Re: How to nullified the correlation

On 19 Jan 2006 00:48:27 -0800, "RAMS" <ramsathish@xxxxxxxxx> wrote:

> Hai all,
>
> I have a data on eye. Both the eyes were affected by Cataract
> at alomst same stage. Half of the patients were treated by a control
> drug and half of the patients were treated by a newly founded drug.
>
> Here our objective is whether the newly founded drug is
> effective over the control drug or not.
>
> If we take both eyes information then we have a problem of lack
> of independence. If we randomly take one eye information then we dont
> know how much it express the sample.
>
> So i wish to take both eyes information and to nullify the
> correlation between the eyes. Anyone please help me.

I think you do not have important information on
each eye, so you do not have any reason not to collapse
information before-hand, in a way that seems sensible to
you -- instead of letting a procedure collapse by its default.

How to collapse? Usually both eyes (I guess) will be the
same, so there is not a problem for them. For the rest,
you could use the average, or the better, or the worse.
Presumably, that will make no difference to the test's result.

If there are *many* where the outcomes differed, that is
probably worth special comment in the written report.

How many ordinal categories (levels) were there?

Are there approximately equal steps between levels?
- The ends might be different: "death" or "totally cured"
might be special outcomes that deserve special attention,
or analyses of *that one* alone.
- The middle might have too many distinctions, resulting
in very low counts. If there were any categories that have
zero (or nearly zero) for counts, they might be absorbed
by a neighbor before assigning integers to the levels.

As someone suggested, you can test on rank-transformations
as a confirmation that the scoring does not make a big difference.

Hope this helps.

--
Rich Ulrich, wpilib@xxxxxxxx
http://www.pitt.edu/~wpilib/index.html
.

Relevant Pages

  • Re: Glyburde 5mg
    ... day with food, breakfast and supper, I am experencing bluriness with my ... The doctor said they are side effects of this drug and might last ... Hopefully this is due to your eyes slowly clearing from your current ...
    (alt.support.diabetes)
  • Re: Glyburde 5mg
    ... day with food, breakfast and supper, I am experencing bluriness with my ... The doctor said they are side effects of this drug and might last ... Hopefully this is due to your eyes slowly clearing from your current ...
    (alt.support.diabetes)
  • Update on Avastin and Lucentis in the Untited Kingdom
    ... A major drug company is blocking access to a medicine that is cheaply ... injecting tiny quantities of a colon cancer drug called Avastin into ... eyes at a higher cost. ... "The drug company hasn't applied for it to be licensed for this use. ...
    (sci.med.vision)
  • Re: Any Dune fans?
    ... > The Dune is all about "spice", a kind of drug, and where it comes from. ... that's why the eyes got Photoshopped to blue. ... Christine in Laitila, Finland ... photos: http://s208.photobucket.com/albums/bb108/christal63/ ...
    (rec.pets.cats.anecdotes)