At 12:25 AM 10/23/00 -1000, Daniel Blaine wrote:
> I'm not sure what you mean by a "no brainer" since I've
> "interpreted your interpretations" to suggest that concepts in
> and around parameter estimation and hypothesis testing are not
> easy ones for our students to grasp.
>
> Aloha. -dan
trying to stay at the intro stat course level ... there is only so much
time ... one has to make choices as to what topics to cover ... which ones
(though perhaps interesting) one has to ignore
i think in a case like this, if one has made a decision to have a unit on
the correlation coefficient ... and presumably also on regression ... then,
one of the things that can and should go ... is a discussion of the
mechanics of a formula like the one (or some variant) for the point
biserial ...
in the context of a correlation, i see no value in discussing the nuances
of the pb formula ... and see this as time taken away from something else
that would be more useful FOR students
in this case, one needs to start with some basic version (and there are
several) of the pearson formula ... and, can then discuss how this is used
in contexts where there are different combinations of TYPES of variables
(both continuous, both dichotomous, one of each) ... is this not the
important thing? after all, they are NOT going to calculate this by hand,
not in the real world for sure ... so, the additional introduction of once
upon a time time saving formulas that we see for pb and the phi coefficient
... totally miss their mark ... and only ADD confusion ... they don't
clarify anything
=================================================================
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
http://jse.stat.ncsu.edu/
=================================================================