On Mon, 10 Jan 2000 18:33:47 -0800, "Dave and Kim Nulton"
<[EMAIL PROTECTED]> wrote:
> Quite frankly Robert the details are proprietary. I suppose I could have
> been more descriptive, but I don't see what the shape of my distribution
> have to do with what it represents. I have received several email replies
> with various recommendations. I'll add transforming to the list. Not being
> a statistician by trade, the little hints I have received should provide a
> good starting point at the library. Thanks again
- just to be counted: I will side whole-heartedly with Robert, and I
think the statisticians with consulting experience will be with us.
Robert describes what happens in the real world. When we *learn* what
has generated the data, when we have pried out the news from people
who were sure that we had no need of it, it has been -- too often --
revealing and important. Not 100% of the time, and maybe it is not
much more than 25% of the time (but it has certainly been no less). I
don't want someone to say, "What a TOTAL idiot you have to be, to
ignore XXX!" I don't want someone to say THAT about my advice.
Or, in this case, you should not be able to say, "The folks on
sci.stat.edu thought that a little bit of xxx would be okay" , since
the folks on sse want to warn you that good statistical advice is
still an art; if you only provide a caricature of the data, you might
get back only a caricature of an answer, no matter how inspired a
guesser your advisor may be.
BUT YOU win (and deserve) a reputation as a flake, an unreliable
screw-up, if you miss an important, *obvious* issue, even once per 10.
Or something like that. And if you knew what was "obvious" then you
wouldn't be asking for that advice.
--
Rich Ulrich, [EMAIL PROTECTED]
http://www.pitt.edu/~wpilib/index.html
