Take-Home Exams at Ridgeline University
Ridgeline University's faculty senate must vote on whether to replace timed, closed-book final exams with 72-hour take-home exams in all 1,400 undergraduate courses. A two-year pilot in 60 courses produced three findings. Mean course grades rose 0.35 GPA points relative to matched non-pilot sections. Reported honor-code violations fell from 41 to 12 per term, but an anonymous survey found that 38% of pilot students admitted to unauthorized collaboration. Alumni recruiters surveyed by the career office said they weigh internships and interviews far more heavily than transcripts. Faculty are split 54–46, with opponents arguing that the pilot confounds grade inflation with genuine learning and that the honor-code numbers reflect reduced detection rather than reduced cheating. The provost has asked for a decision grounded in the evidence actually available, not in the pilot's stated goals.
1. Take a position on whether Ridgeline should expand take-home exams to all courses. Justify it using only the evidence in the setup, and say which piece of evidence you weight most heavily and why.
Read the worked answer
Position: Expand, but only after the measurement problem is fixed - a conditional yes, not an endorsement of the pilot as it stands.
The evidence is weaker than it looks, and the weakness is the point. A 0.35-point grade rise is close to what you would predict from open-book conditions and from instructors grading their own students' polished work, so it is nearly uninformative about learning. The honor-code figure is worse: 12 reported violations against a 38% self-reported collaboration rate implies detection catches a small fraction of what students say they do. So the pilot does not establish that learning improved.
What it does establish is that the current format is not doing what faculty assume. Recruiters barely read transcripts, so the signal argument for closed-book exams is weak; and report counts are a poor proxy for integrity either way.
That points to a narrower conclusion than either side wants: expand where integrity can be independently measured - shared rubrics, external sampling of a portion of exams, and a validated integrity measure rather than self-report - and hold the rest. The most heavily weighted evidence is the gap between 12 and 38%, because it tells you the pilot's headline integrity result is an artifact of detection.
2. State the strongest objection to your position and explain whether it defeats your argument or merely qualifies it.
Read the worked answer
The objection: the conditional expansion is a bureaucratic dodge. Auditing 1,400 courses is expensive, faculty will resist external sampling, and in the meantime the university has already normalized a format in which a third of students admit to cheating. If take-home exams cannot be made honest at scale, the honest conclusion is to keep closed-book exams and improve them, not to phase in a system whose integrity is unmeasured.
Assessment: this is a serious objection and it partly bites, but it qualifies rather than defeats the position. It defeats the version of my argument that says expansion is safe now; it does not defeat the claim that the pilot's evidence cannot justify either expansion or the status quo.
Why it does not defeat the position: the objection assumes closed-book exams are the honest baseline, and the setup gives no integrity data for non-pilot courses at all. The comparison is between a measured format with known problems and an unmeasured one assumed to be clean.
Where the objection lands: it sets a burden. If a one-term audit shows external sampling is infeasible or that audited collaboration rates exceed roughly the closed-book baseline, the conditional yes collapses and the answer becomes 'retain closed-book, redesign the questions.' That is a real concession, not a rhetorical one.
3. What single additional piece of evidence would most plausibly change your mind, and why that one rather than others?
Read the worked answer
The evidence: an independent text-matching and proctoring audit of a random sample of pilot exams, compared against the same audit applied to a matched sample of closed-book exams, reporting unauthorized-collaboration rates for both.
Why this one: it is the only proposed study that replaces the two uninterpretable numbers in the setup with a direct comparison. The grade rise cannot be interpreted without knowing whether it reflects learning or inflation; the audit does not settle that either, but it settles the prior question of whether take-home exams are honest enough to be scaled at all. If the audit shows pilot collaboration at, say, 30% and closed-book at 5%, my conditional yes fails on its own terms and I should oppose expansion.
Why not other evidence: a survey of faculty preferences would only re-measure the 54–46 split. Employer opinion is already known and is nearly irrelevant, since recruiters ignore transcripts regardless of format. A study showing take-home exams predict later course performance better would be interesting, but it addresses learning, not the integrity constraint that makes expansion politically and institutionally impossible right now.
A second-best candidate would be a controlled comparison using a common externally graded question across formats - useful, but slower and it still leaves the cheating question open.