dm2icc.bt: Bootstrap Confidence intervals for dbICC
In wtagr/dbicc: Generalized reliability based on distances

Description Usage Arguments Value Author(s) References See Also Examples

View source: R/dm2icc.bt.R

Nonparametric bootstrapping can be used to construct confidence intervals for the Distance-based Intraclass Correlation Coefficient (dbICC) based on samples of the subjects with replacement.

1	dm2icc.bt(dmat, nsub, nmea, nB, probs = c(0.025, 0.975), adhoc = TRUE)

`dmat`	a distance matrix or an object of `dist`, of dimension `sum(nmea)*sum(nmea)`. Note that the structure of the distance matrix, with the rows or columns is grouped by subjects or individuals. The details refer to Figure 4 and Table 2 of Xu el at.(2020).
`nsub`	number of subject or individual.
`nmea`	a vector containing number of the measurement for each subject or individual; if `nmea` is a scalar, it means each subject shares the same number of the measurement.
`nB`	number of bootstrap.
`probs`	a vector of probabilities with values in [0,1]. `c(.025, .975)` is default.
`adhoc`	a logical variable, whether to apply the ad hoc correction when estimating the dbICC from a bootstrap sample. Default is `TRUE`.

estimates of underlying dbicc sample quantiles in probs. 95

Meng Xu mxu@campus.haifa.ac.il

Xu, M., Reiss, P. T., and Cribben, I. (2020). Generalized reliability based on distances. Biometrics, to appear. https://arxiv.org/abs/1912.07137.

plotdmat,dm2icc

##Point estimates of dbICC

# Generation function for R^2 points from multi-normal distribution
R2gen<-function(nsub,nmea,m=1,variance,sds=NULL,pt=FALSE){
    if (is.null(sds)==FALSE) set.seed(sds)
    if (length(nmea)==1) nmea<-rep(1,nsub)*nmea
    sig1<-diag(rep(variance,2))
    mu<-c(0,0)
    sig2<-diag(rep(1,2))
    t<-MASS::mvrnorm(nsub,mu,sig2)
    e<-MASS::mvrnorm(sum(nmea),mu,sig1/m)
    p<-matrix(apply(t,2,rep,times=nmea),ncol=2)+e#(I*J)x2
    if (pt==TRUE) return(list(t=t,p=p))
    if (pt==FALSE) return(p)
}

# set the number of the point
I <- 10

# set the number of the measurement for each point
J <- 4

# generate the sample of R^2 points
varl <- .25 # variance of the 2-d normal distribution
pij <- R2gen(I,J,variance=varl)

# calculate the squared distance matrix via Euclidean distance
distmat<-as.matrix(dist(pij))

##Bootstrap Confident intervals

dm2icc.bt(distmat, I, J, 500)