Re: Testing the Homogeneity and Sufficiency of empirical data
- From: Richard Ulrich <Rich.Ulrich@xxxxxxxxxxx>
- Date: Tue, 13 Sep 2005 13:30:10 -0400
On Mon, 12 Sep 2005 08:29:59 EDT, Idoia <idoia30@xxxxxxxxx> wrote:
> Hello,
>
> In order to be able to use statistical tools to calculate reserves for a Professional Indemnity and a Medmal book, my state insurance department requires that I prove that the data that I´m using is Homogeneous and Sufficient for statistical analysis.
>
> ¿Any ideas on which methods I can use to test this?
In the 3 or 4 examples that I have seen where a state
or federal regulation required details like that, the
government provided very explicit guidelines as to
what was acceptable, and how to do it. Or, what the
competing statistical options were. Your options,
I think, are: (1) Ask your boss. (2) Google for the law.
In other words, do what they say to do; don't re-invent
the wheel. Even if you invent a better wheel, you will
have to show someone (possibly, bureaucrats rather than
statisticians) that it is a wheel, and then, that it is better.
>
> For Homogeneity, I´ve read that histograms are used:
[snip, detail]
They probably want something more formal than looking
at histograms with arbitrary groupings.
>
> For Sufficiency, I found a paper stating that your mimimum sample size for estimating the mean within d% of error with a% confidence level is:
[snip, detail]
If that is how they want to estimate "sufficiency", you still
need to look up their details to see what error range
they consider acceptable.
[ snip]
--
Rich Ulrich, wpilib@xxxxxxxx
http://www.pitt.edu/~wpilib/index.html
.
- Follow-Ups:
- References:
- Prev by Date: Re: Help for random number generator
- Next by Date: use informative subject lines (was Re: Humbly, attentively and waiting feed-back)
- Previous by thread: Testing the Homogeneity and Sufficiency of empirical data
- Next by thread: Re: Testing the Homogeneity and Sufficiency of empirical data
- Index(es):
Relevant Pages
|
