On Feb 24, 2011, at 4:50 PM, Denis Kazakiewicz wrote:
> Dear R people
> Could you please help with following
>
> Trying to compare accuracy of tumor size evaluation by different
> methods. So data looks like
>
> id true metod1 method2 ...
> 1 2 2 2.5
> 2 1.5 2 2
> 3 2 2 2
>
> etc.
>
> Could you please give a hint how to deal with that.
> Seems like {merror} does not suite to me because I am trying to compare
> accuracy of measurements with their true known values not just overall
> agreement of methods.
> Moreover sample size is ridiculously small (33 patients) so ANOVA is not
> much of help (or is it?)
> Any suggestions, hints and even guesses are highly appreciated. I am
> stuck a bit.
Denis,
I would suggest that you start here:
http://www-users.york.ac.uk/~mb55/meas/meas.htm
This covers various resources pertaining to the design and analysis of
measurement studies, primarily based upon methods by Bland and Altman.
HTH,
Marc Schwartz