Since the 1970s, SSI has developed and produced many programs with lasting scientific impact.
The company’s first important product was MULTIVARIANCE, by Professor Jeremy Finn, State University of New York at Buffalo. The program performs univariate and multivariate analysis of variance for any design up to ten factors, and gives exact solutions in both unbalanced and balanced designs. It includes analysis of covariance, discriminant analysis, canonical analysis, repeated measures analysis, and Potthoff-Roy analysis of longitudinal data, all of which were at the cutting edge of the time. The program was leased to several hundred mainframe sites during the 1970s.
The company’s popular structural equation modeling program LISREL of Professors Karl Jöreskog and Dag Sörbom, University of Uppsala, have helped define the field of structural equation modeling. They remain the most advanced programs available for analysis of large linear systems when both dependent and independent variables are subject to error, and/or when qualitative variables are included in the analysis.
Another important SSI program is HLM – a program for the analysis of complex study designs under the hierarchical linear model, by Professors Stephen Raudenbush and Anthony Bryk, University of Chicago, and Richard Congdon. The popularity of HLM and the methodological approaches that it embodies have made last impact on education sciences.
Last but not the least, a main area of SSI has been item analysis and test scoring, which is important in educational and psychological measurement. The earliest of these programs is BILOG, written by Robert Mislevy, University of Maryland, and Darrell Bock, University of Chicago. The program is the first true enabler of psychometric analysis of educational tests using modern item response theoretic methods. Its success led to the development of BILOG-MG, by Michele Zimowski (National Opinion Research Center), Eiji Muraki (Tohoku University), Robert Mislevy (University of Maryland), and Darrell Bock (University of Chicago), which remains available to this day.
Another important legacy program is MULTILOG, by Professor David Thissen, University of North Carolina at Chapel Hill. The program popularized many multiple-categorical item response models. Multilog presaged many features in PARSCALE, a program written by Eiji Muraki, Tohoku University, and Darrell Bock, University of Chicago. PARSCALE serves similar purposes for large-scale analysis and remains popular today.
The final program in this area is TESTFACT, by Douglas Wilson and Robert Wood, University of London Examinations Council, and Professor Robert Gibbons, University of Illinois, Chicago. This program performs classical test scoring, item analysis, and item factor analysis. TESTFACT has had major influence on the design and computational approaches of modern programs such as IRTPRO and flexMIRT®.