Re: anecdotes

Truman Prevatt (truman.prevatt@netsrq.com)
Mon, 6 Jan 1997 13:04:21 -0400

>Susan Evans writes "I have no objections to anyone experimenting with their
>own horse; however whatever results you obtain are not statistically viable
>and really can't be reliably applied to 'every' horse." The anecdotal
>accounts on the list are fun to read, but only prove that thus and so applied
>to that perspm's particular horse at that particular time.

Several years ago the Univ of South FL, Sarasota wanted to offer a
statistics course and had no one available to teach it. So I got somehow
recuritied to teach it, not my favorite subject to teach. One thing
everyone should keep in the back of their minds when reading what "this or
that compound or procedure does" is:

a. who is doing the testing?
b. are the resultes published results in a refereed forum?
c. do the statistics support the conclusions?

One common fallacy is to believe someone who has done testing - valid
statistically or not. Just remember what I drilled into my students heads
- "statistics don't lie but statisticians sure do". Or as Mark Twain put
it, "There are liers, damn liers and then there are statisticians".

Truman

Truman Prevatt
Sarasota, FL