[R-sig-Epi] R-sig-Epi Digest, Vol 18, Issue 1
Ian Fellows
ifellows at ucsd.edu
Thu Oct 4 18:16:59 CEST 2007
-----Original Message-----
From: r-sig-epi-bounces at stat.math.ethz.ch
[mailto:r-sig-epi-bounces at stat.math.ethz.ch]On Behalf Of
r-sig-epi-request at stat.math.ethz.ch
Sent: Thursday, October 04, 2007 3:00 AM
To: r-sig-epi at stat.math.ethz.ch
Subject: R-sig-Epi Digest, Vol 18, Issue 1
Send R-sig-Epi mailing list submissions to
r-sig-epi at stat.math.ethz.ch
To subscribe or unsubscribe via the World Wide Web, visit
https://stat.ethz.ch/mailman/listinfo/r-sig-epi
or, via email, send a message with subject or body 'help' to
r-sig-epi-request at stat.math.ethz.ch
You can reach the person managing the list at
r-sig-epi-owner at stat.math.ethz.ch
When replying, please edit your Subject line so it is more specific
than "Re: Contents of R-sig-Epi digest..."
Today's Topics:
1. Probability of falling incidence (Ralf Finne)
2. Re: Probability of falling incidence (Kevin Viel)
----------------------------------------------------------------------
Message: 1
Date: Wed, 03 Oct 2007 17:24:28 +0300
From: "Ralf Finne" <Ralf.Finne at syh.fi>
Subject: [R-sig-Epi] Probability of falling incidence
To: <r-sig-epi at stat.math.ethz.ch>
Message-ID: <4703D04D020000EE00004826 at valhall.syh.fi>
Content-Type: text/plain; charset=UTF-8
Hi everybody.
Below are the incidence numbers for end-stage renal disease in the
Helsinki region for the past five years.
The following question has arisen:
What is the probability that the decreased incidence for 2006
is due to random variablilty and not caused by any systematic effect?
2002 75
2003 73
2004 69
2005 65
2006 28
Are there any solutions available in R?
Very thankful for your answers
Ralf Finne
Svenska yrkesh?skolan
Vasa Finland
------------------------------
Message: 2
Date: Wed, 03 Oct 2007 09:38:54 -0500
From: Kevin Viel <kviel at sfbrgenetics.org>
Subject: Re: [R-sig-Epi] Probability of falling incidence
To: r-sig-epi at stat.math.ethz.ch
Message-ID: <001b01c805cb$1fa7f4e0$723a7cce at win.sfbrgenetics.org>
Content-Type: text/plain; charset=us-ascii
> -----Original Message-----
> From: r-sig-epi-bounces at stat.math.ethz.ch
> [mailto:r-sig-epi-bounces at stat.math.ethz.ch] On Behalf Of Ralf Finne
> Sent: Wednesday, October 03, 2007 9:24 AM
> To: r-sig-epi at stat.math.ethz.ch
> Subject: [R-sig-Epi] Probability of falling incidence
>
> Hi everybody.
>
> Below are the incidence numbers for end-stage renal disease
> in the Helsinki region for the past five years.
>
> The following question has arisen:
>
> What is the probability that the decreased incidence for
> 2006 is due to random variablilty and not caused by any
> systematic effect?
>
> 2002 75
> 2003 73
> 2004 69
> 2005 65
> 2006 28
>
> Are there any solutions available in R?
>
> Very thankful for your answers
Ralf,
This is not enough data for us to be able to suggest a test. We would
need to know the study design, with particular emphasis on the statistical
sampling strategy (stratified, simple random sampling, etc.). 28 cases out
of 1000 p-y is much different from 280 cases out of 10000 p-y, for instance.
Unfortunately, I am only lurker on this list and cannot suggest a
procedure in R.
HTH,
Kevin
Kevin Viel, PhD
Post-doctoral fellow
Department of Genetics
Southwest Foundation for Biomedical Research
San Antonio, TX 78227
------------------------------
_______________________________________________
R-sig-Epi mailing list
R-sig-Epi at stat.math.ethz.ch
https://stat.ethz.ch/mailman/listinfo/r-sig-epi
End of R-sig-Epi Digest, Vol 18, Issue 1
****************************************
Hello Ralf,
My first instinct is to assume that your data is from a poisson
distribution. If you have information on renal failure on a finer timescale,
I'd look to see if this assumption holds.
We can test the equality of two poisson rates using the binomial
distribution. Under the null hypothesis that the rate of the first 4 data
points is the same as the last, the distribution of the last rate,
conditional upon the five year rate (23+75+73+69+65) is:
y5 - binomial(n=23+75+73+69+65,p=1/(1+4))
so using an exact binomial test we have:
> sum(dbinom(0:23,23+75+73+69+65,1/(1+4)))*2
[1] 2.626615e-09
so, if there was no difference in 2006, the probability that we would see a
rate as extreme as 23 is VERY small <.0000001.
With so few years, the poisson assumption, though reasonable, is impossible
to check. I would be interested in other opinions on this.
Ian
More information about the R-sig-Epi
mailing list