Unfortunately, the session on mixed models, chaired by Douglas Bates, was full. That is a shame, but instead I attended a session on user interfaces. Since I’m trying to get people to use R-Project (solely because I want to collaborate with them using R), these user interfaces might be of help of pulling people over to the R-side of things.
Washington Junger showed a user interface aimed at statistical analysis of epidemiological data, called Epi-R. He wanted researchers in Brasil to be able to use R, without the need for learning some of the complexities associated with it. The user interface consists of an R-library that is able to perform many of the often used functions by epidemiologists. Unfortunately, the demonstration of the actual interface failed due to computer problems (blame ‘windooze’), so it is difficult to comment on that.
The next presentation was by Erin Hodgess, and she presented a plugin package for undergraduate time-series analysis. A promising approach at bringing R-Project to students, and to (carefully!) let them get familiar with a command based statistical language. It has been conceived as an extension to the RCommander, developed by John Fox, and basically consists of some nice very point-and-click options related to time series analysis. Wouldn’t anybody want a tutor who writes such tailored software for you?
The third user interface focuses on environmental statistics, presented by Rudolf Dutter. Also based on RCommander, perhaps the most interesting feature for general use, is their helper function that eases the reading of various types of text-files. Or, it must be their ability of interactively defining the plotting regions, which is quite nice as well.
Three interfaces, three focusses. To summarize, I do feel that user interfaces almost by definition take away some of the flexibility present in R-Project, but they surely lower the threshold for people unfamiliar with R to use it in their specific filed of research.