Stata

History

Stata version 1.0, designed by Southern Californians William Gould and Sean Becketti, was released in 1985. Written in C for the first generation of MS-DOS computers, Stata 1.0 was principally a regression package with data management features. Basic graphing and programming features were introduced later that year with Version 1.3, and new statistics (analysis of variance, logit, and probit) with 1.5. Many versions later, data management, regression, graphing and programming continue to be core strengths of Stata, although now these are much expanded (e.g., including perhaps a hundred kinds of regression) and complemented by a full array of modern statistical tools. Each new version brought incremental extensions, punctuated by more radical jumps such as a graphical user interface of Version 8 in 2003. Functionality of programs written for earlier versions of Stata has been protected through a version control feature. Users who find a command line interface more efficient than point-and-click can accomplish most tasks using either or both.

Varied and sometimes personal accounts of Stata history, written from the viewpoints of participants, were published in a special issue of The Stata Journal for Stata’s 20th anniversary in 2005 and in a 30-year retrospective book edited by Enrique Pinzon in 2015. The 2005 articles include a detailed history and timeline of Stata development by Nicholas Cox [Page 1612]and a conversation with William Gould, along with stories about the first out-of-house Stata book and launching Stata Technical Bulletin, precursor to The Stata Journal. The 2015 book includes reflections by Becketti and assessments by other authors on Stata’s contributions to epidemiology, biostatistics, public health, public policy, microeconomics, political science, and psychology.

Platforms

Stata is available for Windows, Macintosh, and Linux/Unix computers. Licenses are not platform-specific, so a user could, for instance, install copies under the same license on a Linux system at work, a Windows computer at home, and a Macintosh laptop for travel. Stata data sets, programs, and other data can be used interchangeably across these platforms.

Flavors, Versions, and Updates

Stata comes in a variety of flavors from Small Stata, inexpensive and meant for students, through progressively more capable IC, SE, and MP (multicore/multiprocessor) versions. Unlike modular statistical packages, all Stata flavors have the same complete set of features and documentation, so that even Small Stata can, for example, fit a mixed-effects generalized structural equation model. The main difference among flavors involves the size of programs and data sets that they can handle. Small Stata is limited to no more than 99 variables and 1,200 observations; Stata/MP can analyze 10–20 billion observations or more given current computers and should expand to 281 trillion as hardware advances.

...