Overall concordance correlation coefficient (OCCC)
Overall concordance correlation coefficient (OCCC) for agreement on a continuous measure based on Lin (1989, 2000) and Barnhart et al. (2002).
epi.occc(dat, na.rm = FALSE, pairs = FALSE) ## S3 method for class 'epi.occc' print(x, ...) ## S3 method for class 'epi.occc' summary(object, ...)
dat |
a matrix, or a matrix like object. Rows correspond to cases/observations, columns corresponds to raters/variables. |
na.rm |
logical. Should missing values (including |
pairs |
logical. Should the return object contain pairwise statistics? See Details. |
x, object |
an object of class |
... |
further arguments passed to |
The index proposed by Barnhart et al. (2002) is the same as the index suggested by Lin (1989) in the section of future studies with a correction of a typographical error in Lin (2000).
An object of class epi.occc with the following list elements (notation follows Barnhart et al. 2002):
occc: the value of the overall concordance correlation coefficient (rho.o^c),
oprec: overall precision (rho),
oaccu: overall accuracy (chi^a),
pairs: a list with following elements (only if pairs = TRUE, otherwise NULL;
column indices for the pairs (j,k) follow lower-triangle column-major rule
based on a ncol(x) times ncol(x) matrix),
ccc: pairwise CCC values (rho_jk^c),
prec: pairwise precision values (rho_jk),
accu: pairwise accuracy values (chi_jk^a),
ksi: pairwise weights (ksi_jk),
scale: pairwise scale values (v_jk),
location: pairwise location values (u_jk),
data.name: name of the input data dat.
Peter Solymos, solymos@ualberta.ca.
Barnhart H X, Haber M, Song J (2002). Overall concordance correlation coefficient for evaluating agreement among multiple observers. Biometrics 58: 1020 - 1027.
Lin L (1989). A concordance correlation coefficient to evaluate reproducibility. Biometrics 45: 255 - 268.
Lin L (2000). A note on the concordance correlation coefficient. Biometrics 56: 324 - 325.
## Generate some artificial ratings data: set.seed(1234) p <- runif(10, 0, 1) x <- replicate(n = 5, expr = rbinom(10, 4, p) + 1) rval <- epi.occc(dat = x, pairs = TRUE) print(rval); summary(rval)
Please choose more modern alternatives, such as Google Chrome or Mozilla Firefox.