detect.suppression: detect suppression effects in regression models

In eatRest: eatRest

Description

This function detects suppression effects in regression models.

Usage

detect.suppression ( dat , dependent , independent , full.return = FALSE , xlsx.path = NULL )

Arguments

dat	data.frame with data to be used
dependent	dependent variable in regression model
independent	character vector of independent variables in regression model
full.return	if FALSE a data.frame as a quadratic matrix with suppression effects (TRUE/FALSE) of independent variables is returned if TRUE a data.frame with all calculated terms ist returned
xlsx.path	full path of Excel file that results should be written to

Details

formulae (13.39a) and (13.39b) decribed in Bortz (1999) page 446 are used

if full.return=TRUE a data.frame is returned.

Columns are:

rownames: <dependent variable> ~ <independent variables> | <independent variable that is tested for suppression>

multiple.reg: logical, indicates wether there are 2 (FALSE) or more than 2 (TRUE) independent variables in the regression model

dep: dependent variabel in regression model

pred: independent variable that is investigated on suppression effect

preds: independent variables in regression model besides pred

cor_pred_c: correlation of pred and dependent variable

cor_pred_fitted_c: correlation of predicted pred by indepenent variables and dependent variable

r.sq_pred: R squared from model predicting pred by independent variables

rterm.minus: right term in formula (13.39a)

rterm.plus: right termn in formula (13.39b)

rterm.minus.diff: difference of rterm.minus and cor_pred_c

rterm.plus.diff: difference of cor_pred_c and rterm.plus

(positive difference of rterm.minus.diff or rterm.plus.diff indicates suppression effect)

rterm.minus.log: logical value of formula (13.39a)

rterm.plus.log: logical value of formula (13.39b)

suppression: logical, rterm.minus.log | rterm.plus.log

if full.return=FALSE a data.frame as quadratic matrix is returned:

rows and columns are independent variables

diagonal includes suppression for suppression effect of variable in multiple regression

triangles include suppression for bivariate independent variables, "row" suppresses "column"

Value

depends on options full.return

Author(s)

Martin Hecht

References

for formulae used by detect.suppression see

Bortz, J. (1999). Statistik fuer Sozialwissenschaftler. 5. Auflage. Berlin: Springer. p. 446

eatRest documentation built on May 2, 2019, 6:25 p.m.