RelRep: Reliability Representativeness of Coefficient Alpha or Omega

In melissagwolf/dynamic: DFI Cutoffs for Latent Variable Models

Reliability Representativeness of Coefficient Alpha or Omega

Description

This function evaluates how well a reliability summary index like alpha or omega represents the conditional reliability of a distribution of composite scores. It compares the conditional reliability function to a summary index and outputs a representativeness plot, a table of representativeness indices,and the full conditional reliability table for each possible sum score.

Usage

RelRep(
  data,
  items = c(names(data)),
  rel = "alpha",
  missing = "NA",
  method = "CI",
  width = NULL,
  raw.low = NULL,
  raw.high = NULL
)

Arguments

data	The original data to which the model was applied.
items	Column names of the items on the scale being evaluated (entered as strings). If omitted, all variables in the data will be used.
rel	Reliability coefficient to analyze. Options are "alpha" (the default) or "omega".
missing	The missing data indicator in the data. Not needed in R, only present to simply use of this function in a Shiny application.
method	how the test interval is created. Options are "CI" (the default), "width", or "raw". "CI" uses a 95 "width" builds an interval using a predetermined relative distance from the reliability coefficient (e.g., .05 from alpha). "raw" builds an interval using a predetermined raw values (e.g., .70 to .90)
width	Only required if method="width". Specifies a predetermined relative distance from the coefficient to each bound of the interval. The total width of the interval will be twice this value (e.g., if .05 is entered, the total interval width is .10 because it will span .05 above the coefficient and .05 below the coefficient)
raw.low	Only required if method="raw". Manually specifies the lower bound of the test interval. Must be between 0 and 1.
raw.high	Only required if method="raw". Manually specifies the upper bound of the test interval. Must be between 0 and 1.

Value

Conditional reliability table and Reliability Representativeness plot and table.

Author(s)

Daniel McNeish & Denis Dumas

Maintainer: Daniel McNeish <[e-mail]>

Examples

# "Example" dataset has 12 items on a 1-5 Likert scale

#Example using the first 8 items in "Example" for testing coefficient alpha with a 95% Bayes credible interval
ex1<-RelRep(data=Example,
items=c("X1","X2","X3","X4","X5","X6","X7","X8"),
rel="alpha",
method="CI")

#Example using odd items in "Example" to build a test interval that is .05 above and below coefficient omega
ex2<-RelRep(data=Example,
items=c("X1","X3","X5","X7","X9","X11"),
rel="omega",
method="width",
width=.05)

#Example using even items in "Example" to specify a test interval for coefficient alpha between 0.70 and 0.80
ex3<-RelRep(data=Example,
items=c("X2","X4","X6","X8","X10","X12"),
rel="alpha",
method="raw",
raw.low=.70,
raw.high=.80)

melissagwolf/dynamic documentation built on June 29, 2024, 6:24 p.m.