LOCATION: IDRE Portal https://idre.ucla.edu/idre-portal
This workshop will present several workflows and tools for conducting and documenting quantitative research to ensure transparency and reproducibility of the data processing and analysis that underlie reported statistical results. There will be a particular emphasis on strategies instructors can use to integrate these methods into their classes and advising, so that students are immersed in principles and practices of transparency early in their research training.
The workshop is intended primarily for faculty, staff and graduate students who teach, serve as TAs, or otherwise support courses on quantitative methods, as well as those who supervise theses, dissertations or other forms of independent student research. Participants from both undergraduate and graduate programs are welcome. The focus will be on empirical research in the social sciences, but much of the material covered will also be applicable in relevant fields of the natural sciences, and participants from all disciplines are welcome.
The workshop will introduce two approaches to reproducible quantitative research:
- The TIER Protocol: The TIER Protocol consists of specifications for comprehensive replication documentation for an empirical study, and guidelines for a workflow in which the construction of that documentation is integrated throughout the entire research process. Constructing editable command files containing code that executes all the steps of data processing and analysis required to generate the results reported for a study is central to this workflow. This workflow can be adopted by users of any programmable statistical software (e.g., Stata, R, SPSS, SAS, Matlab, and many others).
- Markup files with embedded code: In this approach, the text of the paper and code that generates the statistical results are integrated in a single document. The text is written in a markup language, and chunks of statistical code are embedded throughout the document. This source file is sent to an engine that renders the complete paper, in which the tables and figures produced by the code chunks are displayed in place of (or along with) the code that appears in the markup file. The final paper can be rendered in a variety of formats, including HTML, pdf, Word and Latex. The workshop will demonstrate how this workflow can be implemented using either R (with R Markdown) or Stata (with an appropriate add-on).
The workshop will also discuss how on-line file-sharing platforms can be used to facilitate both of these approaches to reproducible research. The focus will be on the Open Science Framework (OSF), but the discussion will apply to any file-sharing platform (e.g., Drobox, Google Drive, GitHub, or an institutional server).
In addition, time will be devoted to a broader discussion of underlying purposes and principles: Why should research be reproducible? And what are the educational benefits of teaching students to use reproducible methods when they conduct quantitative research?
Project TIER is supported by the Alfred P. Sloan Foundation