Alongside my research, I develop open source tools and resources that make methods from my work easier to use for others. Source code for all projects is available on GitHub.
mlstats
mlstats is an R package for multilevel descriptive statistics and data preparation. It computes within-group and between-group correlations, intraclass correlation coefficients (ICCs), and descriptive statistics for nested data, such as repeated measurements nested in persons, using either frequentist or Bayesian estimation.
Results are formatted according to APA standards and can be exported as publication-ready tables. The package also provides helpers to decompose variables into their within-group and between-group components for Random Effects Within-Between (REWB) models.
Open Source LLMs for Content Analysis
Together with colleagues, I created llm-content-analysis.com, a tutorial website on using ready-to-use, open source Large Language Models (LLMs) from Hugging Face for the standardized content analysis of texts. It is intended as a resource for researchers and practitioners in the social sciences and related fields, and requires no prior experience with LLMs.
The step-by-step tutorials walk through common tasks with three types of models: category-specific encoder models, task-specific encoder models, and universal decoder models. The site also serves as the companion website to our paper in Publizistik (in German), including a schematic guide and an empirical example study, and hosts the materials of a hands-on workshop.