drawDD: Draw a double dissociation based on regression coefficients
In axzinker/axBoost: axBoost: A collection of useful functions

drawDD

R Documentation

Draw a double dissociation based on regression coefficients

Description

Draws a double dissociation based on regression coefficients.

Usage

drawDD(
  data,
  predUpper,
  predLower,
  critUpper,
  critLower,
  covUpper = NULL,
  covLower = NULL,
  covResidAsDep = FALSE,
  plotCov = TRUE,
  moderator = NULL,
  moderatorSD = c(-1, 1),
  modLabSeq = 1,
  plotRange = 1,
  modBothPaths = FALSE,
  robust = FALSE,
  printReg = FALSE,
  title = ""
)

Arguments

`data`	A data frame containing the numeric columns predUpper, predLower, critUpper, critLower, and optionally covariates and a moderator.
`predUpper`	String with the name of the predictor of the upper double dissociation path.
`predLower`	String with the name of the predictor of the lower double dissociation path.
`critUpper`	String with the name of the criterion of the upper double dissociation path.
`critLower`	String with the name of the criterion of the lower double dissociation path.
`covUpper`	Optional string vector with the names of the covariates for the criterion of the upper double dissociation path.
`covLower`	Optional string vector with the names of the covariates for the criterion of the lower double dissociation path.
`covResidAsDep`	If FALSE (the default), a multiple regression is computed. If TRUE, the residuals of the covariate regression are used as dependent variable.
`plotCov`	Include covariates in the lower part of the plot. Default is TRUE, but can be switched off (e.g., if variable names are too long)
`moderator`	String with the name of the Moderator of the upper and lower direct path from predictor to criterion
`moderatorSD`	Defines the standard deviations for which the slope lines are plotted. Default is -1 and 1. For example, for experimetal group / control group design use c(0,1).
`modLabSeq`	Moderation label sequence (1 or 2). Cope with the lm() function, which puts out the labels in 2 different orders pred:mod (which is 1), or mod:pred (which is 2). Default is 1.
`plotRange`	Plot range of the square plot of a (marginally) significant interaction (only works with default of 1 SD at the moment).
`modBothPaths`	If TRUE, a moderator effect is computed for both paths. Default is FALSE.
`robust`	Logical value if robust parameter estimation is used (robust::lmRob() and WRS2::pbcor()). Default is FALSE.
`printReg`	Print the summary () of the regression coefficients. Default is FALSE.
`title`	Title of plot

Details

Assuming a double dissociation, the function computes two regressions:

critUpper ~ predUpper + predLower

critLower ~ predLower + predUpper

If covariates are givem, the function is computed as:

critUpper ~ covUpper + predUpper + predLower

critLower ~ covLower + predLower + predUpper

If covResidAsDep is false, this is done as a multiple regression. If it is true, this is done for the residuals of the covUpper/covLower regressions as dependent variable. It works also in combination with moderators.

If a moderator is given, the regressions are computed as:

critUpper ~ predUpper * moderator + predLower

critLower ~ predLower * moderator + predUpper

If modBothPaths = TRUE, the regressions are computed as:

critUpper ~ predUpper * moderator + predLower * moderator

critLower ~ predLower * moderator + predUpper * moderator

All variables are scaled before computing the regressions to cope with multicolinearity in the moderated case.