dummy.code {psych}

Create dummy coded variables

Description

Given a variable x with n distinct values, create n new dummy coded variables coded 0/1 for presence (1) or absence (0) of each variable. A typical application would be to create dummy coded college majors from a vector of college majors. Can also combine categories by group. By default, NA values of x are returned as NA (added 10/20/17)

Usage

dummy.code(x,group=NULL,na.rm=TRUE,top=NULL,min=NULL)

Arguments

x

A vector to be transformed into dummy codes

group

A vector of categories to be coded as 1, all others coded as 0.

na.rm

If TRUE, return NA for all codes with NA in x

top

If specified, then just dummy code the top values, and make the rest NA

min

If specified, then dummy code all values >= min

Details

When coding demographic information, it is typical to create one variable with multiple categorical values (e.g., ethnicity, college major, occupation). dummy.code will convert these categories into n distinct dummy coded variables.

If there are many possible values (e.g., country in the SAPA data set) then specifying top will assign dummy codes to just a subset of the data.

If using dummy coded variables as predictors, remember to use n-1 variables.

If group is specified, then all values of x that are in group are given the value of 1, otherwise, 0. (Useful for combining a range of science majors into STEM or not. The example forms a dummy code of any smoking at all.)

Value

A matrix of dummy coded variables

Author(s)

William Revelle

Examples

new <- dummy.code(sat.act$education)
new.sat <- data.frame(new,sat.act)
round(cor(new.sat,use="pairwise"),2)
#dum.smoke <- dummy.code(spi$smoke,group=2:9)
#table(dum.smoke,spi$smoke)
#dum.age <- dummy.code(round(spi$age/5)*5,top=5)  #the most frequent five year blocks