A university department where the large majority of students in every course receive an A or A- is often discussed purely as a question of academic rigor — are standards too lenient. That framing misses a separate, more precise problem: regardless of whether the underlying standard is appropriately rigorous or not, compressing the distribution of grades toward the top of the scale genuinely reduces how much information a grade conveys about a student's actual relative performance, which is a measurement problem independent of the rigor question.
Why compression is a measurement issue, not just a leniency issue
A grading scale's usefulness as a measurement instrument depends partly on how well it discriminates between genuinely different levels of underlying performance — a scale where 90% of students land in the top one or two categories has much less room to distinguish an exceptional student from an adequate one than a scale where performance is spread more evenly across the available range. This is true even if the underlying standard for what counts as "A-level work" hasn't changed at all — the problem isn't necessarily that the bar moved, it's that more students are clustering near whatever the top of the scale represents, leaving less differentiating information in the grade itself.
What this actually costs, concretely
A transcript where nearly every grade is an A conveys meaningfully less information to anyone trying to use it to distinguish between candidates than a transcript with genuine variation would — an employer or graduate admissions committee comparing two candidates with nearly identical, uniformly high GPAs has effectively lost the resolution that grades used to provide for exactly this comparison. This is a real, measurable cost that falls specifically on the people relying on grades as a signal of relative performance, not on the students receiving the grades, who individually benefit from the more generous outcome.
Why this happens gradually and is hard to reverse unilaterally
Grade compression tends to worsen gradually and somewhat independently across institutions and even across departments within the same institution, driven by a mix of pressures — course evaluation incentives tied partly to grades, competitive pressure not to disadvantage one's own students relative to peer institutions or departments with looser grading, and the discomfort of being an outlier that grades unusually strictly in a context where peers don't. This creates a collective-action problem: a single department that unilaterally tightens its own grading distribution puts its own students at a comparative disadvantage relative to peers elsewhere, without actually solving the broader measurement problem for anyone.
What actually addresses the underlying measurement issue
Some institutions have experimented with reporting a course's grade distribution directly on the transcript alongside the individual grade, which restores some lost information without requiring any single department to grade more strictly in isolation — a reader can see not just the grade received but where it sat within that specific course's actual distribution. Norm-referenced or curved grading approaches directly address distributional compression by construction, though they carry their own trade-offs and aren't appropriate for every kind of assessment. Standardized external assessments, where available, provide an additional signal less vulnerable to institution-specific grade compression.
What this means for anyone relying on grades as a signal
- Recognize grade inflation as a measurement precision problem, separate from and in addition to any question about academic rigor
- Where available, look at a grade's position within its course's actual distribution, not just the letter grade in isolation
- Weight standardized external assessments more heavily when comparing candidates from grading contexts known to be heavily compressed
- Understand that unilateral grading reform by one department or institution addresses fairness locally but doesn't resolve the broader signal-loss problem across the wider system
None of this is an argument that students deserve lower grades — it's a recognition that a compressed grading scale, whatever its cause, measures relative performance with less precision, and that cost lands on whoever is trying to use the grade as a signal after the fact.