Vol. 12 - Grading for progress and prioritization
Grades should encourage practices over outcomes.
I do not enjoy grading. Or grades.
I think they often stand in between students and learning, by introducing performance anxiety where there should be personal growth. They encourage conformity in seeking the right answer over the creative mistake.
Every time I have to assign a grade to a specific work (not to a student, to a work that this student turned in), I spend time thinking about what this grade should be measuring. But this year, I started thinking a lot more about what the process of grading should encourage.
I am starting a new class for biology undergraduates, in which we explore how various complex systems (agent-based models, cellular automata, individual-based models, etc.) can be used to represent biological systems, and how we can use our specific knowledge to ask more general, abstract questions. This is a challenging class to teach because there is a fair amount of coding involved, and this is an overwhelming skill to acquire for a lot of biology majors.
To encourage experimentation, I decided (for the coding component of the class) to assess practices over outcomes. Whether the code gives the "right" result is not something I am particularly interested in (explaining why the result feels off is a more interesting use of student's biology knowledge anyway). My rubric uses items like "Code style", and "Code organization", and "Reproducibility" instead. This is something we discuss early in the class: failing to get the right result is a natural part of the scientific process; but sloppy work is not excusable.
For each of these assessment items, I use a directed scale where each rank starts with "and". For code style, for example, it goes like this: one point if the variable names are explicit, two points if there are comments in the code, three if the comments focus on "why". This helps remove the question of "what skill do I need to pick up first?" because to get three points, students first need to use explicit variable names and document their code properly.
Helping students prioritize also happens at the macro scale, where we use different items from the rubric as we move into more advanced projects. For the very first project, evaluation uses the items focused on biological discussion and interpretation; the second project adds minimal good practices in documentation (comments, docstrings); the third project uses both of these families of items; the final project uses the entire rubric, introducing new elements.
Looking at the current distribution of grades (and the state of the ongoing final projects), students have been doing really well grade-wise, in a way that tracks my own observation of their increasing familiarity with different tools and concepts. A big part of designing this rubric was to make sure that there is as little room for ambiguity as possible. Evaluating whether the work of my colleagues is "impactful" or "highly impactful" is measuring my definition of "high impact" much more than it is evaluating the quality of the work; using ambiguous assessment criteria for student's work is preventing them from assessing their own learning.
And with all that said, stay tuned for Vol. 13 next week, for a meditation of three months of newsletter, and some advice to maintain a good writing schedule for research papers.