What is Selection ratio?
Also called Selection rate, Hiring ratio
The selection ratio is the number of people hired divided by the number assessed. It is the lever that decides how much a valid test is worth: the fewer people you take from a given pool, the further up the score distribution your cut falls, and the more predictive accuracy converts into better hires.
Validity is what the instrument can do; the selection ratio decides how much of it you keep
Hold a test at a validity of 0.35 and a base rate of 50%. Hire one applicant in twenty and 77.7% of those hired succeed — a gain of 27.7 points. Hire one in five and it is 69.7%, a gain of 19.7. Hire four in five and it is 54.9%, a gain of 4.9. Hire nineteen in twenty and it is 51.5%, a gain of 1.5 points.
The same instrument delivers roughly eighteen times the lift at one hire in twenty as at nineteen in twenty. The most common way to waste a valid test is to administer it to everyone and then hire nearly all of them, which is what happens whenever assessment is introduced as a formality on top of a pipeline that was already going to fill.
It is the number that connects the recruiting budget to the assessment budget
Money spent on sourcing enlarges the applicant pool, which lowers the selection ratio, which raises what a given validity coefficient is worth. Those are two lines in two different budgets and they are almost never modelled together, even though one of them sets the return on the other.
The same mechanism runs in reverse and is less visible. Dropping a cut score to fill headcount raises the selection ratio. The test carries on producing identical scores and identical reports while its contribution to the decision quietly collapses. Nothing on any dashboard flags it.
The limit almost everyone forgets: a low selection ratio is only worth having over a good pool
The Taylor–Russell arithmetic holds the base rate fixed while the selection ratio varies. Real sourcing does not. A campaign that multiplies applications without improving their quality lowers the selection ratio and lowers the base rate at the same time, and the two effects work against each other.
One hire in fifty from a weak pool is not better than one in five from a strong one. Before treating a falling selection ratio as good news, ask whether the extra applicants resemble the ones you already had.
The realistic range, and what to do outside it
Schmidt and Hunter take 0.30 to 0.70 as the practical range for most settings. Below it you are in high-volume screening, where the arithmetic is generous and the binding constraints become candidate experience, throughput and adverse impact rather than validity. Above it you are effectively not selecting, and the honest conclusion is that the assessment is being used for development, placement or onboarding rather than for a decision.
That is not an argument against assessing at a high selection ratio. It is an argument for saying out loud which job the assessment is doing, because a tool bought on a selection business case and used at a ratio of 0.9 will not produce the results the business case promised.
One test, seven selection ratios — validity 0.35, base rate 50%
- 1 hire per 20 assessed → 77.7% of hires succeed (gain 27.7 points)
- 1 in 10 → 74.1% (gain 24.1)
- 1 in 5 → 69.7% (gain 19.7)
- 2 in 5 → 63.8% (gain 13.8)
- 3 in 5 → 59.2% (gain 9.2)
- 4 in 5 → 54.9% (gain 4.9)
- 19 in 20 → 51.5% (gain 1.5)
Taylor–Russell, computed from the bivariate normal. The validity coefficient is identical on every line; only the proportion hired changes.
Not the same as yield ratio
A yield ratio measures survival across one stage of a funnel — applications to screens, screens to interviews. The selection ratio is the whole funnel end to end: people hired divided by people considered. A pipeline reported as six healthy stage conversions never states the one number the utility arithmetic needs, which is their product. Six respectable-looking percentages can multiply to a selection ratio nobody in the room has seen.
How AssessAll handles it
AssessAll's hiring insights compute six stage conversions — invited to started, started to completed, completed to scored, scored to shortlist, shortlist to offer, offer to hire — against everyone invited to assess (lib/hiring/insights.ts). Two honest caveats follow from that. The denominator is candidates invited, not everyone who applied, so the true selection ratio for the requisition is lower than anything the dashboard shows. And the six ratios are displayed separately and never multiplied, so the end-to-end figure the utility models need is not on the screen. Multiply them yourself before quoting a selection ratio to a finance team.
Sources
- Taylor, H. C. & Russell, J. T. (1939). The relationship of validity coefficients to the practical effectiveness of tests in selection: discussion and tables. Journal of Applied Psychology, 23(5), 565–578
- Schmidt, F. L. & Hunter, J. E. (1998). The validity and utility of selection methods in personnel psychology. Psychological Bulletin, 124(2), 262–274 — the 0.30–0.70 practical range, p. 263
Read next
Related terms
More terms beginning with S
Check a selection process against the four-fifths rule
Free, no signup, computed in your browser — with the remedy, not just the verdict.
Last reviewed