STREAMS (Structural Equation Modeling Made Simple) has been designed with two main purposes in mind:

To be a tool for teaching and learning structural equation modelling (SEM); and

to be a productivity tool for modellers.

These aims are reached in the following ways:

One language and one interface for multiple SEM programs. STREAMS offers a consistent interface to several SEM programs, with generation of model statements, control of program execution, post-processing of program output, and presentation of results in a model viewer. This also makes it easy to switch between programs, to take advantage of their unique strengths.

User-friendly tools for modeling. Models are set up with a simple and powerful language. The user interface is equipped with numerous tools designed to support use of the language, and estimation, presentation, and interpretation of models.

A transparent data interface. STREAMS takes care of the minute details of data specification and setup for all the modeling programs, relieving the user the burden of dealing with program specific rules and conventions for data management.

Models for complex data. Easy specification of large and complex models for multiple groups of cases and multi-level data.

Efficienct estimation. In STREAMS start values are automatically copied from previously estimated models, which usually yields substantial improvements in program execution time, and makes it possible to achieve a solution for models which are otherwise difficult to estimate.

Data management tools. STREAMS prepares data for modeling, including computation of covariance matrices, and takes care of output from modelling programs, such as computed factors scores.

Project management. Efficient functions for management of data and models are included in the system.

STREAMS comes with ample documentation and a large set of example projects.

STREAMS currently supports the following model fitting programs::



EQS (Structural Equation Systems) has been developed by Peter Bentler at the Psychology Department at UCLA and is distributed by Multivariate Software Ltd.. The first versions of the system were developed in the early 1980s, and offered an alternative to  LISREL's matrix based  method of model specifion, namely model specification through specification of equations. Later versions of EQS also offer a path diagram interface, through which the user draws the model as a diagram, which by the program is then translated into equations. EQS also offers a wide variety of advanced estimation techniques, and offers, in particular, techniques useful for dealing with non-normal data,. and for computing robust test-statistics. The program also computes a wide range of goodness-of-fit measures, and other useful statistics.

STREAMS 3.0 supports EQS 6.1 . 

Back to Top 


LISREL (LInear Structural RELations) has been developed by Karl Jöreskog and Dag Sörbom at the Department of Statistics, Uppsala University in Sweden and is distributed by Scientific Software Inc.. The first versions of LISREL appeared in the early 1970s and the contributions made by Jöreskog and Sörbom are regarded as the ground-breaking work in structural equation modeling. During the three decades that LISREL has existed,  numerous improvements and extensions have been made. Thus, in addition to the somewhat complicated but extremely powerful matrix based language of the original program there is now also a simplified, English-like, language (SIMPLIS) as well as a path diagram interface. The program offers a wide range of estimation methods and goodness-of-fit statistics, and it usually estimates the models very efficiently.

STREAMS 3.0 supports LISREL 8.50 to LISREL 8.70 .

Back to Top


Mplus has been developed by Linda and Bengt Muthén at Muthén and Muthén, and the program implements, among other things, advanced SEM procedures developed by Bengt Muthén in his research and the Department of Education and Information Studies at UCLA.. This is a newcomer among SEM programs, being released in 1998, but it builds upon Muthéns LISCOMP program from 1987. Mplus provides all the standard SEM techniques, and it also offers a rich variety of advanced techniques for dealing with more complex data and models, such as non-normal data, ordinal data, two-level data, missing data, weighted data, multiple cohort data. The Mplus program also offers an integration between latent class analysis and ordinary latant variable modeling, thereby extending the modeling techniques into second-generation SEM.

STREAMS 3.0 supports Mplus 1 to Mplus 3. 

Back to Top

STREAMS also supports Amos 4, which program is, however, no longer distributed.

Copyright © 1999-2005  MultivariateWare. All rights reserved.
Revised: 03-04-2006.