contrastTest {linearModel}

Test Contrasts

Description

Contrast testing function. Designed to test contrasts of parameter estimates from a linear model.

Usage

contrastTest(
  estVec,
  n,
  dfModel,
  dfError,
  mse,
  C = NULL,
  test = c("scheffe", "bonferroni", "hsd", "lsd"),
  ...
)

Arguments

estVec

numeric vector of parameter estimates for comparison

n

numeric vector indicating the sample size for the parameter estimates, if a single value is given it is assumed to apply to all estiamtes

dfModel

numeric value for the model degrees of freedom

dfError

numeric value for the error or residual degrees of freedom

mse

numeric value for the mean squared error from the model

C

numeric matrix, each row is a contrast that should sum to zero, see details

test

character, indicating which testing method should be used, see details

...

currently ignored

Details

The test argument can be one of the following: 'scheffe','bonferroni','hsd', or 'lsd'. 'hsd' is the Tukey HSD test. 'lsd' is th Fisher LSD test. The other two are the Scheffe test and Bonferroni adjustment.

The matrix C is the contrast matrix. Each row is a separate contrast. The number of columns of C must be equal to the length(estVec). Row names for C are retained in the output, but they are not required.

Value

Object of class anova and data.frame

Examples

data(genericData)

mod <- lm(Y~A,data=genericData)
dfModel <- anovaTable(mod)['Model','df']
dfError <- anovaTable(mod)['Residual','df']
mse <- anovaTable(mod)['Residual','MS']
meanVec <- aggregate(Y~A,FUN=mean,data=genericData)$Y
n <- aggregate(Y~A,FUN=length,data=genericData)$Y

## can add names for ease of interpretation with the output
names(meanVec) <- c('group 1','group 2','group 3')
contrastTest(estVec=meanVec,n=n,dfModel=dfModel,dfError=dfError,mse=mse,test='hsd')

## each group vs the mean of the other two
C <- rbind(c(1,-0.5,-0.5),c(-0.5,1,-0.5),c(-0.5,-0.5,1))
## row names are not required but are helpful
row.names(C) <- c('1 vs 2+3','2 vs 1+3','3 vs 1+2')
contrastTest(estVec=meanVec,n=n,dfModel=dfModel,dfError=dfError,mse=mse,C=C,test='scheffe')