trimbt.ccmc: One sample percentile bootstrap test with cluster correction...

In kevinprinsloo/CCMC: Cluster Based Correction For Multiple Comparisons

Description

One sample percentile bootstrap test with cluster correction for multiple comparisons

Usage

trimbt.ccmc(x, nullval = 0, tr = 0.2, alpha = 0.05, bt = TRUE,
  nboot = 599)

Arguments

x	A list or a matrix. In the first case x[1] contains the data for the first group, x[2] the data for the second group, etc. Length(x) = the number of groups = J. If stored in a matrix, columns correspond to groups.
nullval	The null value against which to compare the trimmed mean - default = 0.
tr	The amount of trimming - default 0.2. Set tr=0 for a t-test on means.
alpha	Alpha level - default 0.05.
bt	Set to TRUE to compute critical t values, p values and confidence intervals based on boostrap-t distributions, otherwise use standard calculations - default = TRUE
nboot	Number of bootstrap samples - default = 599

Details

One sample t-test on trimmed means with cluster correction for multiple comparisons

Value

A list of univariate t values, p values and confidence intervals, as well as cluster-based statistics.

estimate = mean or trimmed mean

ci = confidence interval

tval = t values

pval = p values

cluster.th = cluster threshold

cluster.map = vector of cluster IDs

cluster.sig = statistical significance based on cluster test

References

NA

Examples

# no effect: 
set.seed(21)
x <- matrix(rnorm(100), ncol = 5)
trimbt.ccmc(x,nullval=0,tr=0,alpha=.05,bt=FALSE,nboot=599)
# cluster of length 3:
set.seed(21)
x <- matrix(rnorm(100), ncol = 5)
x[,3:5] <- x[,3:5] + 1
trimbt.ccmc(x,nullval=0,tr=0,alpha=.05,bt=FALSE,nboot=599)
# use bootstrap-t thresholds
trimbt.ccmc(x,nullval=0,tr=0,alpha=.05,bt=TRUE,nboot=599)
# get cluster statistics for cluster 1:
out <- trimbt.ccmc(x,nullval=0,tr=0,alpha=.05,bt=TRUE,nboot=599)
c2sum <- sum(out$tval[out$cluster.map==1]^2)

kevinprinsloo/CCMC documentation built on May 14, 2020, 12:35 a.m.