Re: Dixons up-and-down sequential allocation test

From: Bob Wheeler (bwheeler_at_echip.com)
Date: 11/19/04 

Date: Fri, 19 Nov 2004 11:54:22 -0500

There is a large literature, but tell me why I should take the time to
point it out to someone who hides behind a handle.

crina wrote:
> I am interested in the same problem. Did you ever found the answer?
> Please let me know if you did ASAP. CRINA
>
>
> On Wed, 6 Jun 2001 14:35:59 +0200, rothi wrote:
>
>>I'm doing some research to establish the ED50 of a cocktail of
>
> medication.
>
>>The sequential up-down method for small samples appeals to me,
>
> because it
>
>>substantially reduces the number of participants in the trial.
>>
>>However, in similar trials in which analysis is done by the
>
> up-and-down
>
>>method the number of participants is estimated by conservative power
>>analysis. The advantage of the small sample size is then gone.
>>
>>Is anyone familiar with the up-and-down procedure as described by
>
> Dixon and
>
>>Massey (Introduction to Statistical Analysis, 4th ed, McGraw-Hill)? I
>
> have
>
>>the following questions:
>>
>>1. Can you tell me when the procedure for small samples is
>
> appropriate.
>
>>2. What are the advantages of the procedures with large N's, compared
>
> to the
>
>>small sample variant.
>>
>>3. In Dixon's text book I did not find how to determine the
>
> confidence
>
>>interval for small samples.
>>Lichtman (1998) determines the 95% confidence interval in the test
>
> for small
>
>>samples as:
>>
>> 1.96 x dose increment x (SQR 2/n).
>>
>>in which n= the last n trials and 1.96 reflects the 0.05 alpha level.
>>
>>Is this equation correct? It would mean that the CI depends primarily
>
> on the
>
>>dose increment.
>>
>>Finally what is meant by the term "last n trials".
>>
>>Thank you for your help!
>>
>>Rothi
>
> 

-- 
Bob Wheeler --- http://www.bobwheeler.com/
         ECHIP, Inc. ---
Randomness comes in bunches.