There is Only One Statistical SoftwaRe

I show how a variety of tasks in econometric computing, usually performed through spreadsheet manipulation and use of one or more software packages, can be accomplished effectively through R, substituting many different tools and interfaces with just one, free, open and platform-independent software environment ; how in this environment the need for manual intervention in summarizing or documenting the results of research is virtually eliminated through comprehensive automated interfacing; and how new possibilities in econometric programming arise from the distinctive features of the R system. I briefly introduce the R Project and give examples of R’s scalability, illustrating the different available levels of usage, from useR of precompiled procedures to programmeR of new methods, to show that R can be far less difficult to use than its reputation goes. Not being tied to any particular GUI and/or editor, R can be used together with a general-purpose editor of your choice. Graphical user interfaces are available as well, so that R can pretty much be anything for everyone. I discuss reproducibility of results as compared with point-and-click statistical software, giving a quick overview of the logging facilities and interfacing possibilities of R geared towards the production of statistical reports and scientific papers. I describe some implementation examples where programming challenges are overcome by taking advantage of object-orientation features, of the status of functions as first-class objects, of the special ‘dots’ argument, of implicit recursion and of explicit parsing and evaluation of dynamic text. The concept of abstraction is discussed and applied to tasks and data types. Lastly, I give some examples of useful integration with other statistical disciplines through which R overcomes the limitations of typical ‘single-purpose‘ applications, allowing researchers to use one interface and one syntax instead of having to switch between many tools.