This index calculates a value such that a certain fraction of difference between methods will be numerically smaller than this. The TDI is a measure which esentially is a number K such that the interval [-K,K] contains the limits of agreement.
TDI(y1, y2, p = 0.05, boot = 1000, alpha = 0.05)
A list with 3 components. The names of the list are preceeded by the criterion percentage, i.e. the percentage of the population that the TDI is devised to catch.
The numerically computed value for the TDI. If boot
is
numeric, a vector of median and a bootstrap c.i. is appended.
The approximate value of the TDI
Limits of agreement
Measurements by one method.
Measurements by the other method.
The fraction of items with differences numerically exceeding the TDI
If numerical, this is the number of bootstraps. If FALSE
no confidence interval for the TDI is produced.
1 - confidende degree.
Bendix Carstensen, bendix.carstensen@regionh.dk
If boot==FALSE
a single number, the TDI is returned.
If boot
is a number, the median and the 1-alpha
/2 central interval
based on boot
resamples are returned too, in a named vector of length 4.
LI Lin: Total deviation index for measuring individual agreement with applications in laboratory performance and bioequivalence, Statistics in Medicine, 19, 255-270 (2000)
data(plvol)
pw <- to.wide(plvol)
with(pw,TDI(Hurley,Nadler))
Run the code above in your browser using DataLab