Subject: Re: EPR Approach to Intro Stat: Relationships Between Variables Date: July 18, 1996 22:48 EDT From: Donald Macnaughton <email@example.com> (formerly firstname.lastname@example.org) To: email@example.com, firstname.lastname@example.org
On July 16, 1996, Will G Hopkins wrote about my posting of July 14: > if you present this sort of stuff to students in those words > you'll lose them, IMHO. I agree. However, my discussion in the posting is for teachers and (as I note in the posting) is condensed. I discuss in refer- ences 1 and 2 below how I recommend the material be presented to students. > I agree with your concepts, but the way to get it across is to > use the vernacular and concrete examples. I believe that when we are presenting statistical concepts to students we must use carefully chosen familiar terminology. Thus it's very important to use the vernacular when possible, although some new terms must be introduced because not all statistical concepts have names in the vernacular. I also agree that it's important to use concrete examples. I recommend that teachers use two types: 1. Teacher-chosen examples are helpful for introducing new con- cepts. 2. Student-chosen examples are helpful because they ensure that the students are considering examples that are of interest. Extensive class discussion of both types of examples is important because it helps the teacher to determine when the students have mastered the concepts. Because the concepts of statistics build on one another, an introductory statistics course can be success- ful only if the teacher ensures that the students have mastered each concept before moving on to the next. I further discuss terminology and examples in references 1 and 2 below. I shall discuss the use of "frivolous" examples in sta- tistics courses in my paper at the Joint Statistical Meetings in Chicago. > <snip> > I teach that a variable is something that can take on various > values. This is correct, of course. But if a teacher doesn't attach var- ables *to* something, the variables are left floating in a haze, completely disconnected, which is certainly not correct. Instead (as I discuss further in reference 3), whenever you have a vari- able in empirical research, entities are always present and the variable is always associated with a property of these entities. Because the entities are often the most tangible aspect of em- pirical research, I believe that we should make them explicit, hence the definition A "variable" is equivalent to a property of an entity. [Note added on December 1, 1996: I now believe a more precise definition is: A "variable" is a formal representation of a property of entities.] I emphasize that the teacher must precede this definition with a characterization of entities and properties of entities (using carefully chosen terminology and many examples), and follow this definition with a careful discussion of examples of variables and their values. > <snip> > I then go on to say that stats is all about: > > (a) Finding one or two numbers to summarise the values of a > variable (e.g. mean and SD for a numeric variable; frequency > table for a nominal variable). These numbers are called simple > statistics. > > (b) Finding one or two numbers to summarise the relationship > between two or more variables (e.g. correlation coefficient, > effect size, relative risk). There's no agreement as to what > to call these, so I call them effect statistics. (I saw a paper > lately in which they were called measures-of-effect statistics, > but that's too much of mouthful.) > > (c) Estimating likely values of these statistics in a > population on the basis of a sample of that population. I > emphasise the importance of confidence intervals here, and I > have few kind words for hypothesis testing. > > That's it. The end. No more mystery. Although this description certainly characterizes some of the ac- tivities of statistics, I think it can be enhanced by showing students the role that the field of statistics plays in empirical research. That role is to provide a broad set of techniques that help satisfy an important goal of empirical research, which is to study relationships between variables as a means to predicting and controlling the values of variables. LINK Material about the entity-property-relationship approach to the introductory statistics course is available at http://www.matstat.com/teach/ ----------------------------------------------------------- Donald B. Macnaughton MatStat Research Consulting Inc. email@example.com Toronto, Canada ----------------------------------------------------------- REFERENCES 1. Macnaughton, D. B. (1996) "The Introductory Statistics Course: A New Approach." This 8000-word draft paper is available at the web site above. 2. Macnaughton, D. B. (1996) "The Entity-Property-Relationship Approach to Statistics: An Introduction for Students." This 21,000-word draft paper is available at the web site above. 3. Macnaughton, D. B. (1996) "EPR Approach to Intro Stat: Enti- ties and Properties. Posted to the sci.stat.edu UseNet news- group (= EdStat-L) on June 25/26, 1996. Available at the web site above.