falsediscovery: p-value and confidence interval adjustment for multiple...

In wadetj/timsRstuff: Various utilities

p-value and confidence interval adjustment for multiple comparisons

Description

takes numeric vector of p-values and adjusts for multiple comparisons using false discovery rate (benjamini and hochberg 1995) and bonferroni (method="bon") optionally takes numeric vectors of estimates and standard errors and produces confidence intervals adjusted for multiple comparisons (benjamini et al. 2005)

Usage

falsediscovery(
  pvalues,
  qstar = 0.05,
  method = "fdr",
  dp = 5,
  ci = FALSE,
  ses = NULL,
  estimates = NULL
)

Arguments

pvalues	numeric vector of p-values
qstar	overall alpha level (default=0.05)
method	"fdr" (default) or "bon"
dp	rounded decimal places (default=5)
ci	FALSE (default) or TRUE to produce adjusted confidence intervals
ses	numeric vector of standard errors, required if ci=TRUE
estimates	numeric vector of estimates, required if ci=TRUE

Value

dataframe with starred p-values and optionally adjusted confidence intervals that meet adjusted significance level.

Examples

#from benjamini and hochberg 1995
pvals<-c(0.0001, 0.0004, 0.0019, 0.0095, 0.0201, 0.0278, 0.0298, 0.0344, 0.0459, 0.3240, 0.4262, 0.5719, 0.6528, 0.7590, 1.00)
falsediscovery(pvals)
falsediscovery(pvals, method="bon")

#from https://www.statisticshowto.com/benjamini-hochberg-procedure/
#note that p-values 0.039 and 0.041 are still considered significant even though higher than fdr
pvals<-c(0.205, 0.074, 0.060, 0.042, 0.041, 0.039, 0.008, 0.001, c(rep(0.3, 17)))
falsediscovery(pvals, qstar=0.25)

#example from regression model
library(datasets)
data(swiss)
otemp<-with(swiss, lm(Infant.Mortality~(Fertility+Agriculture+Examination+Education+Catholic)^2))
pvals<-summary(otemp)$coefficients[,4]
#remove intercept
pvals<-pvals[-1]
falsediscovery(pvals)
falsediscovery(pvals, qstar=0.2)
falsediscovery(pvals, method="bon")
#with adjusted CIs
betas<-summary(otemp)$coefficients[,1][-1]
stderrs<-summary(otemp)$coefficients[,2][-1]
falsediscovery(pvalues=pvals, ci=TRUE, estimates=betas, ses=stderrs)
falsediscovery(pvalues=pvals, ci=TRUE, estimates=betas, ses=stderrs, method="bon")

wadetj/timsRstuff documentation built on April 12, 2025, 5:54 p.m.