nullsim: Simulate null hypothesis (size differences only)

In bbolker/cpcbp: common principal components/back-projection analysis

Description

Analyze multivariate data and simulate (a) null set(s) of data assuming that the different groups have a common first principal component and that their means differ only along the CPC1 axis; extract F statistics

Usage

nullsim(x, f, P1, plot.it = FALSE)
nullFstats(x,f,n=1,bpanova=FALSE)

Arguments

x	a numeric data matrix
f	grouping variable (factor)
P1	the result of running phillips.cpc on the data; if missing, will be recomputed
plot.it	produce a descriptive plot? (STUB)
n	number of replicates
bpanova	(logical) use bp.anova to compute F statistics (alternatively use anova on back-projected data, without BP error correction)

Value

nullsim: a numeric matrix the same shape as x. nullFstats: a matrix of replicate F statistics for the null case.

Author(s)

Ben Bolker

Examples

  s1 = simdata(offset=10,offset2=10)
  plot_multigrp(s1)
  s1n = nullsim(s1$data,s1$f)
  plot_multigrp(s1n,s1$f)

bbolker/cpcbp documentation built on May 11, 2019, 9:28 p.m.