Selection Accuracy Calculator
If a hiring test is 80% accurate, how good are my hires?
Not 80%. On a pool where 30% would succeed, a test that passes 80% of good candidates and rejects 70% of the rest gives a hit rate of 53.33% — a lift of 23 points over hiring at random, which is real but far from the headline. It also rejects 20% of the good candidates, and you never find out which ones.
A test that is 80% accurate does not give you an 80% chance of a good hire. On an ordinary pool it gives about 53%.
The difference is the base rate, and it is almost never in the sales material.
Built by Muhammad Shahzeb Anwar, MSc Industrial and Organizational Psychology, who also writes the paid version of this tool. The base rate does more work than the test does, and it is the number nobody is given.
What the test is actually worth
Two accuracy figures, because passing good candidates and rejecting poor ones are different skills.
Why an accurate test still gives ordinary hires
Accuracy is quoted as a property of the test. The hit rate is a property of the test and the pool it is used on, and it is the hit rate you actually experience.
Work it through. Two hundred candidates, of whom 30% - sixty people - would succeed. The test passes 80% of them, so 48 good candidates get through. It rejects 70% of the other 140, so 42 poor candidates also get through. You hire from a pool of 90, of whom 48 succeed. That is 53.33%.
The test moved you from 30% to 53%, which is a genuine and valuable improvement. It is simply not 80%, and the gap between those two numbers is where disappointment with assessment products comes from.
The cost that never shows up
Twelve good candidates were rejected. They were capable, they would have succeeded, and the process filtered them out.
You will never learn this. Rejected candidates do not come back to report how well they did somewhere else, so the errors a selection process makes in that direction are structurally invisible while the errors in the other direction - the bad hire - are extremely visible.
That asymmetry pushes organisations toward ever stricter screening, because the visible error gets all the attention. On a low base rate, stricter screening throws away good candidates faster than it removes poor ones.
What raising the base rate does
Change the base rate in the calculator and watch the hit rate move far more than it does when you change the test.
That is the practical conclusion. Attracting a better pool - the job advert, the pay, the reputation, where you look - does more for the quality of your hires than a more accurate instrument applied to the same pool. The test is a multiplier on what you already have.
What this will not tell you
This is arithmetic on figures you supply. It does not know whether your test really has the accuracy claimed for it, and published validity figures are usually measured on a different population from yours. It cannot tell you whether a test is lawful or fair to use; the four-fifths ratio is a separate question and there is a separate tool for it.
Questions people actually ask
If a test is 80% accurate, why aren't 80% of my hires good?
Because accuracy is a property of the test and the hit rate depends on the pool as well. On a pool where 30% would succeed, a test that passes 80% of good candidates and rejects 70% of the rest produces a hit rate of about 53%.
What is a base rate in selection?
The proportion of your candidate pool who would succeed in the role anyway. It sets the ceiling on what any selection method can achieve, and it is usually missing when accuracy figures are quoted.
How many good candidates does a hiring test reject?
On the figures this page opens with, 20% of them - twelve people out of sixty. You never find out which, because rejected candidates do not report back on how well they did elsewhere.
Is it better to improve the test or the candidate pool?
Usually the pool. Changing the base rate moves the hit rate substantially more than improving the instrument does, which is why the advert, the pay and where you look matter more than the assessment.
Cite this tool
Free to use and free to quote. If a figure here is useful in something you are writing, a link back is all that is asked.
Muhammad Shahzeb Anwar, MSc Industrial and Organizational Psychology. Selection Accuracy Calculator (Base Rates). Shahzeb Speaks. https://www.shahzebspeaks.com/selection-accuracy-calculator-free-tool/
The other free tools
Each one answers a single question and refuses the ones it cannot. All free, none asks for an email address.
Want it on a better pool, side by side?
The paid Selection Accuracy and Base Rates tool runs the same method on two pools at once, so you can see what attracting better candidates is worth against what a more accurate test is worth - which is the comparison that decides where the budget should go.
See the full tool