A department receives its examination results. The cohort averaged sixty-four. Eleven students are below the pass mark, four are excellent, the rest are somewhere in the middle.
Every one of those facts is about ranking. None of them tells the faculty what to teach differently next term — which is the question they actually need answered, and the one a score has never been able to address.
The gap is not a failure of assessment design. A score is a summary, and summarising is precisely the operation that discards the information teaching depends on.
What a mark cannot distinguish
Two students score fifty-eight. One knew the material and could not organise it under time pressure. The other worked systematically through a structure and did not know enough to fill it.
They need opposite interventions. The mark cannot separate them, and neither can a class average built from a hundred such marks.
Multiply that across a cohort and the consequence becomes clear: faculty know how many students are struggling and not what they are struggling with. Remediation is therefore generic — more revision, more practice — because the data supports nothing more specific.
The information that would be useful
If clinical reasoning is a sequence rather than a single act, the informative question is where in the sequence students diverge.
At information gathering. Did they ask the questions that would have distinguished the possibilities, or did they take a history that was thorough and undirected?
At interpretation. Did they attach the right weight to what they found — treating an incidental finding as significant, or dismissing a significant one?
At differential formation. Did the plausible alternatives appear at all, or did they commit early to the first plausible answer?
At investigation choice. Did the tests they ordered discriminate between their own differentials, or were they a panel ordered by habit?
At safety. Did they recognise what needed escalating, and when?
Each of these is a distinct teachable skill with a distinct remedy. A cohort that is strong at gathering and weak at differential formation needs something entirely different from one with the reverse profile — and no examination mark distinguishes the two.
The argument for assessing the path rather than the endpoint is developed in how clinical reasoning is actually assessed.
What changes when the data is dimensional
Faculty holding per-dimension data across a cohort can do three things that a mark does not support.
Adjust teaching to the cohort's actual weakness. If a batch consistently under-performs at investigation choice, that is a curriculum signal rather than a set of individual problems, and it is addressable in a way that "average 64" is not.
Target remediation. A struggling student with a specific dimensional profile gets specific practice rather than being sent to revise broadly.
Detect the quiet failure. The student passing comfortably while consistently weak in one dimension is invisible to a mark and visible in a profile. Historically this student is found late — sometimes in internship — and the cost of finding them late is high.
The measurement traps
Dimensional data is more useful than a score and it introduces failure modes worth naming in advance.
Aggregation undoes it. Averaging nine dimensions into one number reproduces the original problem with more steps. If the composite is what gets reported, nothing has changed.
Comparison becomes tempting. Per-student dimensional data invites ranking students against each other, which is the same summarising instinct that caused the problem. The comparison worth making is a student against their own trajectory.
Volume creates noise. One encounter tells you very little about a dimension. A pattern across many encounters is what carries signal, and treating individual results as diagnostic produces false conclusions in both directions.
Measured things get taught. If a dimension is scored, it will be optimised for. That is acceptable when the dimensions genuinely describe good practice and corrosive when they are a proxy for it.
Where the data has to come from
The binding constraint on all of this is observation. Dimensional assessment requires watching someone reason, repeatedly, which is faculty time — the scarcest resource in every medical college, and the same bottleneck examined in competency-based medical education in practice.
This is where simulation has a specific and limited role: it produces observed encounters at a volume faculty cannot, and it produces them in a structured form that aggregates.
SYNTAX is Truffaire's clinical simulation platform — 500 cases across five specialties and 43 sub-specialties, delivered as voice-based encounters where information must be elicited, with structured review across nine dimensions. Our interest in this argument is disclosed.
What it produces for faculty is cohort-level dimensional data. What it does not do is observe a student with a real patient, and no aggregate should be read as though it did.
Frequently asked questions
Does this replace examinations?
No. Examinations certify; this informs teaching. They answer different questions and the failure is expecting one to do the other's job.
How much data is needed before a pattern is meaningful?
Enough encounters that individual case difficulty averages out. A handful is anecdote; a term's worth of encounters is a profile.
Should students see their own dimensional data?
Yes, with framing. A student who learns they are weak at differential formation has something to work on. One who receives nine numbers without interpretation has a new source of anxiety.
Does this create pressure to game the dimensions?
It can, and the protection is that the dimensions describe genuinely good practice rather than proxies for it. Anything scored will be optimised toward — the question is whether optimising toward it is desirable.
Can this identify faculty teaching quality?
It can indicate where a cohort is weak, which is not the same as attributing that to an individual teacher. Using it that way would degrade the data quickly, for the same reasons described in operational settings.
Where to start
Take your last cohort's borderline students and ask, for each, where in the reasoning sequence they actually failed.
If the honest answer is that nobody knows — that the mark records the outcome and not the cause — that gap is the case for dimensional assessment, and it is the same gap the remediation programme has been working around.
If you are evaluating simulation for a college or teaching hospital, get in touch.