# RE: st: RE: Interobserver agreement for continuous variables

 From BISSERY Alvine To "'statalist@hsphsun2.harvard.edu'" Subject RE: st: RE: Interobserver agreement for continuous variables Date Fri, 18 Oct 2002 09:16:17 +0200

```Dear Nitin,

For your problem, you have to verify one hypothesis:
calculate the mean and standart deviation of  measures for each subject.
Then plot the sd versus the mean.
If there is no relation between mean and sd (you can test it with spearman
test), you have to do an ANOVA, with the variable "rater" as factor to
calculate inter and intra raters variability . ..

If your continuous variable is Y, then do:

loneway Y rater

It will give you the intraclass correlation coefficient, and the within
subject standart deviation

But you have a problem with your data: if you want to mesure inter rater
variability, I think that it would be better to have exactly the same rater
for all subjects. So maybe it would be better to keep only subjects wich
have the same raters.

When you will have done ANOVA, you will be able to calculate the limits of
acceptability (Bland  Altman), with the within subjcet standart deviation,
and then have an idea of inter and intra rater variability.

Hope it helps

Alvine

> -----Message d'origine-----
> De: Nitin Jain [mailto:drnitinjain@yahoo.com]
> Date: vendredi 18 octobre 2002 06:44
> À: statalist@hsphsun2.harvard.edu; n.j.cox@durham.ac.uk;
> srmillis@mindspring.com
> Objet: Re: st: RE: Interobserver agreement for continuous variables
>
>
> Hi Nick and Scott,
> Thank you for the response. But the concord and other
> tests contained in it are only for 2 observers/raters.
> I am looking for something that calculates
> interobserver agreement for a continuous variable WHEN
> THERE ARE MORE THAN 3 RATERS AND THE number of raters
> vary for each subject ?
> I look forward to hearing from you.
> Nitin
>
> --- Nick Cox <n.j.cox@durham.ac.uk> wrote:
> > Nitin Jain
> >
> > > 1. Is there a way in STATA to compute the
> > > interobserver/interrater agreement/variability for
> > a
> > > continuous variable (equivalent of the kappa
> > statistic
> > > for categorical variable)?
> >
> > [R] kappa points to -concord- as a contribution in
> > this territory.
> > -findit concord- shows the historical trail. Pick
> > the latest software, although the first write-up is
> > the fullest.
> >
> > In addition to the graphs and measures computed
> > by -concord-, let me add a reference to the
> > -vplplot-
> > program on SSC and the really excellent paper
> >
> > McNeil, D.R. 1992. On graphing paired data.
> > American Statistician 46, 307-11.
> >
> > which lies behind it.
> >
> > Nick
> > n.j.cox@durham.ac.uk
> >
> > *
> > *   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/
>
>
> =====
> Nitin Jain, MBBS, MSPH
> Research Associate,
> Duke University, Durham, NC 27710
> (919) 684-5603
> School of Public Health
> University of North Carolina at Chapel Hill
> (919)968-9828
>
> __________________________________________________
> Do you Yahoo!?
> Faith Hill - Exclusive Performances, Videos & More
> http://faith.yahoo.com
> *
> *   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/
```