varContrasts: Set Factor Contrasts

In lmSupport: Support for Linear Models

Description

Calculates contrast matrix for a specified contrast type. Options include DUMMY, POC, HELMERT, EFFECTS

Usage

varContrasts(TheFactor, Type = "DUMMY", RefLevel = length(levels(TheFactor)), 
       POCList = NULL, Labels = NULL)

Arguments

TheFactor	factor from dataframe
Type	type of contrast, Options include DUMMY (default), POC, HELMERT, or EFFECTS
RefLevel	Reference level for contrast. Only applies to DUMMY, HELMERT, and EFFECTS. For DUMMY: RefLevel is numeric index of control/reference category (i.e. coded 0 for all regressors). For HELMERT: RefLevel = 1 indicates reverse HELMERT (i.e., last vs. earlier, second to last vs. earlier, etc), RefLevel = 'Highest Level' indicates forward HELMERT (i.e., first vs. later, second vs. later, etc). For EFFECTS: RefLevel is numeric index of excluded level.
POCList	if Type = POC, a list of Contrasts is required in POCList; e.g., list(c(1,0,-1), c(-1,2,-1)). Best to provide as whole numbers. Function will re-scale to unit weighted contrasts.
Labels	if Type = POC, Labels can be provided. If NULL (Default), contrast labels are POC1, POC2, etc.

Details

Use the contrast matrix with contrasts() to set contrast for a specific factor in dataframe.

Value

Returns contrast matrix for indicated type of contrast.

Author(s)

John J. Curtin [e-mail]

See Also

contrasts

Examples

  d = data.frame(f=factor(c('f1', 'f2', 'f3')))
  contrasts(d$f)
  
  ##set as DUMMY with last category as reference
  contrasts(d$f) = varContrasts(d$f, Type='DUMMY', RefLevel = 3) 
  
  ##set as POC with user defined labels
  contrasts(d$f) = varContrasts(d$f, Type='POC', POCList = list(c(2,-1,-1),c(0,1,-1)), 
                   Labels = c('f1_v_f2f3', 'f2_v_f3')) 
                   
  ##set as reverse HELMERT                 
  contrasts(d$f) = varContrasts(d$f, Type='HELMERT', RefLevel = 1) 
  
  ##set as EFFECTS, excluding f3 vs. grand mean contrast
  contrasts(d$f) = varContrasts(d$f, Type='EFFECTS', RefLevel = 3) 

Example output

Registered S3 methods overwritten by 'lme4':
  method                          from
  cooks.distance.influence.merMod car 
  influence.merMod                car 
  dfbeta.influence.merMod         car 
  dfbetas.influence.merMod        car 
   f2 f3
f1  0  0
f2  1  0
f3  0  1
   f1_v_f3 f2_v_f3
f1       1       0
f2       0       1
f3       0       0
    f1_v_f2f3 f2_v_f3
f1  0.6666667     0.0
f2 -0.3333333     0.5
f3 -0.3333333    -0.5
   f3_v_Earlier f2_v_Earlier
f1   -0.3333333         -0.5
f2   -0.3333333          0.5
f3    0.6666667          0.0
   f1_v_Mean f2_v_Mean
f1         1         0
f2         0         1
f3        -1        -1

lmSupport documentation built on May 2, 2019, 2:14 p.m.