Ghosts of Statistics Software: Past, Present, and Future

• In Event: Educational Statisticians SIG Business Meeting

Abstract

The purpose of this presentation is to reflect on the status of statistical software since the 1960’s, as well as to project ahead a bit. Although many of the current statistical techniques have been around for decades, and even centuries, computing capability has driven their applicability. Mainframes became available after World War II, microcomputers about 1975-1980, and the PC around 1981. Obviously today’s computers are smaller, have more computing power, and are faster than ever. Just compare a deck of computer cards to rotary calculators to cassette tapes to CDs to cloud computing. All of these have occurred since I joined SIG:ES.

For statistical computing, until the release of BMD in 1961, researchers were reliant upon non-commercially available code. Along came SPSS (1968) and SAS (1971). Windows 1.0 was released in 1985 leading to Windows versions of those and other statistics software packages.

Today we have statistics software in the cloud, using big data, with open source software, and/or on an iPad. The future is already here. We can conduct univariate, bivariate and multivariate analyses via the iPad, (b) do online SEM through WebSEM in R, and (c) use cloud-based statistics packages. Statistical software not making these advances will be left behind. The days of submitting your job on computer cards and waiting 24 hours, only to find an error, are long gone. These are the days of instant results, so we had better be ready.

Author

• Richard G. Lomax, The Ohio State University