Selection Ratio

What selection ratio is

Selection ratio (نسبة الاختيار) is the number of people hired divided by the number of people who entered the selection stage. The smaller the result, the wider the field the organisation had to choose from. The closer it comes to one, the narrower that field becomes, until hiring turns into acceptance rather than choice.

It is a ratio, not a measure of quality. Nothing in it says that the people hired were the better candidates, or that the people turned away were unsuitable. What it says is how much choice the organisation had at the moment it decided. Its value lies in being the condition inside which every selection tool works: an excellent tool with no choice in front of it does nothing.

Why the denominator decides the selection ratio

The numerator is not in dispute: it is the number of people hired. The denominator, though, can take four different values in a single hiring process, and each of them produces a figure that someone will call the selection ratio. Take a request to fill one role, to which 240 people applied:

• 90 were excluded for failing an explicit requirement, such as a professional licence the role demands, which left 150.
• 60 of those passed the initial screening interview.
• 24 reached the assessment stage.
• 3 were hired.

Measured against all applicants, the ratio is 3 divided by 240, which is 0.0125, or 1.25%. Measured against those who remained after the automatic exclusion, it is 3 divided by 150, which is 0.02, or 2%. Against those who passed the initial screening, it is 3 divided by 60, which is 0.05, or 5%. Against those who reached assessment, it is 3 divided by 24, which is 0.125, or 12.5%.

The last of these figures is ten times the first, and nothing in the organisation changed between them. A selection ratio reported without its denominator therefore cannot be compared with anything: not with an earlier period, not with another role, and not with another organisation.

The denominator that serves the practical question is the number of people who actually entered selection, meaning those who were measured with a selection tool. Someone excluded on an explicit requirement never entered a selection at all. Including them in the denominator shrinks the ratio, and the shrinkage says more about the job advertisement than about the selection.

Why a low selection ratio does not mean a high standard

A selection ratio results from two sides: how many vacancies the organisation opened, and how many people came forward. The second side is only partly in the organisation’s hands. It moves with the state of the market, with how clear the advertisement was, and with what people know about the organisation as an employer.

A low ratio can therefore point in two opposite directions. It may mean that a large pool of qualified people applied and the choice widened. It may equally mean that a broadly worded advertisement flooded the process with people who did not fit the description, so that the denominator grew with applicants who were never in the competition. The two situations produce the same figure, but they do not call for the same decision.

A second indicator tells them apart: the share of candidates who passed the initial screening. In the example above, 60 of the 150 who remained after the explicit requirement passed it, which is 40%, a share that suggests an advertisement that reached people close to the description. Had 12 of the 150 passed, which is 8%, the problem would have lain in sourcing rather than in the standard. The remedy would then lie in how candidates are found and attracted, not in how they are chosen.

How selection ratio affects what a selection test appears to show

Predictive evidence for a test, one of the routes described under test validity, compares scores with what later showed up in the work. An organisation, however, sees performance in the job only for the people it accepted, and that narrows the range of scores over which the relationship is measured. Selection ratio is the quantity by which that narrowing is measured. The smaller it is, the more alike the people hired will be on the trait that was measured, and the weaker any relationship between score and outcome will appear compared with the relationship across all applicants.

A practical caution follows. An organisation that picks one candidate in forty and then checks its tool on the people it hired can find a weak relationship, and it would be wrong to conclude from that alone that the tool does not work. The weakness here is an effect of how selective the hiring was, not a verdict on the tool. It grows as the organisation becomes more selective, which is to say as it does better at choosing.

Where a selection tool has room to work, given the selection ratio

The usefulness of a selection tool is not measured by its validity alone. It depends on its validity together with the selection ratio and with the share of candidates who would have succeeded without any tool. That third quantity can be left out of the calculation entirely, and the example below shows how much it can change the result.

Go back to the 24 candidates who reached assessment. Suppose that, had all of them been hired, half would have succeeded, which is 12. If three are hired at random, the expected number of successes is 3 multiplied by 0.5, which is 1.5. If the tool were perfectly accurate, all 3 would succeed. The distance the tool is able to cover is 1.5 out of 3: at best it could raise success from half of those hired to all of them.

Now change a single assumption. If 90% of those candidates would have succeeded without any tool, a random choice of three would be expected to produce 3 multiplied by 0.9, which is 2.7, against a perfect 3. The whole distance is then only 0.3. The selection ratio is the same in both cases and the tool is the same in both cases, but the room in front of the tool has shrunk to a fifth.

The figures in both examples are assumed. They are there to show the effect of a change in the underlying success rate, and they are neither a measurement of any tool nor a reference on which to base a purchase. The definition of selection ratio does not settle how much a particular tool adds.

Two conclusions follow. Spending money on a tool for a role in which nearly everyone who reaches assessment succeeds is spending on a short distance. The same tool can have a larger effect where the people reaching assessment differ more from one another. The share who would have succeeded without a tool may not be known, and it is hard to establish within a single organisation. We found no published source on which to base that share for the Saudi market. Acknowledging that it is unknown is still more useful than ignoring it, because ignoring it produces an expectation that no tool can meet.

When the selection ratio reaches one

If everyone who entered selection is hired, the selection ratio becomes one, and whatever is spent on tools has no effect on the decision, however accurate they are: there is no decision whose quality could improve. A second use remains for the tool in that case, and it should not be dismissed: learning what induction and training each new hire will need, which is the subject of employee onboarding.

This situation can arise where the occupation is scarce, where the pay sits below the market, and where the hire is required within a period that leaves no room to wait. The response is not to raise the standard. A stricter standard with no candidates produces an open vacancy, not a better hire, and the question then becomes one of how long the vacancy stays open and why.

How batch hiring changes the selection ratio

The calculation so far concerned a role in which three people were hired. In batch hiring, such as opening one hundred identical vacancies for a season or a new branch, the selection ratio has to be interpreted differently. The capacity side of that kind of hiring is covered under high volume hiring.

Take one hundred vacancies to which 400 people applied. The ratio is 100 divided by 400, which is 0.25. That is twenty times the ratio of the first example measured against all applicants, yet the organisation still has a real choice here: there are three hundred applicants it will turn down.

The difference is that the room for choice in a single hire runs out quickly, since one withdrawal among three finalists removes a third of the choice. In a batch, the room remains even after ten or twenty withdrawals. That is why a batch figure is examined alongside the number who withdrew after receiving an offer. If offers go to one hundred people and 30% of them withdraw, 70 start work and 30 vacancies remain. Those thirty are then filled from the three hundred who were turned down, so the second round has a selection ratio of 30 divided by 300, which is 0.1. The organisation ends up more selective than it planned, choosing from candidates it had rejected in the first round, and that is the point at which the reason for each first rejection is worth recording before it is needed.

How selection ratio differs from quality of hire, cost per hire and the ratio of interviews to hires

• Quality of hire. That is a judgement on the outcome of hires after they have happened, while selection ratio describes the conditions in which the decision was taken. A low selection ratio does not promise high quality; it promises only that high quality was possible.
• Cost per hire. A low selection ratio can raise the cost, because every candidate assessed takes time. The two figures can move in opposite directions, and examining them together gives a truer picture than examining either one alone.
• The ratio of interviews to hires. It measures the effort spent within one stage, while selection ratio measures what lies between the two ends of the process. The first can rise while the second stays where it was, if more interviews are held with the same number of candidates.

What distorts a selection ratio

• A denominator that changes between two periods. An organisation that added an automatic exclusion halfway through the year changed its denominator, so a change in the ratio reflects the change in procedure rather than any change in the number of people applying.
• Reposting the advertisement. Applicants to a role that is advertised again may be counted twice, depending on the system, which inflates the denominator with no new candidates. The place to check this is the applicant tracking system itself.
• Combining unrelated roles in one figure. A selection ratio for the whole organisation combines a role with three hundred applicants and a role with four, and produces an average that describes neither.
• Counting withdrawals as rejections. A candidate who entered assessment and then withdrew was not turned down. Counting them in the denominator lowers the ratio and hides the fact that the organisation lost a candidate it wanted.

Before a selection ratio is compared with anything

A selection ratio describes a condition, not a result. That is why it is reported together with its denominator, the role and the period; without them it is not really a figure at all. There is also no correct selection ratio to prefer over another, because the figure describes the market for the occupation as much as it describes the organisation.

The figure is most useful for answering two questions. Is there any real choice in this role at all? And is the room in front of our tool worth what we spend on it? If the answer to the first is no, the work lies in sourcing and in the terms of the offer, not in stricter screening. If the room in the second is narrow, a simpler tool is enough, and the money saved can move to onboarding, because what cannot be gained through selection may still be gained through what comes after it.

How candidates’ data is processed, and what an organisation may ask of candidates, have their own sources; the definition of selection ratio settles neither question.

This is an explanation of the concept, not legal advice.

Related terms

• Cost per Hire
• Test Validity
• High Volume Hiring
• Structured Interview
• Candidate Experience