elisr – Exploratory Likert Scaling in R

Description

An alternative to Exploratory Factor Analysis (EFA) for metrical data in R. Drawing on characteristics of classical test theory, Exploratory Likert Scaling (ELiS) supports the user exploring multiple one-dimensional data structures. In common research practice, however, EFA remains as go-to method to uncover the (underlying) structure of a data set. Orthogonal dimensions and the potential of overextraction are often accepted as side effects. ELiS confronts these problems. As a result, elisr provides the platform to fully exploit the exploratory potential of the multiple scaling approach itself.

The two workhorses: disjoint() & overlap()

elisr comes with two user functions: disjoint() and overlap().

• disjoint() searches highly consistent scale fragments in the haystack of variables.
• overlap() enhances disjoint()’s fragments – selecting items so that they can overlap across fragments.

With a typical case in mind, the practical difference between them is: disjoint() is set up to produce sharp and disjoint scale fragments. Sharp and disjoint fragments feature a high internal consistency. Thus, items within such a fragment share a strong linear relationship with each another. The thing with disjoint() is, it allocates any item to a particular fragment. This is whereoverlap() steps in. Passing fragments to overlap(), the function’s underlying algorithm tries to enrich each fragment. The emerging scales are flavored with items from your specified data frame, but the algorithm ignores those that are already built into a fragment (step 1). Later on, we will talk about the inclusion criterion in greater detail. To get to the point: Using overlap() an item can appear in more than one of the enriched fragments. In doing so, we overcome the splitting effect induced by disjoint(). These basic principles will unfold one step at a time in the companion.

Note: The last section is part of elisr’s vignette. If you are interested, you can read on there.

Install from GitHub (development version)

There are multiple ways to get elisr. I’ll show you three, sorted by different levels of R expertise (pro, skilled and novice). If you don’t understand a given installation instruction – move on to the next.

Pro

If you are an advanced R user simply download elisr from github (e.g., with devtools).

Skilled

I stick with devtools and install_github() to install elisr, but feel free to use whatever you like.

devtools::install_github(sbissantz/elisr)

Novice

To install the development version, copy and paste the following snippet into your R console. You will be guided through the installation process. What the snippet does, is (1) to check if the R package devtools is available on your system and if not (2) asks if you want to install it. If so, (3) it installs elisr via devtools’ function install_github(). After the installation you need to load and attach elisr. Simply type library(elisr) and elisr warmly welcomes you.

if (!requireNamespace("devtools", quietly = TRUE)) {
  msg <- "devtools is not installed, want to install it? Type 'yes' or 'no'."
  answer <- readline(prompt = message(msg))
  switch(answer,
         yes = {
           install.packages("devtools")
           devtools::install_github("sbissantz/elisr")
         },
         no = {
           stop("devtools is required to proceed the installation of elisr.",
                 call. = FALSE)
         },
         stop("Please answer 'yes' or 'no'.")
  )
} else {
  devtools::install_github("sbissantz/elisr")
}