Report results from normality tests

This repository provides two functions report_ols() report_normality() to automatically generate the results from the most common normality tests as text. If you have questions, feedback, or further ideas, feel free to contact me.

Installation

The package is not yet published on CRAN. Nevertheless, you can install this package from Github. Type in the following lines of code:

install.packages("remotes")

remotes::install_github("huelemeier/report-normality-tests")

Load the package every time you start R:

library(reportnormalitytests)

Functions

The report_ols() works with:

ols_normality_test # Shapiro-Wilk normality test, Anderson-Darling normality test, one-sample Kolmogorov-Smirnov test and Cramer-von Mises normality test

The report_normality() is compatible with:

lillie.test() # Lilliefors (Kolmogorov-Smirnov) normality test
cvm.test() #Cramer-von Mises normality test
ad.test() #Anderson-Darling normality test
sf.test() # Shapiro-Francia normality test
shapiro.test() # Shapiro-Wilk normality test
pearson.test() # Pearson chi-square normality test
ks.test() # Kolmogorov-Smirnov test (one-sample and two-sample; alternatives: "two-sided", "less", "greater";  exact: TRUE, FALSE)

Examples

# report_ols() function
report_ols(ols_test_normality(iris$Sepal.Length))
We performed a Shapiro-Wilk normality test, and it showed the distribution departed significantly from normality (W = 0.98, p = .010). Based on the two-tailed Asymptotic one-sample Kolmogorov-Smirnov test, we have sufficient evidence to say our data came from normal distribution (D = 0.09, p = .189). We calculated a Cramer-von Mises test of goodness-of-fit test and it showed the distribution of our data departed significantly from the normal distribution (W = 50, p = .000). The Anderson-Darling normality test indicated the distribution of our data departed significantly from normality (W = 0.89, p = .023).



# report_normality() function
report_normality(lillie.test(iris$Sepal.Length))
The Lilliefors (Kolmogorov-Smirnov) normality test suggested iris$Sepal.Length departed significantly from normality (D = 0.09, p = .006).

report_normality(cvm.test(iris$Sepal.Length))
We calculated a Cramer-von Mises normality test, and it showed the distribution of iris$Sepal.Length departed significantly from the normal distribution (W = 0.13, p = .047).

report_normality(ad.test(iris$Sepal.Length))
We performed an Anderson-Darling normality test, and it showed the distribution of iris$Sepal.Length departed significantly from normality (W = 0.89, p = .023).

report_normality(sf.test(iris$Sepal.Length))
We performed a Shapiro-Francia normality test, and it showed the distribution of iris$Sepal.Length departed significantly from normality (W = 0.98, p = .026).

report_normality(shapiro.test(iris$Sepal.Length))
We performed a Shapiro-Wilk normality test, and it showed the distribution of iris$Sepal.Length departed significantly from normality (W = 0.98, p = .010).

report_normality(pearson.test(iris$Sepal.Length))
Based on the Pearson chi-square normality test of goodness of fit, the sample's distribution of iris$Sepal.Length did not match that of the population's (X2(12) = 17.4, p = .135).

report_normality(ks.test(iris$Sepal.Length, 'pnorm', exact = TRUE, alternative = "less"))
Based on the Exact one-sample Kolmogorov-Smirnov test, we have sufficient evidence to say iris$Sepal.Length came from our assumed distribution (D = 1, p = .000). We applied a one-tailed test assuming the cumulative distribution function lies below the null hypothesis.

report_normality(ks.test(iris$Sepal.Length, iris$Sepal.Width))
Based on the Asymptotic two-sample Kolmogorov-Smirnov test, we have sufficient evidence to say iris$Sepal.Length and iris$Sepal.Width did not come from our assumed distribution (D = 0.99, p = <.05). We applied a two-tailed test.