Skills and psychometric assessments

What assessments does OpusHire provide?

OpusHire serves coding assessments, English proficiency assessments and psychometric assessments covering verbal, numerical, logical, situational-judgement, personality and emotional-intelligence dimensions. All of them are drawn from maintained item banks and scored on the server, so results are comparable between candidates rather than being a raw count of correct answers.

Each family answers a different question. Coding assessments test whether someone can do the work. English proficiency matters where the role is client-facing or the working language is not the candidate's first. Psychometric measures speak to reasoning and working style, and are the ones most often misused — which is why they are presented as one input among several rather than a verdict.

Why does server-side scoring matter?

Scoring on the server means the answer key never reaches the browser and the score cannot be tampered with by the candidate. It also means every candidate is scored by the same code against the same key, so a score from March and a score from October mean the same thing and can be compared directly.

Client-side scoring is common in assessment tooling and quietly fatal: if the browser knows the right answers, the result is unverifiable. Beyond the integrity argument, server-authoritative scoring is what makes standardisation possible at all — scores are converted into standard scores and percentile bands against the item bank, rather than reported as a raw mark whose difficulty nobody can reconstruct later.

• Standard scores and percentiles, so "70%" on an easy set and a hard set are not silently treated as equal.
• Maintained item banks with retirement flags, so overexposed items can be pulled from circulation.
• Reliability tracking on the banks, so a section that stops measuring consistently is visible rather than assumed sound.

How do you stop assessment questions from leaking?

Items are drawn from maintained banks and carry retirement flags, so an item that has been over-used or has appeared publicly can be pulled from circulation without rebuilding the assessment. Combined with server-side scoring — the answer key never reaches the browser — this addresses the two ways assessment integrity usually fails.

Leakage is inevitable over a long enough period; the question is whether a platform can respond to it. A fixed question set with client-side scoring cannot: once the items circulate, every score it has produced becomes suspect and there is no remedy short of replacing the whole assessment.

Bank health is also tracked rather than assumed. Reliability statistics on each bank make it visible when a section has stopped measuring consistently, which is the signal that items need review — not something to discover after a year of decisions have been made on the results.

Can assessment results screen candidates out automatically?

No. A requisition can define an advisory minimum, and candidates below it are flagged for human review — never rejected automatically. The flag exists to direct a reviewer’s attention, and a reviewer can act on it or disregard it with the full result in front of them.

Assessments are also the place where accessibility and fairness obligations bite hardest, and where an automated cut-off is most likely to produce a disparate outcome nobody intended. Keeping a human in the loop is both the defensible position and the one that produces better hiring, because a borderline score on one dimension is frequently irrelevant next to what the rest of the evidence shows.

Frequently asked questions