make.wide: Create data frame for Fitting Conjoint Measurment Models by...

In MLCM: Maximum Likelihood Conjoint Measurement

Create data frame for Fitting Conjoint Measurment Models by glm

Description

make.wide and make.wide.full generate a n x q - 1 matrix from an n x 2 column subset of a data frame storing the results of a conjoint measurement experiment, where n is the number of trials and q is the number of levels per dimension in the stimulus set tested. Currently, make.wide.full is limited to data sets with only 2 stimulus dimensions. The columns code covariates for all but the first stimulus level, which is constrained to be 0, along each dimension. These columns take the value 0 unless one of the stimuli in the trial corresponded to a level along that dimension, in which case it takes a 1 or a -1, depending on which of the two stimuli represented that level. If both stimuli represent the same level for a dimension, then they cancel out and the column contains a 0. This function is used for each dimension along which the stimuli vary to create a design matrix for each dimension. The final design matrix is constructed inside the mlcm method by putting together the design matrices from each dimension.

Usage

make.wide(d)

make.wide.full(d)

Arguments

d	a n x 2 column data frame. The columns give the indices of the levels of the dimensions along which the two stimuli presented in a trial vary.

Details

This is a helper function, normally used inside mlcm, and not typically exploited by the casual user.

Value

A data frame with n rows and q - 1 columns

D2–Dq

For each dimension along which the stimulus can vary, there are q - 1 columns coding the absence or presence of that level of the dimension in the stimulus. If the level is present, then the value is -1 or 1 as a function of which of the two stimuli contained that level, unless both do, in which case it is, also, 0.

Author(s)

Kenneth Knoblauch

MLCM documentation built on March 18, 2022, 7:31 p.m.