README.Rmd

---
output: github_document
---

<!-- README.md is generated from README.Rmd. Please edit that file -->

```{r, include = FALSE}
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "man/figures/README-",
  out.width = "100%"
)
```

# raoBust: Robust Rao (score) tests for Generalized Linear Models

<!-- badges: start -->
[![R-CMD-check](https://github.com/statdivlab/raoBust/workflows/R-CMD-check/badge.svg)](https://github.com/statdivlab/raoBust/actions)
[![codecov](https://codecov.io/github/statdivlab/raoBust/coverage.svg?branch=main)](https://app.codecov.io/github/statdivlab/raoBust)
<!-- badges: end -->

`raoBust`, at its core, gives you all the important information from `glm()`, but also with model misspecification-robust Rao tests (also called score tests) and Wald tests. 

Robust score tests have outstanding error rate performance in small samples, **and** when your data is not drawn from a parametric family (i.e., always). It's shocking to me (Amy) how well they perform. They have a reputation for being conservative in small samples, but I would argue that this is a *very good thing*. 

For now (because it's what I need for my work) I only have implementations for Poisson GLMs with log links. If you have another specific case you'd like to request, please let me know at [Issues](https://github.com/statdivlab/raoBust/issues) and label it as a "feature request". 

## Installation

You can install the development version of raoBust from [GitHub](https://github.com/) with:

``` r
# install.packages("devtools")
devtools::install_github("statdivlab/raoBust")
```

## Example

This is a really silly example to show you the syntax. It looks just like `glm()` on the input side, but outputs a summary table that includes robust Wald and Rao tests as well as others. 

```{r example}
library(raoBust)
glm_test(dist ~ speed, data = cars, family=poisson(link="log"))
```

## People

This software project was started by Amy D Willis, with input and insights from former lab member David S Clausen. Sarah Teichman is an amazing collaborator and coauthor. 

All errors are Amy's fault.