variables that may be created as functions of other variables; <br> * tests of univariate and multivariate normality that are obtained for all continuous variables; <br> * bootstrap estimates that may be obtained of the covariance matrix of correlations and other associations measures for which no asymptotic theory is available; <br> * univariate and multivariate multinomial probit models that may be estimated for any set of jointly dependen