Subject: Re: Eight Features of an Ideal Intro Stat Course
         (Response to comments by Rolf Dalin)

     To: (July 31, 1998)
         EdStat-L (August 4, 1998)

   From: Donald B. Macnaughton <>

Referring to a July 17 post of mine, Rolf Dalin writes (on July 

> I think your suggestion for the introductory statistics course
> is very good.  It is very much like the way I'm teaching since
> a few years in courses in "quantitative methods in /subject/".
> I'm also using relationships between variables as the unifying
> concept. 

It is always gratifying to see thoughtful teachers who have inde-
pendently discovered the unifying power of the concept of a rela-
tionship between variables.

> My experience is the same as yours that students find it boring
> and/or meaningless to start with a lot of studies of descrip-
> tive univariate statistics.  

It is an unfortunate fact of contemporary statistics that tradi-
tion has handcuffed some introductory statistics teachers to bor-
ing univariate distributions.

> I got some interesting feedback from a group of ... business
> students ... after the first introductory lecture of the
> course.  I gave them a printout of data ....  It was intended
> to be a starting point for group discussions ... [of] the con-
> cept of "variable". ...  But already before splitting up in
> groups to discuss, they immediately started to state questions
> about relationships between variables.  They simply didn't
> bother with anything less than that.   

As Rolf's students illustrate, once we have introduced the con-
cept of a relationship between variables to students, they are 
quick to recognize the value of relationships.  They are espe-
cially quick to recognize the value if 

- we emphasize that relationships provide the most accurate known 
  objective means for predicting or controlling the values of any 
  variable and 

- we use practical examples with response variables that students 
  can see clear value in predicting or controlling.

In the following remarks Rolf obliquely refers to two goals for 
the introductory statistics course that I discuss in a paper 

> So I support your approach, but still have a couple of remarks:
> 1) I would use a third goal:  Preparing for further studies (of
> statistical methods and of empirical work).
> We don't want a course in any subject to be a "blind alley" in
> the respect that it teaches methods that cannot be developed
> further in a higher degree of generality or that it gives
> skills that will limit the participant in some respects. ...  I
> would not rank that goal as high as the two you've mentioned,
> but I would like to have it there because ... we could make
> mistakes if we leave it out completely.

I fully agree with Rolf's third goal, which I see as being deriv-
able from the second goal, which is

    to teach students to understand and use some useful sta-
    tistical methods in empirical research.

When we are implementing this (second) goal in the introductory 
course we must draw up short at many points (due to the broadness 
of the field of statistics and the shortness of the course).  At 
each of these points the responsible approach is (subject to time 
constraints) to lay the proper groundwork for students' further 
work in later courses.

> 2) I would also have as an important feature of a statistics
> course a complete list of objectives (derived from the goals)
> to inform the participants of what is expected of them. 

I agree.  I believe a formal *hierarchical* organization of ideas 
is the easiest way to understand any body of thought.  Thus for 
the introductory statistics course a few goals should be stated 
at the top, and these should then be broken into lower-level more 
detailed objectives.

Donald B. Macnaughton   MatStat Research Consulting Inc      Toronto, Canada


Macnaughton, D. B. 1998.  Eight features of an ideal introductory 
   statistics course.  This paper is available at

Home page for Donald Macnaughton's papers about introductory statistics