group_perc_diff_test: Group percentage difference test

In netjack: Tools for Working with Samples of Networks

Description

This function implements the group percentage difference test on a network statistic. This test assesses if the percent change in the network statistic due to the network manipulation is significantly different between groups. Percent change is calculated as the difference between the target and original statistic divided by the original statistic.

Usage

group_perc_diff_test(netSampleStatSet, grouping.variable,
  p.adjust = "BH", non.parametric = F)

Arguments

netSampleStatSet	Input NetSampleStatSet
grouping.variable	character name of sample level grouping variable
p.adjust	character string for requested multiple comparisons adjustment. Defaults to Benjamani-Hochberg
non.parametric	Logical. if true, test is performed using Wilcox test. If false, t-test. Defaults to false.

Details

If the sample has 2 groups, this test is performed using a t-test or Wilcox test. If the sample has 3 or more groups, the test is performed using a 1-way ANOVA, or Kruskal-Wallis test. Differences are tested at each network manipulation.

Value

A data frame containing original and adjusted p.values.

Examples

data(GroupA)
GroupA_Net = as_NetSample(GroupA, 1:20, node.variables = list(community = c(rep(1, 10), rep(2,10))),
  sample.variables = list(group = c(rep(1, 10), rep(2,10))))
Jackknife_GroupA_Net = net_apply(GroupA_Net, node_jackknife)
GlobEff_GroupA_Net = net_stat_apply(Jackknife_GroupA_Net, global_efficiency)
group_diff_test(GlobEff_GroupA_Net, grouping.variable = "group")

netjack documentation built on July 8, 2019, 1:02 a.m.