Hiring, and the filters ยท 4.4
References, and what they can legally say
References, and what they can legally say. What is actually the case, and how it compares with what is repeated.
A practical software reference for this part of the discussion is Monitask's overview of productivity tracking software.
References are collected almost universally, predict very little, and in one large economy have been narrowed by fear of litigation to a confirmation of dates.
What they can say
In the United States there is generally no legal prohibition on giving a substantive reference, and many employers restrict themselves to confirming job title and dates anyway. That is a policy adopted to avoid defamation claims rather than a requirement, and it has become an industry norm on the strength of a risk that is small.
Several states provide qualified immunity for employers giving truthful references in good faith, and the norm has not changed much in response.
Elsewhere the position differs. Some jurisdictions impose a duty of care to make a reference accurate and not misleading, which cuts both ways: an employer may not damage a former employee unfairly and may not conceal something material either.
What they predict
Little. Predictive validity of references for job performance is low in the meta-analytic literature, and the reasons are structural: the candidate chooses the referee, the referee knows the candidate chose them, and almost nobody supplies a negative reference in writing.
A method in which the subject selects the witnesses is not a method for finding out anything unwelcome.
The letters, where they still matter
In academic and clinical hiring, substantive letters remain standard. Research analysing them has found systematic differences by gender: letters written for women tend to be shorter, more likely to use words describing effort and diligence, and less likely to use words describing ability and achievement.
The pattern has been found repeatedly and the writers are typically unaware of it. It is a clear example of a mechanism operating inside people who would sincerely deny holding the preference.
The back channel
The informal call to somebody the candidate did not nominate. It produces more candid information and it is unstructured, unrecorded, unverifiable, and available only to candidates whose industry the hirer already knows.
Which makes it simultaneously the most informative reference practice and the one most likely to entrench existing networks.
What to use them for
Verification, which they do reliably. Dates, title, and whether the person is eligible for rehire, where that question is answered.
Treating them as evidence about performance is treating a formality as a measurement, and it is the reason references consume time out of proportion to what they contribute.
If you want them to be useful
Ask specific behavioural questions rather than general ones. Ask about a defined situation. Ask what the person would need support with, which is answerable without being negative. And ask the same questions of every referee for every candidate, for the reasons given in the entry on interviews.
A structured reference is a marginal improvement on an unstructured one and remains weaker than a work sample.
The timing question
Taking references before an offer risks alerting a current employer; taking them after weakens their function, since withdrawing an offer is costly and rare.
Most employers resolve this by making offers conditional and then almost never acting on what comes back, which is an honest description of what the step has become.
For the person being referenced
Ask directly what a referee will say. Most will tell you, and the question is not awkward. Where a former employer has a stated dates-and-title policy, say so in advance rather than letting the hirer discover a thin response and draw a conclusion.
The reference nobody asks for
A former direct report. Almost never requested, frequently the most informative source about anybody who has managed people, and available for the asking with the candidate's permission.
Where a bad reference is actionable
In several jurisdictions a reference that is inaccurate and damaging can give rise to a claim, and in some the duty runs to the recipient as well: an employer who conceals something material may be liable to the new employer.
The practical consequence is the dates-and-title norm, which protects against both duties by conveying nothing.
Two questions worth the call
What did this person find difficult, and what would you want their next manager to know. Both are answerable honestly without being negative, and both produce more than a general request for comment.
Keeping your own record
Ask for a written reference when you leave, while the people who worked with you are still there. Managers move, firms restructure, and a letter obtained on the way out costs nothing and cannot be obtained later.
What this rests on
- Reference immunity statutes exist in many American states and are published; the dates-and-title norm is a policy choice rather than a legal requirement.
- Meta-analytic estimates of reference validity are low and are reported in the personnel selection literature.
- Research on gendered language in academic recommendation letters has been published repeatedly since the early two-thousands with consistent findings.
- Duties of care in giving references differ by jurisdiction; nothing here is legal advice.
For broader context, consult EEOC reference-check guidance.