Began assistantship with North Carolina State University Computing Center. I was assigned to work with the Statistics Department.
Created general analysis of variance program controlled by analysis of variance language similar to the notation of Kendal. Program was written on IBM 1410 assembler. Dr. A. Grandage, author of IBM 650 analysis of variance programs, advised on Analysis of Variance calculations. "Statistical programs for the IBM 650-Part I, Communications of the ACM, Volume 2, Issue 8"
June - Aug 1963
Summer fellowship in Physical Oceanography, Woods Hole Oceanographic Institute
Sept 1963 - May 1964
Resumed assistantship with North Carolina State Computing Center. Wrote multiple regression program with a compiler that generated machine code for transforming data. Dr. A. Grandage advised on the Doolittle procedure for inverting matrices.
June 1964 - May 1966
Employed with IBM Federal Systems Division at the Pentagon, Washington. DC.
I was assigned to work with the National Military Command Center, the information processing branch of the Joint Chiefs of Staff.
Project: Rewrite and enhance the Formatted File System, a generalized data based management system for retrieval and report writing.
Implemented three of the five major components: retrieval, sorting, and file update.
Innovated the idea of a uniform Lexical Analyzer for all languages in the system with a uniform method of handling all error messages within the system.
With the experience in this environment, I saw the power of the self-defining file for providing overall structure to the information processing world.
It became obvious that I could put statistical procedures in the same formatted file framework. At the same time, manuals for PL/1 appeared in the IBM library. The Lexical design of PL/l was an improvement over that used in the Formatted File System.
I was recruited by North Carolina State University Statistics Department to rewrite analysis of variance and regression programs for the IBM 360.
I saw this as an opportunity to develop the Statistical Analysis System (SAS).
I wrote the analysis of variance program while independently developing the SAS software for inputting and transforming data.
Presented conceptual ideas of SAS to members of the Committee on Statistical Software of the University Statisticians of Southeast Experiment Station (USSERS). The meeting was held in Athens, GA. Individuals present:
Frank Verlinden, North Carolina State University
Anthony J. Barr, North Carolina State University
Walt Drapula, Mississippi State University
Jim Fortson, University of Georgia
Jim Goodnight and I cooperated in putting his regression program into SAS.
This procedure was invaluable to pharmaceutical and agricultural scientists in analysis of experiments with missing data.
Developed language for describing regression and analysis of variance model, and preprocessor for creating dummy variables
Developed regression and statistical routines that made practical the analysis of variance methodology within the regression framework
Release of 1972 version of SAS. This was the first release to achieve wide distribution. SAS was now recognized as a major system in statistical computing.
Credits for SAS 72 as described in SAS 76 Users Guide:
Anthony J. Barr
Language translator; data management and supervisor; ANOVA, DUNCAN, FACTOR, GUTTMAN, INBREED, LATTICE, NESTED, PLAN, PRINT, RANK, SORT, SPEARMAN
James H. Goodnight
CANCORR, CORR, DISCRIM, MEANS, PLOT, PROBIT, REGR, RSQUARE, RQUE, STANDARD, STEPWISE.
Jolayne W. Service
"A User's Guide to the Statistical Analysis System"
Carroll G. Perkins
HARVEY, HIST, PRTPCH: A Guide to the Supplementary Procedures Library for the Statistical Analysis System
37,000 total lines of code with distribution:
I had developed and implemented the language, data management, and interface to operating system.
June 1973 - May 1976
I rewrote the internals of SAS: Data Management, report writing and the compiler.
John Sall joined us in 1973 (approx.).
Release of 1976 version of SAS.
The 76 version was a functionally complete system for statistical computing and business data analysis.
I wrote the systems portion of the software.
Credits in the SAS 1976 manual:
Anthony J. Barr
Language translator; data management and supervisor; GUTTMAN, NESTED, PRINT, SORT
James H. Goodnight
ANOVA, CLUSTER, DISCRIM, GLM. MEANS, NEIGHBOR, NLIN, PROBIT, RSQUARE, STANDARD, STEPWISE, TTEST, VARCOMP
John P. Sall
AUTOREG, BMDP, CONTENTS, CORR, DUNCAN, EDITOR, FACTOR, FREQ, MATRIX, OPTIONS, PLAN, RANK, SA572, SCORE, SPECTRA, SYSREG, function library