Subject: Re: EPR Approach to Intro Stat: 
             Relationships Between Variables

   Date: July 18, 1996 20:48 EDT

   From: Donald Macnaughton <>

     To:, ,

On July 17, 1996 Ronan Conroy wrote:

> At 1:50 pm 16/7/96, Bryan W. Griffin wrote:
>> The emphasis [in the EPR approach] is clearly placed on pre-
>> diction and control, yet I think the most important goal of
>> research is totally ignored in your discussion--explanation.
> I would add another important activity: estimation.

I agree that *estimation* is an important activity of statistics.  
However, I believe that we can view "estimation" as being essen-
tially equivalent to "prediction".  The two concepts are equiva-
lent because both are equivalent to "determining" the values of 
properties of entities.

(I describe two useful minor distinctions between the concepts of 
"estimation" and "prediction" in the appendix.)

Therefore, the EPR approach covers the concept of "estimation" 
under the concept of "prediction".

Material about the entity-property-relationship approach to the 
introductory statistics course is available at

Donald B. Macnaughton      MatStat Research Consulting Inc.          Toronto, Canada

There are two senses in which we can distinguish between the con-
cepts of "estimation" and "prediction".

One sense of the distinction is that researchers tend to "estim-
ate" the values of "parameters" and "predict" the values of vari-
ables.  However, parameters and variables are two different names 
for the same concept--properties of entities.  

(The term "parameter" is generally used to signify properties of 
three special types of entities:  populations, distributions, and 
model equations.  The term is used for these three special types 
of properties to distinguish them from the other properties that 
are usually present in the same discussion, which are, of course, 
called "variables".  The distinction is useful because the two 
types of properties have a critical difference--parameters are 
generally properties that are assumed to be constant throughout 
an empirical research project, while variables are properties 
that are generally assumed to vary in one or more senses in an 
empirical research project.)

A second distinction between "estimation" and "prediction" is 
that the term "estimation" is often applied when researchers are 
determining the value of a property in the present or past, while 
the term "prediction" is usually applied when researchers are de-
termining the value of a property in the future.

I thank Bruce Frey for helpful comments pertaining to this post-