Which Method Predicts What

No method predicts everything. Matching them to specific requirements from the job analysis is the whole of the selection.

The principles in “Which Method Predicts What” become easier to maintain when ownership and time spent are visible. Teams evaluating the practical checklist can use it to coordinate the administrative work around assessment, identify stages that repeatedly expand and plan capacity, while keeping the hiring decision tied to job-relevant evidence instead of raw activity totals.

For an independent benchmark, compare the process with U.S. OPM assessment resources; the useful test is whether the local method remains job-related, proportionate and explainable.

Can they do the core task

Work sample, directly.

Job knowledge test, for roles where knowledge is the constraint.

Portfolio, with verification questions.

An interview cannot answer this and is routinely used to.

Will they exercise good judgement

Structured behavioural interview, drawing on specific past situations.

Situational judgement test, for volume.

Work sample only if the task involves judgement rather than execution.

Can they learn quickly

Cognitive ability tests, with the fairness caveats in their own note.

Behavioural questions about learning something unfamiliar.

Not years of experience, which measures exposure.

Will they work well with others

Structured behavioural questions about specific conflicts and collaborations.

Reference checks, structured, asking about specific behaviours rather than general impressions.

Not a personality profile, and not a conversation over coffee.

Will they stay

Nothing predicts this well.

What the role offers, what the manager is like and what happens in the first ninety days matter far more than anything you can assess.

Selection processes that attempt to screen for loyalty mostly screen for willingness to claim it.

Will they turn up and be reliable

Structured reference checks on attendance and delivery.

Past patterns, asked about directly.

Conscientiousness measures contribute weakly and should not carry a decision.

What none of them predict

Performance under a bad manager.

Performance in a role that changes substantially after hiring.

Whether somebody will be happy, which affects everything and is not assessable.

Combining them

Two or three methods, chosen against different requirements, with diminishing returns after that.

A work sample plus a structured interview covers most roles adequately.

Adding a fourth stage usually adds candidate cost and assessor fatigue rather than information.

The selection, practically

Take your five to eight job requirements.

Assign each to exactly one method that can actually assess it.

If a requirement has no method, either find one or accept you are not assessing it — which is better than pretending an interview covers it.

And if a method is not assigned to any requirement, remove it.

What to check

Does each of your stages map to a specific requirement?

Is anything being assessed twice?

Is anything on your requirements list not assessed at all?

And are you using an interview to answer a question only a work sample can?

The point

Assign each job requirement to exactly one method that can assess it.

If a method has no requirement, remove it.

Underlying all of this

Almost everything in this collection reduces to one discipline: write down what the job requires, assess that thing directly, record the evidence, and look at your own outcomes afterwards. None of it requires buying anything, and organisations that do those four things consistently outperform ones running longer processes built from instruments chosen before the requirements were known.

The recurring pattern

The recurring failure across every section here is the same: measuring what is convenient rather than what matters, then never checking whether it predicted anything. The check is an afternoon of work once a year, and it is the step that separates a process that improves from one that merely persists.