In statdivlab/rigr: Regression, Inference, and General Data Analysis Tools in R

A key feature of many exploratory analyses is obtaining descriptive statistics for multiple variables. In the rigr package, we provide a function descrip() with improved output for descriptive statistics for an arbitrary number of variables. Key features include the ability to easily compute summary measures on strata or subsets of the variables specified. We go through examples making use of these key features below.

Descriptive statistics with descrip

Throughout our examples, we'll use the fev dataset. This dataset is included in the rigr package; see its documentation by running ?fev.

## Preparing our R session
library(rigr)
data(fev)

First, we can obtain default descriptive statistics for the dataset simply by running descrip().

descrip(fev)

Since we input a dataframe, we can see that all variables have the same number of elements given in the N column. None of our variables have any missing values, as seen in the Msng column.

Rather than specifying the whole dataframe, if we are interested in only the variables fev and height, we can input only those two vectors into the descrip() function, as below.

descrip(fev$fev, fev$height)

Descriptive statistics for strata

Suppose we wish to obtain descriptive statistics of the fev and height variables, stratified by smoking status. To do this, we can use the strata parameter in descrip:

descrip(fev$fev, fev$height, strata = fev$smoke)

In the output, we can see that overall descriptive statistics, as well as descriptive statistics for each stratum (smoke = 1, smoke = 2) are returned in the table.

Descriptive statistics for subsets

Now suppose we only want descriptive statistics for height and FEV for individuals over the age of 10. We first create an indicator variable for age > 10 outside of the descrip() function, and then give this variable to the subset parameter.

greater_10 <- ifelse(fev$age > 10, 1, 0)
descrip(fev$fev, fev$height, subset = greater_10)

Above/Below

Suppose we want to know the proportion of individuals with FEV greater than 2, stratified by smoking status. We can use the strata argument as before, in addition to the above parameter to obtain this set of descriptive statistics:

descrip(fev$fev, strata = fev$smoke, above = 2)

From the output, we can see that 96.92% of the individuals in this dataset who smoke (smoking status 1) had an FEV greater than 2 L/sec, and 71.99% of the individuals in this dataset who were nonsmokers had an FEV greater than 2 L/sec.

statdivlab/rigr documentation built on June 23, 2024, 12:01 p.m.