Author Description

Building on the results of a decade of research in statistics, computer science, and artificial intelligence, the <i>TETRAD II</i> program provides a powerful, flexible aid for specifying linear structural equation models and for constructing and using Bayesian networks -- causal models of discrete valued variables. <i>TETRAD II</i> employs algorithms based on a novel theory of causal inference that can help with the following kinds of problems:<br> <br> * Selecting regressor variables that directly influence an outcome variable and reducing the number of regressors needed for prediction; <br> * Finding alternatives to a given model that explain the data equally well or better; <br> * Detecting the existence of latent variables; <br> * Finding structural equation models with latent variables; <br> * Locating unidimensional measurement models; <br> * Respecifying a given linear structural equation model; <br> * Writing <i>LISREL, EQS</i>, and <i>CALIS</i> input files; <br> * Building a Bayesian network; <br> * Using a Bayesian network to classify and predict; and <br> * Generating Monte Carlo data from causal models. <br><br> Utilizing graph analysis algorithms rather than numerical techniques, <i>TETRAD II</i> relies on heuristic search techniques characteristic of many artificial intelligence programs. The manual accompanying the program disks is organized into two parts. The first describes information common to the entire enterprise, and the second contains detailed descriptions of each of the nine modules. The chapters include information which:<br> <br> * describes what the program does in general terms, gives an overview of the modules, and describes several common research contexts in which TETRAD II can help;<br> * provides an abridged glimpse at the theoretical foundation upon which TETRAD II is built;<br> * describes how to operate the program, and it documents how to enter and exit the program, and gives several examples of how to use the procedures; and