Quality Education / Pressure And The Pointer

The completion figure that improves by changing who is admitted

The clearest illustration of the problem involves no rule breaking at all.

Take an institution judged on the proportion of its students who finish. There are two broad ways to improve that figure. One is to teach better, support students more effectively, fix the parts of the course where people get stuck, and make the sequence itself more survivable. That work is slow, expensive and uncertain. The other is to enrol students who were more likely to finish in the first place, by tightening entry, discouraging applicants with complicated circumstances, or quietly shifting the marketing toward the confident and already well prepared.

The second path is faster, cheaper and produces a better figure sooner. It also makes the institution worse at the thing society might most want it to do, which is to take people who are not already advantaged and get them somewhere. Nothing improper has occurred. Admissions policy is legitimately the institution's own business, and everyone involved can explain their reasoning without embarrassment. The measure has simply rewarded selection over teaching, because it could not tell the difference, and a measure that cannot tell the difference will eventually be satisfied by whichever of the two is cheaper to supply.

The example generalises to nearly every measure in use. Any figure that can be improved by changing the input rather than the process will eventually be improved that way somewhere, because that route is always available and usually cheaper. This is the test worth running on any proposed measure before adopting it: ask what the laziest available way to improve it would be, and then ask whether you would be content with an institution that took it. If the answer is no, the measure needs a companion that moves the other way when the shortcut is taken, so that the shortcut leaves a mark somewhere visible.