The study was presented at ICAME43 in Cambridge, UK. The **presentation slides** can be found [here](osf.io/ambt7).
The manuscript is available as a [**preprint**](psyarxiv.com/jr8yk/) on *PsyArXiv*:
- Sönning, Lukas & Jason Grafmiller (2022). Seeing the wood for the trees: Predictive margins for random forests. *PsyArXiv preprint*.
**Data** used in the study have been published on TROLLing:
- Grafmiller, Jason. 2022. The genitive alternation in 1960s and 1990s American English: Data from the Brown and Frown corpora, https://doi.org/10.18710/R7HM8J, DataverseNO, V1.
An **R package** for implementing predictive margins is available from Github [here](github.com/jasongraf1/predictiveMargins). There is also a **vignette** with basic information on how to use the package.
The folder **R scripts** contains two files:
- **script_predictive_margins**: R code for reproducing the analyses and figures in the paper, as an html file and a Quarto/RMarkdown script.
- **tutorial_predictive_margins**: A short **tutorial** showing how to manually construct predictive margins from a random forest model, as an html file and a Quarto/RMarkdown script.
**Images** created for this study can be found in the folder "figures". They are published under a Creative Commons Attribution 4.0 licence (**CC BY 4.0**), which means that the licence terms for their use are quite generous.
