Type: Package
Title: Robust Coefficient Alpha and Omega with Missing and Non-Normal Data
Version: 0.7.2
Date: 2023-08-26
Depends: rsem, lavaan
Author: Zhiyong Zhang and Ke-Hai Yuan
Maintainer: Zhiyong Zhang <johnnyzhz@gmail.com>
Description: Cronbach's alpha and McDonald's omega are widely used reliability or internal consistency measures in social, behavioral and education sciences. Alpha is reported in nearly every study that involves measuring a construct through multiple test items. The package 'coefficientalpha' calculates coefficient alpha and coefficient omega with missing data and non-normal data. Robust standard errors and confidence intervals are also provided. A test is also available to test the tau-equivalent and homogeneous assumptions. Since Version 0.5, the bootstrap confidence intervals were added.
License: GPL-2 | GPL-3 [expanded from: GPL]
LazyLoad: yes
NeedsCompilation: no
Repository: CRAN
Packaged: 2023-08-27 01:22:16 UTC; zzhang4
Date/Publication: 2023-08-27 06:50:07 UTC

Robust Coefficient Alpha and Omega with Missing and Non-Normal Data

Description

Cronbach's alpha and McDonald's omega are widely used reliability or internal consistency measures in social, behavioral and education sciences. Alpha is reported in nearly every study that involves measuring a construct through multiple test items. The package 'coefficientalpha' calculates coefficient alpha and coefficient omega with missing data and non-normal data. Robust standard errors and confidence intervals are also provided. A test is also available to test the tau-equivalent and homogeneous assumptions. Since Version 0.5, the bootstrap confidence intervals were added.

Details

Package: coefficientalpha
Type: Package
License: GPL

Author(s)

Zhiyong Zhang and Ke-Hai Yuan Maintainer: Zhiyong Zhang <zzhang4@nd.edu>

References

Zhang, Z. & Yuan, K.-H. (2014). Robust Coefficients Alpha and Omega and their Confidence Intervals: Methods and Software.

Robust Cronbach's alpha and McDonald's omega

Description

Calculate alpha or omega for a given data set.

Usage

alpha(y, varphi = 0.1, se = FALSE, test = TRUE, 
complete =FALSE, auxiliary = NULL, drop, silent = TRUE)
omega(y, varphi = 0.1, se = FALSE, test = TRUE, 
complete =FALSE, auxiliary = NULL, drop, silent = TRUE)

Arguments

y

Data

varphi

Downweight rate

se

Whether to estimate standard error. It might take significant more time if se is requested with many items.

test

Whether to test the assumption of alpha (tau equivalent test) or omega (homogeneous test).

complete

Calculate alpha/omega only based on complete data (listwise deletion)

auxiliary

Provide a matrix or data frame of auxiliary variables for missing data analysis.

drop

The row number of cases to be dropped from the analysis.

silent

Whether to print information of the analysis.

Author(s)

Zhiyong Zhang and Ke-Hai Yuan

References

Zhang, Z. & Yuan, K.-H. (2014). Robust Coefficients Alpha and Omega and their Confidence Intervals: Methods and Software.

Examples

data(example)

alpha(example, varphi=.01)
omega(example, varphi=.01)

Bootstrap se and CI for alpha and omega

Description

Bootstrap se and CI for alpha and omega.

Usage

bootstrap(y, type="omega", alpha=.95, nboot=1000, ci="bc", plot=FALSE, 
varphi=0, complete=FALSE, auxiliary=NULL, silent=FALSE)

Arguments

y

Data

type

omega: coefficient omega. Otherwise, coefficient alpha.

alpha

Confidence level.

nboot

Number of bootstrap samples to use

ci

bc: Bias-corrected CI. Otherwise, the percentile CI is used.

plot

Whether to plot the bootstrap density.

varphi

Downweight rate

complete

Calculate alpha/omega only based on complete data (listwise deletion)

auxiliary

Provide a matrix or data frame of auxiliary variables for missing data analysis.

silent

Whether to print information of the analysis.

Author(s)

Zhiyong Zhang and Ke-Hai Yuan

References

Zhang, Z. & Yuan, K.-H. (2014). Robust Coefficients Alpha and Omega and their Confidence Intervals: Methods and Software.

Examples

data(example)

boot.test<-bootstrap(example, type='alpha', nboot=10, plot=TRUE)

An example data set

Description

An example data set with 10 variables.

Usage

data(example)

Alpha/Omega related plot

Description

Generate plot related to alpha or omega. Three plot can be generated. (1) The weight plot will plot the weight associated with alpha calculation. (2) Profile plot will plot the cases with smallest weights and the average value. (3) The diagnostic plot plots the alpha according to different values of the tuning parameter.

Usage

## S3 method for class 'alpha'
plot(x, type="weight", profile=5, interval=0.01,
center=TRUE, scale=FALSE, w1=FALSE, numbered=FALSE, pos="topright",...)
## S3 method for class 'omega'
plot(x, type="weight", profile=5, interval=0.01,
center=TRUE, scale=FALSE, w1=FALSE, numbered=FALSE, pos="topright",...)

Arguments

x

Results from the function alpha or omega.

type

Three types of plots can be generated. type="weight" generates the plot of weight of each case. type="profile" generates a profile plot for the smallest weight. type="weight" generates the diagnostic plot for varphi with an interval of .01.

profile

Number of cases used on the profile plot. At most 10.

interval

The interval used in the diagnostic plot. The default is 0.01.

center

Whether to center the data in the profile plot.

scale

Whether to scale the data using variance parameters in the profile plot.

w1

Whether to plot the weight for means

numbered

Whether to number the profile plot

pos

Position of legend. If pos=NULL, no legend is plotted.

...

Options can be passed to the plot function.

Author(s)

Zhiyong Zhang and Ke-Hai Yuan

References

Zhang, Z. & Yuan, K.-H. (2014). Robust Coefficients Alpha and Omega and their Confidence Intervals: Methods and Software.

Examples

data(example)

res<-alpha(example, varphi=.01)
## diagnostic plot
plot(res, type='d')

## alpha with varphi=.01 & standard error
res<-alpha(example, varphi=.01, se=TRUE)
## confidence interval
summary(res)

## weight plot
plot(res)
# or
plot(res, type='w')

## profile plot
plot(res, type='p')

Print alpha/omega and its confidence interval.

Description

Print alpha/omega and its confidence interval.

Usage

## S3 method for class 'alpha'
summary(object, type = "raw", prob = 0.95,...)
## S3 method for class 'omega'
summary(object, type = "raw", prob = 0.95,...)

Arguments

object

Results from the function alpha or omega.

type

If not raw, the CI based on transformation will be provided. The transformed CI will be always in the range of [0,1].

prob

Alpha level for confidence interval. By default, the 95

...

Option can be passed to the summary function.

Author(s)

Zhiyong Zhang and Ke-Hai Yuan

References

Zhang, Z. & Yuan, K.-H. (2014). Robust Coefficients Alpha and Omega and their Confidence Intervals: Methods and Software.

Examples

data(example)

res<-alpha(example, varphi=.01)
## diagnostic plot
plot(res, type='d')

## alpha with varphi=.01 & standard error
res<-alpha(example, varphi=.01, se=TRUE)
## confidence interval
summary(res)

Test the tau-equivalence and the homogeneity of items

Description

Calculate alpha or omega for a given data set.

Usage

tau.test(y, varphi = 0.1, complete = FALSE, drop)

Arguments

y

Data

varphi

Downweight rate

complete

Calculate alpha/omega only based on complete data (listwise deletion)

drop

The row number of cases to be dropped from the analysis.

Author(s)

Zhiyong Zhang and Ke-Hai Yuan

References

Zhang, Z. & Yuan, K.-H. (2014). Robust Coefficients Alpha and Omega and their Confidence Intervals: Methods and Software.

Examples

data(example)
tau.test(example)