Thanks Nick, I've been using anova to study this problem as well - what I'm
looking to do is find something like a slope and intercept that you'd get
from a regression to describe the metric from assay1 as a function of
assay2, but with a confidence interval based on the observed variation of
the measurements in the two assays.
In other words, the _averages_ of the two assays are indeed paired for each
batch observation, but the relative variance of the measurements differ.
-----Original Message-----
From: Nick Cox [mailto:[email protected]]
Sent: Tuesday, May 11, 2004 1:20 PM
To: [email protected]
Subject: st: RE: RE: RE: unpaired regression
Your problems looks to me like -anova-,
the flavor depending on what "separate"
means. It is not regression without
pairing. I don't know what "unpaired regression"
would be.
Nick
[email protected]
> -----Original Message-----
> From: [email protected]
> [mailto:[email protected]]On Behalf Of
> Wallace, John
> Sent: 11 May 2004 20:57
> To: '[email protected]'
> Subject: st: RE: RE: unpaired regression
>
>
> Can anyone comment on whether Scott's suggestion would be
> appropriate for
> the problem I'm working on? The difference in R^2 between the samples
> indicates that it might be problematic.
>
> John Wallace | Research Associate | Test Method Development
> AFFYMETRIX, INC. | 3380 Central Expressway | Santa Clara, CA
> 95051 | Tel:
> 408-731-5574 | Fax: 408-481-0435
>
> -----Original Message-----
> From: Wallace, John [mailto:[email protected]]
> Sent: Monday, May 10, 2004 10:08 PM
> To: '[email protected]'
> Subject: st: RE: RE: unpaired regression
>
> Doesn't that imply a relationship between the observations
> though? Wouldn't
> it be equally valid to end up with them lined up like
> +-------------------------+
> | batch assay1 assay2 |
> |-------------------------|
> 1. | Btch1 5400 .905 |
> 2. | Btch1 5320 .898 |
> 3. | Btch1 5670 .9 |
> 4. | Btch2 8600 .943 |
> 5. | Btch2 7840 .955 |
> 6. | Btch2 7550 .962 |
>
> In the original line-up, the coefficient of determination is
> 0.968. In the
> second one above, its 0.8.
>
>
> -----Original Message-----
> From: Scott Merryman [mailto:[email protected]]
> Sent: Monday, May 10, 2004 6:42 PM
> To: [email protected]
> Subject: st: RE: unpaired regression
>
> How about lining up the measurements?
>
> Something like
>
> . l
>
> +-------------------------+
> | batch assay1 assay2 |
> |-------------------------|
> 1. | Btch1 5400 . |
> 2. | Btch1 5320 . |
> 3. | Btch1 5670 . |
> 4. | Btch1 . .9 |
> 5. | Btch1 . .905 |
> |-------------------------|
> 6. | Btch1 . .898 |
> 7. | Btch2 8600 . |
> 8. | Btch2 7840 . |
> 9. | Btch2 7550 . |
> 10. | Btch2 . .962 |
> |-------------------------|
> 11. | Btch2 . .955 |
> 12. | Btch2 . .943 |
> +-------------------------+
>
> . by batch: replace assay2 = assay2[_n +3]
> (12 real changes made, 6 to missing)
>
> . drop if assay1 == .
> (6 observations deleted)
>
> . l
>
> +-------------------------+
> | batch assay1 assay2 |
> |-------------------------|
> 1. | Btch1 5400 .9 |
> 2. | Btch1 5320 .905 |
> 3. | Btch1 5670 .898 |
> 4. | Btch2 8600 .962 |
> 5. | Btch2 7840 .955 |
> |-------------------------|
> 6. | Btch2 7550 .943 |
> +-------------------------+
>
>
> Scott
>
> ________________________________________
> From: [email protected]
> [mailto:[email protected]] On Behalf Of
> Wallace, John
> Sent: Monday, May 10, 2004 7:48 PM
> To: '[email protected]'
> Subject: st: unpaired regression
>
> I have two measures of batch performance on which I'd like to
> perform a
> regression.� The measurements are taken on separate samples
> from the batch,
> and typically look something like:
> ����������� Assay1 Assay2
> Btch1��� 5400���� ����
> Btch1��� 5320���� ����
> Btch1��� 5670���� ����
> Btch1��������������� 0.900
> Btch1��������������� 0.905
> Btch1��������������� 0.898
> Btch2��� 8600���� ����
> Btch2��� 7840���� ����
> Btch2��� 7550���� ���
> Btch2��������������� 0.962
> Btch2��������������� 0.955
> Btch2��������������� 0.943
> ...etc (on for multiple batches which show correlated
> measures for the two
> assays)
> -collapse- ing them to batch averages and then performing the
> regression is
> one approach, but it doesn't take variance of the measures
> themselves into
> account in the regression.� Is there a system for performing
> this type of
> analysis?
> �
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
