sTabl3R: Quick Statistical Testing and Summary Tables in R

Quickly generate general statistical hypothesis tests and summary tables for demographic/phenotypic/clinical variables using a formatted data frame and a specified grouping variable. This project was motivated by a perceived need to accelerate data table production using some reasonable (albeit general) assumptions about the provided data.

In its current form, the package consists of a set of two major R functions to: (1) perform statistical analysis on a given data frame using a pre-specified grouping variable; and (2) generate tables using a combination of knitr::kable and flextable. Some additional functions (≥ v0.5.0) are available to allow the user to access results more easily and quickly identify categorical variables that may need to be recoded. See package documentation for additional details.

To install: The latest version can be found on GitHub, and installed using devtools.

if (!require(devtools)) install.packages("devtools")
devtools::install_github("CodeDepotIV/sTabl3R")

Or for the development version:

devtools::install_github("CodeDepotIV/sTabl3R", ref = "dev")

Trivia: While the name sTabl3R was chosen to signify its role as a statistics table utility in R, it was also done with an intentional nod to Hall of Fame NFL quarterback Ken Stabler.

Quick start:

library(sTabl3R)

data(mtcars)

# Generate a grouping variable
mtcars$Group <- as.factor(rep("GroupA", nrow(mtcars))) # Single group case
res1 <- generate_statistics(mtcars, group = "Group")
summary(res1)
generate_results_tables(res1)

data(mtcars)
res2 <- generate_statistics(mtcars, group = "cyl") # Multiple group levels
summary(res2)
generate_results_tables(res2)

Here's the deal:

The package is built around a central generate_statistics() function. This function function attempts to identify numeric (continuous and ordinal) and categorical variables within a user-supplied data frame.

For each identified numeric variable, Shapiro-Wilk testing is employed to check for normality. Provided that the user does not specify force_nonparametric = TRUE in the arguments, each numeric variable will be subjected either to parametric (t-test or one-way ANOVA) or non-parametric (Mann-Whitney U test or Kruskal-Wallis test) comparisons depending on the number of strata in the grouping variable. For instances where the grouping variable has a single level, only summary statistics will be generated.

For categorical variables, the function will apply either the Chi-squared test or the Fisher’s Exact test based on the expected frequencies in each contingency table, unless the grouping variable has a single level, in which case only summary tables will be generated.

FORMATTING: The data frame should be an R data.frame() object in which the first column is a list of unique identifiers. The group argument should identify a column in the data frame that will be used to distinguish the groups that are to be compared statistically. Please note that if unique identifiers are not present in the first column, unique identifiers will be appended to the dataframe.

The other functions include extract_stats() which allows the user to extract the statistical results for a variable of interest quickly, generate_results_tables() which generates a series of tables displaying all of the results generated following a call to generate_statistics(), and summary() functions to show which variables in the user input were used in the analysis and how they were categorized.

New features as they arrive:

v0.6.0.0 Added flag_high_cardinality() function to scan a dataset for categorical variables that may require recoding.