On the Statistics Education Section’s email listserv, there is an interesting debate going on regarding the software with which statistics education should be taught.
The first email stated:
My dean is a member of a CUPM committee writing a section on using technology in teaching. She has asked me for recommendations for technology in teaching Statistics. I know all the standard ones (TI calculators, Minitab, R, SPSS, SAS, JMP, etc). Are there other forms of technology that should be mentioned here?
Many aspects of the conversation are curious, and in particular, the conversation was driven towards the usefulness of Microsoft Excel for teaching statistics.
At this point, St. Lawrence’s Robin Lock took over, responding:
Always fun to see the Excel for teaching statistics debate every few years - especially for those who may have missed the earlier versions. I have to admit that I rarely use Excel for teaching or doing statistics, so I took a quick look at the latest version of base Excel to see what it now does. To keep things simple, in intro stat we usually worry about two basic kinds of variables, categorical and quantitative, so how would Excel do for displaying the distribution of a single sample of either of these variables. Not too bad for categorical, I see bar charts, pie charts and some other possible options. How about a single quantitative variable? Now I'm stuck. Maybe, with some effort, I can coerce a column chart to look vaguely like a histogram, but how about a dotplot, boxplot or real histogram with a good numeric horizontal scale. There may be ways, but they aren't obvious to me and probably not convenient for students. I'm sure someone will suggest some possibilities! Could I recommend using software for teaching statistics that has no easy way to create a reasonable graphical display of one of the two common types of variables (and the type that most often needs a graphical display)? No. I'm sure there are add-ins that accomplish such tasks which might be reasonable to use in teaching, but without such capabilities I wouldn't want to try. Maybe some of the 80% can show me how, but I would not be willing leave an idea as fundamental as looking graphically at the distribution of a single quantitative variable out of a course.
I wholeheartedly agree.
Note: This was published with Lock’s permission