Subject: Re: Eight Features of an Ideal Intro Stat Course (Response to comments by Rossi Hassad) To: EdStat-L and sci.stat.edu From: Donald B. Macnaughton <firstname.lastname@example.org> Date: Wednesday November 25, 1998 cc: Rossi Hassad <email@example.com>
In a November 23 post I recommend that statistics teachers omit discussing univariate distributions near the beginning of the in- troductory course. I recommend that teachers instead concentrate on discussing relationships between variables. Quoting that post (and taking an opposing point of view), Rossi Hassad writes (on November 23) > This is all very stimulating. I strongly support the logical > progression from univariate to bivariate. As I note in a paper (1998a, sec. 9.1), I agree with Rossi that univariate distributions *logically* precede relationships be- tween variables. However, it is possible -- and, I believe, emi- nently practical -- to begin the introductory course by discuss- ing relationships with NO reference to univariate distributions. I recommend concentrating first on relationships because students find univariate distributions to be *boring* -- students can see little or no obvious practical use of the study of univariate distributions. (I am NOT saying that univariate distributions have no practical uses -- they have many important uses in statistics. I am only saying that univariate distributions have few uses that can be readily appreciated by students at the beginning an introductory statistics course.) On the other hand, relationships between variables have many ob- vious practical uses, as can be seen by their use across almost all empirical research. For example, ALL reputable medical research into the effective- ness of new treatments for diseases can be easily viewed as studying relationships between variables. The response variable reflects the amount of disease patients have (possibly in terms of the length of their survival), and the predictor variable(s) reflect the amount(s) of treatment(s) administered to the pa- tients. I submit that it is difficult to find examples of univariate dis- tributions that are of real practical interest to beginning stu- dents. I invite Rossi and other proponents of teaching univari- ate distributions near the beginning of the introductory statis- tics course to describe some such examples. (I discuss some pu- tative examples in the earlier post [1998b] and in the paper [1998a, appendix G].) Anyone proposing examples of interesting univariate distributions may also wish to consider the idea that we can make *any* inter- esting univariate distribution of a variable substantially *more* interesting by studying the *relationship* between that distribu- tion and some other variable(s) -- that is, by studying the rela- tionship between the variables. > Students learn best when our teaching strategies are based on > the "constructivist learning theory", Under the constructivist view of learning the teacher structures the course in a way that encourages students to construct their own knowledge through active participation (Moore 1997, sec. 3). I agree that it is important to encourage students to construct their own knowledge. However, - many statistics teachers will agree that a reasonable first goal for the introductory course is to give students a lasting appreciation of the vital role of the field of statistics in empirical research - most empirical research projects can NOT be viewed as studying univariate distributions, but can be easily viewed as studying relationships between variables - relationships between variables are easy to understand (espe- cially when taught in terms of practical examples) and - it appears that students can construct their own knowledge of relationships between variables at least as easily as they can construct their own knowledge of univariate distributions (be- cause relationships are more interesting -- they lead to accu- rate prediction and control). Therefore, I believe it is more effective to encourage students to construct knowledge about relationships between variables than to encourage them to construct knowledge about (boring) univari- ate distributions. > that is they would quickly link Bi to Uni and get the big pic- > ture. There are indeed strong linkages between relationships between variables and univariate distributions. Specifically, relation- ships between variables can be viewed as a *generalization* of univariate distributions. Conversely, univariate distributions can be viewed as a *special case* of relationships between vari- ables. (I discuss these linkages further in the earlier post [1998b]). Thus we can introduce either of the two concepts in terms of the other. However, since relationships between variables are much more interesting and more useful (because they allow more accu- rate prediction and control), I suggest it is more reasonable to start with relationships. After students have a good sense of the use of the concept of a relationship between variables in empirical research, we can then discuss univariate distributions as a special case of relation- ships. At that point we can exploit the linkages Rossi mentions. In addition, we can show students the small but important roles that univariate distributions play in the study of relationships between variables. > Believe me it's as simple as this. Given that two presidents of the American Statistical Association have stated that "students frequently view statistics as the worst course taken in college" (Hogg 1991, Iman 1994), I suggest that the issues are not as simple as Rossi (with his traditional approach) would like us to believe. However, I suggest a solution to the "worst course" problem is available: If we focus on the *practical* study of relationships between variables as a means to accurate prediction and control, I believe we can move statistics from its current sometime back- water status to its rightful role as a respected cornerstone of all empirical research. ------------------------------------------------------- Donald B. Macnaughton MatStat Research Consulting Inc firstname.lastname@example.org Toronto, Canada ------------------------------------------------------- REFERENCES Hogg, R. V. (1991), "Statistical education: Improvements are badly needed," _The American Statistician,_ 45, 342-343. Iman, R. L. (1994), "The importance of undergraduate statistics," _Amstat News,_ Number 215, December 1994, 6. Macnaughton, D. B. 1998a. "Eight features of an ideal introduc- tory statistics course." This paper is available at http://www.matstat.com/teach/ Macnaughton, D. B. 1998b. "Re: Eight features of an ideal intro stat course (response to comments by Gary Smith)." Posted to EdStat-L and sci.stat.edu on November 23, 1998. Available at http://www.matstat.com/teach/p0036.htm Moore, D. S. 1997. "New pedagogy and new content: The case of statistics (with discussion)." _International Statistical Re- view,_ 65, 123-165.
Home page for Donald Macnaughton's papers about introductory statistics