sostools

This is just a collection of functions I use in my research that I use frequently. Other functions I use infrequently can be found in my snippets repo. Since a lot of this is stuff that I use for my personal work, it’s not guaranteed to work flawlessly in other contexts.

Installation

You can install the development version from GitHub with:

# install.packages("devtools")
devtools::install_github("tsostarics/sostools", build_vignettes = TRUE)

Easily reference coefficients in prose

Here is a simple example of referencing the values of a model summary without needing to hand write or format it yourself. See ?get_coef for more information. make_coef_getter is a function factory that provides a shorthand if you have multiple models you’re working with.

library(sostools)
library(tidyverse)
mdl <- lm(mpg ~ cyl * gear, data = mtcars) # Create model
coefs <- enlist_coefs(mdl) # Create coefficients
get_coef1 <- make_coef_getter(mdl) # Using the model object directly
get_coef2 <- make_coef_getter(coefs) # Using the list of coefficients we made
get_coef("cyl:gear", coefs)
#> $(\hat\beta = -0.67, t = -1.27, s.e. = 0.53, p=0.213)$
get_coef1("cyl:gear")
#> $(\hat\beta = -0.67, t = -1.27, s.e. = 0.53, p=0.213)$
get_coef2("cyl:gear")
#> $(\hat\beta = -0.67, t = -1.27, s.e. = 0.53, p=0.213)$

Here we can get the formatted values of a write up easily with in-line R code. You could say something like: We found no significant effect of cylinder (β̂=−0.18,t=−0.09,s.e.=2.05,p=0.931) or gear (β̂=5.14,t=1.41,s.e.=3.63,p=0.168) nor a significant interaction between the two (β̂=−0.67,t=−1.27,s.e.=0.53,p=0.213).

You can also use these with Bayesian models fit with brms, which will give you the β estimate as well as the credible interval.

Contrast coding functionality

This functionality has been exported to a standalone package, please visit tsostarics/contrastable