hypr

hypr: An R package for hypothesis-driven contrast coding - Published in JOSS (2020)

https://github.com/mmrabe/hypr

Repository

hypr is an R package for easy translation between experimental (null) hypotheses and contrast matrices as used for linear regression.

README.Rmd

---
title: "hypr"
output:
  md_document:
    variant: markdown_github
---
```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
library(hypr)
latexImg = function(latex){
  paste0('![](','https://render.githubusercontent.com/render/math?math=', URLencode(latex),')')
}
```
# hypr

[![CRAN](https://www.r-pkg.org/badges/version/hypr)](https://cran.r-project.org/package=hypr)
[![downloads](https://cranlogs.r-pkg.org/badges/hypr)](https://cran.r-project.org/package=hypr)
[![JOSS publication status](https://joss.theoj.org/papers/10.21105/joss.02134/status.svg)](https://joss.theoj.org/papers/10.21105/joss.02134)
[![DOI](https://zenodo.org/badge/208564895.svg)](https://zenodo.org/badge/latestdoi/208564895)
[![Package status](https://github.com/mmrabe/hypr/actions/workflows/r.yml/badge.svg)](https://github.com/mmrabe/hypr/actions/workflows/r.yml)

hypr is an R package for easy translation between experimental (null) hypotheses and contrast matrices as used for linear regression. For an extensive overview of the package functions, see the `hypr-intro` vignette, e.g. by running `vignette("hypr-intro")` after installing the package.

## Installation

Install from CRAN within R using:

```{r, eval=FALSE}
install.packages("hypr")
```

Install the development version in R using `devtools`:

```{r, eval=FALSE}
devtools::install_github("mmrabe/hypr", build_vignettes = TRUE)
```

## Deriving contrast matrices

For a treatment contrast with a baseline and three treatment conditions:

`r latexImg("H_{0_1}: \\mu_1 = 0")`

`r latexImg("H_{0_2}: \\mu_2 = \\mu_1")`

`r latexImg("H_{0_3}: \\mu_3 = \\mu_1")`

`r latexImg("H_{0_4}: \\mu_4 = \\mu_1")`

```{r}
trtC <- hypr(mu1~0, mu2~mu1, mu3~mu1, mu4~mu1)
trtC
```

To assign the contrast matrix to a factor `fac` with an intermediate hypr object:
```{r, eval=FALSE}
contrasts(fac) <- contr.hypothesis(trtC)
```

... or without an intermediate object:
```{r, eval=FALSE}
contrasts(fac) <- contr.hypothesis(mu1~0, mu2~mu1, mu3~mu1, mu4~mu1) 
```

For more information, see `vignette("hypr-regression")`.


## Deriving research hypotheses

To check which reserach (null) hypotheses a given contrast matrix is testing, we can create an empty hypr object.

```{r}
testC <- hypr() # create an empty hypr object
```

A treatment contrast with 4 levels (incl. the baseline) can look as follows:

```{r}
contr.treatment(4)
```

We can now populate the hypr object by setting its contrast matrix. Note that the treatment contrast does not have an intercept. We thus have to add it when populating the hypr object:

```{r}
cmat(testC, add_intercept = TRUE) <- contr.treatment(4) # populate object via contrast matrix
```

Now, the hypr object contains 4 hypotheses, a hypothesis matrix and the contrast matrix identical to the treatment contrast with an intercept added:

```{r}
testC
```

The derived hypotheses can be rewritten as:

`r latexImg("H_{0_1}: \\mu_1 = 0")`

`r latexImg("H_{0_2}: \\mu_2 = \\mu_1")`

`r latexImg("H_{0_3}: \\mu_3 = \\mu_1")`

`r latexImg("H_{0_4}: \\mu_4 = \\mu_1")`

For more information, see `vignette("hypr-contrasts")`.


## Community guidelines

If you want to report a bug, are having technical difficulties or want to recommend features, it’s best to open a [Github Issue](https://github.com/mmrabe/hypr/issues/new/choose). If you want to suggest a specific implementation of a feature or bug fix, you’re welcome to fork the repository and submit a pull request! Alternatively, if you are having problems or questions, you can also send an e-mail ().

Owner

JOSS Publication

hypr: An R package for hypothesis-driven contrast coding

Published

April 27, 2020

DOI

10.21105/joss.02134

Volume 5, Issue 48, Page 2134

Authors

Maximilian M. Rabe
University of Potsdam

Shravan Vasishth
University of Potsdam

Sven Hohenstein
University of Potsdam

Reinhold Kliegl
University of Potsdam

Daniel J. Schad
University of Potsdam, Tilburg University

Editor

Marcos Vital

Tags

psychology linguistics linear regression linear model statistics research methods research hypotheses

View PDF Review Thread Software Archive

Citation (CITATION.cff)

cff-version: 1.2.0
message: "If you use this software, please cite it as below."
authors:
  - family-names: Rabe
    given-names: Maximilian M.
    orcid: https://orcid.org/0000-0002-2556-5644
  - family-names: Vasishth
    given-names: Shravan
    orcid: https://orcid.org/0000-0003-2027-1994
  - family-names: Hohenstein
    given-names: Sven
    orcid: https://orcid.org/0000-0002-9708-1593
  - family-names: Kliegl
    given-names: Reinhold
    orcid: https://orcid.org/0000-0002-0180-8488
  - family-names: Schad
    given-names: Daniel J.
    orcid: https://orcid.org/0000-0003-2586-6823
title: "hypr: An R package for hypothesis-driven contrast coding"
version: 0.2.2
doi: 10.5281/zenodo.3765842
date-released: 2021-08-18

GitHub Events

Committers

Last synced: 7 months ago

Issues and Pull Requests

Last synced: 6 months ago