Re: Dilemma of business meaning and stat result
- From: Russell.Martin@xxxxxxx
- Date: 1 Aug 2005 06:57:51 -0700
RL wrote:
> If the result of statistics is not making business sense, what would you do?
> Believe in the stat analysis or do something else?
As Art said, more details would help people make meaningful
responses. However, the general situation you refer to is
like that which we often face in science, i.e. what the data
(specifically the statistical analysis of the data) tells us
does not jibe with what we expect. The usual case is that
we have a model, theory, or hypothesis that we think the
data will confirm, but it doesn't. This means we don't
understand something, but it is not always clear what. It
may be that our theory is not correct. It may be that our
statistical analysis is based on some assumptions which
contradict those of the theory or how the data were gathered.
It may be that the data simply are wrong (e.g. my measuring
apparatus is broken and I don't know it) or not applicable
to testing the theory like we thought they would be (IOW the
data are not capable of answering the question I'm asking).
So you need to reexamine your situation. You have a model of
what you think the relationships among the data are (even if
you don't think you do, you do, especially since a statistical
analysis assumes a model). You have an idea of how these data
and relationships relate to your business. You have gathered
data and done some analysis. If it was a science experiment,
I'd reexamine my assumptions and theory carefully, reexamine
how I got the data and whether or not I really understood what
the data represent, reexamine the assumptions that the analysis
requires, etc. It may be that your business is to complex for
the data you have and/or the analysis you did to tell you
something sensible about it, so the analysis gives you nonsense.
You may need more or better data, or a model which corresponds
more closely with the realities of your business, or ...
Sometimes it is helpful to go over these things with someone
who is knowledgeable about the field (e.g. your business) but
hasn't gone through the process with you, so they can see
things with a fresh eye. If you can find such a person, accept
what they have to say, the questions they ask, etc., with an
open mind. It may seem obvious to you, but that may be just
the problem. What looks obviously fine to you, because you've
been looking at it so long, may be the source of your confusion.
Cheers,
Russell
--
All too often the study of data requires care.
.
- References:
- Prev by Date: Re: Dilemma of business meaning and stat result
- Next by Date: Re: test of normality
- Previous by thread: Re: Dilemma of business meaning and stat result
- Next by thread: Re: Dilemma of business meaning and stat result
- Index(es):
Relevant Pages
|
