Book a meeting
All articles
Future of Work16 November 2026·6 min read

Remote-Readiness Tests vs Better Role Assessment: Two-Thirds of the Remote Productivity Gap Was Selection, Not Remote Work

A Federal Reserve Bank of New York study found remote call-centre workers answered 12% fewer calls per hour - but two-thirds of that gap was who chose remote jobs, not the effect of remote work. What that means for whether a remote-readiness assessment earns its place in your funnel.

By AssessAll Editorial

A remote-readiness assessment is a pre-hire instrument that scores self-management, written communication and home-working discipline to predict performance outside an office. The research base does not establish it as a construct distinct from predictors you already measure. Most of the observed remote performance gap reflects who applies for remote roles rather than a trait a test can isolate.

That distinction decides whether adding one is a sound investment or an expensive way to re-measure conscientiousness.

The finding that reframes the question

The most careful study of remote productivity separates two things most hiring teams blend together. Emanuel and Harrington's Federal Reserve Bank of New York staff report (no. 1061, May 2023) followed the call centres of a US Fortune 500 firm that hired 344 remote and 1,592 on-site workers into the same jobs between January 2019 and October 2021.

Before the pandemic, remote workers answered 12% fewer calls per hour than on-site workers. The instinct is to read that as the cost of remote work, and to screen for it. The decomposition says otherwise:

  • When offices closed, formerly on-site workers' productivity fell about 4% relative to colleagues who were already remote. That is the causal effect of going remote.
  • Even once everyone was remote, the originally-remote group remained about 8% less productive.
  • Roughly one-third of the original gap was the treatment effect of remote work. About two-thirds was negative selection into remote jobs.

Then the part that matters most for assessment design: the selection gap narrowed as a return to the office came to seem distant, falling to no appreciable difference by winter 2021. The signal a remote-readiness test would have picked up in 2019 was a property of that moment's applicant pool, not a durable attribute of people. When remote roles stopped being a self-selected niche, two-thirds of the thing being measured dissolved.

A test calibrated on a transient labour-market composition is not measuring readiness. It is measuring who was willing to take a remote job when remote jobs were unusual.

What the hybrid evidence says about the risk being priced

If the worry is that distributed work degrades output, the best-identified evidence does not support screening for it. Bloom, Han and Liang's randomised controlled trial in Nature (2024) assigned 1,612 Trip.com employees — 395 managers and 1,217 non-managers — to hybrid or fully on-site work by odd or even birthday, two days at home per week.

What improved

  • Quit rates fell from 7.20% in the control group to 4.80% in the treatment group, a one-third reduction.
  • Among non-managers, attrition fell from 8.59% to 5.33%, a 40% reduction.
  • Among women, it fell from 9.19% to 4.18%, a 54% reduction.
  • Self-reported job satisfaction rose from 7.84 to 8.19.

What did not change

  • Performance reviews showed no difference across four half-year cycles, with equivalence tested inside a band of ±0.5 of a grade on a five-point scale.
  • Promotion rates showed no evidence of a difference, overall or by subgroup.
  • Lines of code for 653 engineers showed no difference.

A selection hurdle exists to reduce the probability of a bad outcome. When a randomised trial finds the performance outcome unchanged and retention materially better, a pre-hire filter for hybrid readiness is pricing a risk the strongest available evidence puts near zero.

Where the remote-specific risk actually sits

Remote work is not consequence-free, but the consequences land somewhere a selection test cannot reach. Gajendran and Harrison's meta-analysis of 46 studies covering 12,883 employees found telecommuting related to perceived autonomy at ρ = .22, job satisfaction at ρ = .10, supervisor-rated or objective performance at ρ = .19, turnover intent at ρ = −.10 and role stress at ρ = −.13.

Coworker relationship quality came out at roughly ρ = .00 overall — but that average conceals the real pattern. At high intensity, 2.5 or more days a week, the relationship was −.19. At low intensity, −.03.

The penalty is relational, and it scales with how much time people spend apart. That is a coordination and work-design variable, not an individual difference. No pre-hire score fixes it; scheduling, overlap hours and communication norms do.

The same holds for the quality findings in the New York Fed data. Hold times rose about 11% and call-back rates about 3% among workers who shifted to remote, and those declines were concentrated among less experienced workers. That points at onboarding and ramp design, not screening. So does the promotion finding: a year after hire, 44.0% of on-site hires had been promoted against 20.9% of remote hires, a gap that vanished once offices closed. Proximity bias in the promotion process is not a candidate defect.

The two approaches, compared

Adding a dedicated remote-readiness assessment

  • What it has to prove: incremental validity over the predictors already in your battery.
  • Evidence available today: no separately validated construct distinct from general predictors.
  • Legal position: a new scored hurdle is a selection procedure, subject to the EEOC's guidance on employment tests and its own adverse-impact scrutiny.
  • Main failure mode: scoring a labour-market artefact, or charging candidates and yourself twice for conscientiousness.

Strengthening the role assessment you already run

  • What it has to prove: job-relatedness, established through job analysis that names asynchronous demands as actual tasks.
  • Evidence available today: the standard predictors, already validated for the work.
  • Legal position: the same standard, against documentation you already maintain.
  • Main failure mode: missing genuinely async-specific demands if the job analysis never looked for them.

The incremental-validity test to run before you buy

  1. Redo the job analysis and name the asynchronous demands — written handover, self-directed planning, working without real-time escalation — as tasks, not traits.
  2. Correlate the candidate instrument against your existing predictors. A high correlation with conscientiousness means you are paying twice for one construct.
  3. Run a hierarchical regression. Does it add variance in performance beyond the current battery?
  4. Check adverse impact on the new hurdle in isolation, not only on the funnel as a whole.
  5. Re-check calibration whenever the applicant pool shifts. The New York Fed data shows how fast a selection signal can decay.
  6. Document the reasoning against the Standards for Educational and Psychological Testing and the SIOP Principles.

Steps 2 and 3 need scores, not opinions, which is where consumption-priced testing helps: running a validity pilot on a few hundred candidates at ₹30 or US$0.50 per assessment costs less than a year of licence fees for an instrument that may add nothing.

When a dedicated remote assessment is the right call

There are conditions where the answer flips, and they share a feature: the remote element is the job, not a context for it.

  • Fully asynchronous teams with little or no overlap. When a role's primary output is written decisions that colleagues act on hours later, written asynchronous judgment is a core task. Assess it directly — as a work sample, scored against a rubric. Scenario-based exercises with AI-assisted grading make that practical at volume, and AssessAll's custom assessment work exists for exactly this kind of role-specific instrument.
  • Roles with no ramp. Contractor and short-engagement models that offer no onboarding cannot rely on the tenure effects that the New York Fed data shows matter. Selection carries more weight because development carries none.
  • Genuinely self-directed work planning. If job analysis surfaces unsupervised prioritisation as a central task, measure it — as a task, with job-relevant content.

In each case the defensible instrument is a work sample of the job's actual demands. That is not a remote-readiness test with better branding; it is a role assessment that finally includes the whole role.

The takeaway

Two-thirds of the remote productivity gap in the best-identified study was selection, and that selection effect faded within two years as the applicant pool normalised. Before buying an instrument to screen for remote readiness, run the job analysis and the incremental-validity check — and fix onboarding, overlap hours and promotion visibility, where the evidence says the remote penalty actually lives.

#remote hiring#selection#incremental validity#work design#skills-based hiring

Measure it, don't guess it.

Start free with 100 credits — or write to solutions@bodhih.com.

Start free