In article <8du6nvsor0c6cern49nm059k1rvrqonlr6@4ax.com>,
 <myhome@manitobatelephonesystem.net> wrote:
|X-No-Archive: Yes   On 25 Sep 2003 22:30:50 GMT,
|roberson@ibd.nrc-cnrc.gc.ca (Walter Roberson) wrote:
|>If one were to answer the questions at random, then the average score
|>would be 39.55,

|>In other words, people who score in that range should probably think
|>of the 'test' as little better than Astrology;

|so you didn't fare too well in the test then?

Hard to say. How would one fair well or badly on that test?

My score was within two standard deviations of what you would expect
if a person were to answer randomly [which I didn't.] The category
I fell in according to the scoring key had a flattering description
of me. Most people would read the matching parts out of the description
and just ignore the rest -- the same way that Astrology says a
lot of things and lets the reader pick out what the reader wants
to hear. I, though, know myself well enough to know that some of the
most flattering parts of the description of my supposed personality
are outright wrong; a better match for me would be in one of the
adjacent categories that was also within two standard deviations of
the central value.


What I'm saying is that the categories close to the score of 40
are written to have something to match and cheer up everyone, because
that's the area that most people are going to land in. Only at
the extremes of the scale can the "test" possibly say anything
non-generic, having consistantly selected out extreme behaviours.
You take one extreme of behaviour and assign it the highest score
in a question; you take another extreme and assign it the lowest
score in the question, and you give all the other answers in the
question a pretty much random value, knowing that all the common
behaviours will cancel each outer out in the end by virtue of
the Central Limit Theorem. Extreme scores -might- be meaningful;
central scores are basically pap.
-- 
How does Usenet function without a fixed point?