I also agree with Philippe Grosjean that there is a need to investigate the effects of different types of GUIs on software "usability". To do that, however, there is a need to develop a set of appropriate metrics, and to understand that the appropriateness of a given metric (and ultimately a given type of GUI) is likely to be conditional on the type of user. My sense is that the heat of the rhetoric in this discussion is driven by perceived differences in implicit metrics and student/user types. If I am training students to become professional biostatisticians who will be analyzing data from clinical drug trials, then my concerns (really the most important metrics) relate to the ability of the student to do the most complete and appropriate analysis possible in a way that can consistently be replicated. In this instance, the most preferred user interface for teaching these students is going to be a text editor or an IDE. On the otherhand, if my students are MBAs who will ultimately be marketing managers, placing them in a situation where they will be "customers" of statistical analyses rather then analysts themselves, then my concerns and appropriate metrics are very different. Specifically, I want to make these students informed buyers of statistical analysis. The best way to make them informed about the process of statistical analysis, and how to use the results of an analysis to make better managerial decisions, is to have them actually conduct several analyses and have them apply the results to a managerial problem in a low-risk, classroom environment. However, in this sort of environment the two things I want to accomplish are: (1) minimizing the time it takes them to be able to conduct a usable (but likely imperfect) analysis and (2) minimizing the frustration level they experience when working with the software. In this instance the best interface will be very point-and-click oriented, which will greatly limit a student's freedom in the types of analyses s/he can conduct, but will keep them out of "trouble" and minimize the time it takes them to use the software in a useful way. So why this long-winded discussion? The point I want to make is that there probably isn't a single "ideal" interface to R. In a related thread, one contributor commented that there were 10 GUIs for R in some form of development, and there could ultimately be 20 (my sense is the author perceived this to be a bad thing). It is possible that 10 different GUIs is the right number, although it is an empirical question. The saving grace is that most (all)? of us are empirical researchers. Dan Putler -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html Send "info", "help", or "[un]subscribe" (in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
Dan Putler wrote:>I also agree with Philippe Grosjean that there is a need to investigate the >effects of different types of GUIs on software "usability". To do that, >however, there is a need to develop a set of appropriate metrics, and to >understand that the appropriateness of a given metric (and ultimately agiven>type of GUI) is likely to be conditional on the type of user.Exactly! And it is the most interesting part of the job! Creating a pale copy of one existing GUI is not a very exciting work. Thinking, designing and validating an original and efficient GUI is a science that could probably be appealing for many of us (after all, we are statisticians and programmers involved in research and teaching). Remember the work done at Xerox's Palo Alto lab at the early ages of the GUIs, where they invented icons, menus,... I believe that such nice adventure could be also conducted among a virtual lab constituted by a community of open source developers around the R project in a field that obviously needs some improvements in the user interface design: data analysis. Best, Philippe Grosjean -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html Send "info", "help", or "[un]subscribe" (in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
From: Philippe Grosjean (phgrosje at ulb.ac.be) Date: Fri Jul 12 2002 - 12:53:39 CEST [...]> probably be appealing for many of us (after all, we are statisticians and > ^^ ?? Not all of us! > programmers involved in research and teaching). Remember the work done atWell, in any event, please do remember that the majority of students taking statistics courses may well be not (future) statisticians, but (future) researchers involved in other things that may include programming and (more likely) teaching. A well designed GUI will certainly "open up" R to users who until now have used (and had to pay for) commercial products to run the boring? assortment of tests they need in their work. Those are now probably a minority (due to the lack of a GUI...?), but once a good GUI gets off the ground, they are bound to become more numerous. It would be nice to have everybody's needs taken into account from the beginning! (Did somebody say altruism there? :)) RenE Bertin -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html Send "info", "help", or "[un]subscribe" (in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
Ah yes - one stable, powerful, flexible, well-supported stats engine, with a variety of GUI frontends that will be used depending on where one is coming from. Hey, that sounds like heaven... Stuart Dr Stuart Leask MA MRCPsych, Clinical Lecturer in Psychiatry University of Nottingham Dept of Psychiatry, Duncan Macmillan House Porchester Road, Nottingham. NG3 6AA. UK http://www.nottingham.ac.uk/psychiatry/staff/sjl.html ----- Original Message ----- From: "Dan Putler" <dputler at scu.edu> To: <r-help at stat.math.ethz.ch> Sent: Thursday, July 11, 2002 7:11 PM Subject: Re: [R] R GUI For Which User?> I also agree with Philippe Grosjean that there is a need to investigatethe> effects of different types of GUIs on software "usability". To do that, > however, there is a need to develop a set of appropriate metrics, and to > understand that the appropriateness of a given metric (and ultimately agiven> type of GUI) is likely to be conditional on the type of user. > > My sense is that the heat of the rhetoric in this discussion is driven by > perceived differences in implicit metrics and student/user types. If I am > training students to become professional biostatisticians who will be > analyzing data from clinical drug trials, then my concerns (really themost> important metrics) relate to the ability of the student to do the most > complete and appropriate analysis possible in a way that can consistentlybe> replicated. In this instance, the most preferred user interface forteaching> these students is going to be a text editor or an IDE. On the otherhand,if> my students are MBAs who will ultimately be marketing managers, placingthem> in a situation where they will be "customers" of statistical analysesrather> then analysts themselves, then my concerns and appropriate metrics arevery> different. Specifically, I want to make these students informed buyers of > statistical analysis. The best way to make them informed about theprocess> of statistical analysis, and how to use the results of an analysis to make > better managerial decisions, is to have them actually conduct several > analyses and have them apply the results to a managerial problem in a > low-risk, classroom environment. However, in this sort of environment the > two things I want to accomplish are: (1) minimizing the time it takes themto> be able to conduct a usable (but likely imperfect) analysis and (2) > minimizing the frustration level they experience when working with the > software. In this instance the best interface will be verypoint-and-click> oriented, which will greatly limit a student's freedom in the types of > analyses s/he can conduct, but will keep them out of "trouble" andminimize> the time it takes them to use the software in a useful way. > > So why this long-winded discussion? The point I want to make is thatthere> probably isn't a single "ideal" interface to R. In a related thread, one > contributor commented that there were 10 GUIs for R in some form of > development, and there could ultimately be 20 (my sense is the author > perceived this to be a bad thing). It is possible that 10 different GUIsis> the right number, although it is an empirical question. The saving graceis> that most (all)? of us are empirical researchers. > > Dan Putler > -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-> r-help mailing list -- Readhttp://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html> Send "info", "help", or "[un]subscribe" > (in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch >_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._. _._ -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html Send "info", "help", or "[un]subscribe" (in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._