Prev: Invalid argument to function QUOTE
Next: SAS and 64 bit systems
From: John Uebersax on 16 Sep 2009 16:52
Hello Group,

Does anyone have SAS code to compute Agresti-Coull binomial proportion
confidence intervals they'd be willing to share?

Ref:
Agresti, Alan, and Coull, Brent A. Approximate is better than 'exact'
for interval estimation of binomial proportions. The American
Statistician 52: 119-126, 1998.

Thanks in advance.

John Uebersax
http://www.john-uebersax.com
From: John Uebersax on 16 Sep 2009 17:44
On Sep 16, 1:52 pm, John Uebersax <jsueber...(a)gmail.com> wrote:
To correct what I wrote earlier, this way of estimating the CI of a
binomial proportion is actually called the Wilson score interval,
after Wilson (1927):

http://en.wikipedia.org/wiki/Binomial_proportion_confidence_interval#Wilson_score_interval

John Uebersax

> Hello Group,
>
> Does anyone have SAS code to compute Agresti-Coull binomial proportion
> confidence intervals they'd be willing to share?
>
> Ref:
> Agresti, Alan, and Coull, Brent A. Approximate is better than 'exact'
> for interval estimation of binomial proportions. The American
> Statistician 52: 119-126, 1998.
>
> Thanks in advance.
>
> John Uebersaxhttp://www.john-uebersax.com

From: "Data _null_;" on 16 Sep 2009 17:16
Did you see this?

http://support.sas.com/documentation/cdl/en/statug/59654/HTML/default/statug_freq_sect027.htm

On 9/16/09, John Uebersax <jsuebersax(a)gmail.com> wrote:
> Hello Group,
>
> Does anyone have SAS code to compute Agresti-Coull binomial proportion
> confidence intervals they'd be willing to share?
>
> Ref:
> Agresti, Alan, and Coull, Brent A. Approximate is better than 'exact'
> for interval estimation of binomial proportions. The American
> Statistician 52: 119-126, 1998.
>
> Thanks in advance.
>
> John Uebersax
> http://www.john-uebersax.com
>
From: John Uebersax on 17 Sep 2009 12:24
Thank you Data_null_ and Peter for the very helpful suggestions!

I'll try to get a copy of the Agresti & Coull (1998) article, to clear
up the terminology. They described, if I recall correctly, several
alternative methods for estimating the confidence interval of a
binomial proportion. Among these was the Wilson score interval, and
another (similar?) approach that has come to be called the Agresti-
Coull interval(?). Note that SAS proc freq reports both a "Wilson"
and an "Agresti-Coull" interval; and in the example at the link
Data_null_ gave the values are identical. I believe Agesti & Coull
(1998) then suggested a simplification of their method that supplies a
2 in the place of 1.96 for a 95% confidence interval, calling the
result the "modified Wald" method.

John Uebersax
www.john-uebersax.com
From: Oliver Kuss on 18 Sep 2009 05:49
On 16 Sep., 22:52, John Uebersax <jsueber...(a)gmail.com> wrote:
> Hello Group,
>
> Does anyone have SAS code to compute Agresti-Coull binomial proportion
> confidence intervals they'd be willing to share?
>
> Ref:
> Agresti, Alan, and Coull, Brent A. Approximate is better than 'exact'
> for interval estimation of binomial proportions. The American
> Statistician 52: 119-126, 1998.
>
> Thanks in advance.
>
> John Uebersaxhttp://www.john-uebersax.com

Dear John,
I once wrote a little macro that computes the Agresti-Coull interval.
You can find it here:
http://www.oliverkuss.de/science/software/binomci.sas

Hope you find it useful,
Oliver
 | 
Pages: 1
Prev: Invalid argument to function QUOTE
Next: SAS and 64 bit systems