What is Base rate?

Also called Base rate of success, Prior probability of success

The base rate is the proportion of applicants who would perform satisfactorily if they were hired with no assessment at all. It sets the headroom available to a selection procedure: where almost everyone would succeed anyway, or almost nobody would, even a highly valid test has very little left to add.

A good base rate weakens the case for testing, which is the opposite of what people expect

Taylor and Russell published the conversion in 1939. Given a validity coefficient, a base rate and a selection ratio, their tables return the proportion of the people you hire who will turn out satisfactory. The coefficient is only one of the three inputs, and it is the only one most vendors discuss.

Hold the test constant at a validity of 0.35 and hire one applicant in five. At a base rate of 50% the success rate among those hired is 69.7%, a gain of 19.7 points. At a base rate of 80% it is 91.9% — a gain of 11.9 points. At 95% it is 98.8%, a gain of 3.8 points. Move the other way, to a base rate of 20%, and the gain falls to 15.5 points.

The gain peaks near a base rate of one half and falls away on both sides. The same instrument, with the same published coefficient, is worth five times as much to one employer as to another purely because of a number that describes neither the test nor the vendor. A validity coefficient on a data sheet cannot tell you what the test will do for you.

Nobody knows their base rate, and the reason is structural rather than lazy

Measuring it properly would mean knowing how the people you rejected would have performed. You did not hire them, so you never find out. Every base rate in circulation is an estimate, and the honest ones say so.

The usual proxy — the share of current employees rated satisfactory — is biased upward, because those people already passed whatever screening was in place when they were hired. Feeding that inflated figure into the tables makes the test look less valuable than it is, because it puts you in the region where there is little headroom left.

The practical discipline is the same one that applies to a validity coefficient: model a range rather than a point. Run the arithmetic at 40%, 50% and 60% and see whether the decision changes. A business case that only works at one assumed base rate is not a business case.

It is a property of your applicant pool, not of the job

The base rate is set by who applies, at the wage you offer, in the labour market you recruit from. Raise the salary, change the job title, widen sourcing into a different channel, and it moves. The same role in two cities has two base rates.

That is why an industry benchmark base rate is close to useless, and why a base rate measured before a hiring freeze does not survive it. If you are going to assume a number, assume your own and write down the date you assumed it.

The same arithmetic outside hiring

Meehl and Rosen made the general case in 1955: a diagnostic sign or cutting score with respectable accuracy can still produce mostly false positives when the condition it detects is rare. The rarer the thing you are looking for, the more of what you find is noise.

Selection is the mirror image of the same result. A rare quality means a low base rate, which means a great deal of headroom and a test that earns its place; a common one means a high base rate and a test that mostly confirms what you would have got anyway.

One test, five base rates — validity 0.35, one hire in five

Base rate 20% → 35.5% of those hired succeed (gain 15.5 points)
Base rate 30% → 48.5% (gain 18.5 points)
Base rate 50% → 69.7% (gain 19.7 points — the maximum)
Base rate 80% → 91.9% (gain 11.9 points)
Base rate 95% → 98.8% (gain 3.8 points)

Taylor–Russell, computed from the bivariate normal rather than read off the printed 1939 tables, so intermediate values are available. Rounded to one decimal place.

Not the same as selection ratio

The base rate describes the pool that arrives; the selection ratio describes the decision you make about it. You cannot improve your base rate by being choosier — being choosier is the selection ratio. Confusing the two is how an organisation concludes that a new test worked when what actually changed is that more people applied.

Selection ratio

Sources

More terms beginning with B

Banding

Check a selection process against the four-fifths rule

Free, no signup, computed in your browser — with the remedy, not just the verdict.

Open the calculator

Last reviewed 2026-09-14