SFBU designs assessment around observable, demonstrable skills. Each is something a student can do in a real or realistic setting that a knowledgeable observer can assess.
“Can analyze a financial statement to identify early indicators of liquidity risk” describes a skill someone can observe. “Understands financial analysis” describes nothing an observer can see. Employers look for skills of the first kind, and graduate programs test them in interviews. Rubrics built on them describe proficient and developing work, and students can use them to plan their next attempt.
A grade compresses a detailed record of a student’s learning into one symbol. Students attend to grades for good reasons, since scholarships, visas, internships and admissions depend on them. When the grade becomes the main thing students aim for, it damages learning. SFBU designs courses where the path to a strong grade runs through the behaviors that produce learning. That is settled early in the sequence described in Course Design.
Mid-challenge, checking an answer against the clues on the board.
A meta-analysis of 435 studies found that feedback has a medium-sized effect on learning (Wisniewski, Zierer and Hattie, 2020). Feedback that carried more information had larger effects. Feedback that arrives only on final assessments comes too late to use, so SFBU courses build low-stakes practice before high-stakes assessment. A software engineering course that collects a design rationale two weeks before implementation gives students feedback while they can still use it. A comment that names where the work succeeds and where it is still forming tells a student what to do next.
The organizing question for any assessment at SFBU is what completing it asks of the student that AI does not supply. Where little is required, the assessment is redesigned around the point where human judgment enters. In practice students document their process, defend their choices, revise after specific critique and apply concepts to contexts they know first-hand. In a graduate strategic management course, students write a competitive analysis of a company. Each student then records a ten-minute video that explains three choices they made and names one thing the AI draft got wrong. Teaching with AI sets out the capacities this evidence is meant to show.
The most durable basis for fair assessment is knowing students well. An instructor who has talked with each student about their work across a semester asks better questions and reads growth in context. Relational Intelligence describes how that knowledge is built.
Take this with you
1
Where a row says “analysis, 20 points,” describe what the analysis looks like at three levels. Students can act on a description.
Prompt 05 in Resources drafts a levels-based rubric from your assignment brief.
2
A design rationale before implementation, an outline before the essay. Name where the work succeeds and where it is still forming.
3
A ten-minute video or conversation: which three choices did you make, and what did the AI draft you started from get wrong? The guidebook’s Appendix A.3 gives a complete rubric for this assignment.
One question to keep
Where in my course does feedback arrive while a student can still use it?
Download the Faculty Guidebook
The full argument, worked examples and sources. PDF.
