Teaching with the TIER protocol and replication assignments using R

Tags: #<Tag:0x00007f79a8b83958> #<Tag:0x00007f79a8b83818> #<Tag:0x00007f79a8b836d8> #<Tag:0x00007f79a8b83598> #<Tag:0x00007f79a8b83458>

I’m attending a project TIER conference at Haverford College, which has the goal of making research transparency becomes standard and ubiquitous in the education of social scientists. Their community has published a bunch of excellent papers on this (e.g. 1, 2). They have a TIER protocol which requires students to submit a complete description of the replication documentation as their assignment. For R-using students, this might mean that they minimally submit a Rmd file, a data file, and the PDF output for an assignment.

This seems well-aligned with rOpenSci’s community values. Some former Project TIER fellows are also active in the rOpenSci community, I see some familiar names, like Amelia McNamara, Mine Çetinkaya-Rundel and Ben Baumer, on their list of people.

They have a collection of syllabi using R and R Markdown. I see we now have rOpenSciEd, and some great blog posts (1, 2, 3), and so I’m curious to know do rOpenSci community members teach in way that are generally consistent with the TIER protocol (i.e. require students to submit reproducible assignments)? Do you have students do assignments that are rep-x (reproductions/replications of published papers) using R and open data, etc. I’ve seen @cboettig’s teaching materials on GitHub that require students to submit reproducible assignments. Rep-x assignments could be a fun topic of future unconf!

I’d love to know how others are doing this kind of stuff in their teaching in domain-specific subjects (rather than stats and compsci classes)

I wonder if it might be useful to identify the rOpenSciEd materials on GitHub with a tag for #replication-assignment to make it easy for instructors to find examples of this important type of assignment?

2 Likes

@lincoln :point_up_2:t3:
I’ve tagged the other rOpenSciEd authors via twitter

2 Likes

I do this in all my classes (undergrad/graduate Bioinformatics, Ecology, Microbiology, California Ecology) at the University of San Francisco.

In the California Ecology class, which is a core class for non-STEM majors, assignment submissions are zip files of project folders (Rproj, Rmd, docx, data, figures, etc) that get uploaded to our LMS (Canvas). For majors classes, the submission process is via Pull Requests on GitHub where all commits must pass Travis checks for code execution (knit without errors; built based on ideas from discussions with @cboettig about how he had things set up for his courses at Cal) and style (lintr) .

The Bioinformatics classes use existing datasets from NCBI’s SRA (Sequence Read Archive) and the Ecology class uses data from NEON (the National Ecological Observatory Network), and all steps including downloading and processing the raw data are scripted.

2 Likes

Several folks have classes that cover replication of published results (which has a bit of a different scope than individual assignments be reproducible):

It might also be useful to look into similar efforts in the jupyter notebook world:
https://nbgrader.readthedocs.io/en/stable/

2 Likes