Ensuring Selection Validity: Removing a Barrier to Neuroinclusive Hiring

When High Scores Predict Low Performance

Zia scored in the 95th percentile on her company’s selection set of assessments. Fast on the cognitive ability test, desirable personality profile, flawless video interview. The automated selection system flagged her as a “top candidate” for the cybersecurity analyst role. Meanwhile, Marcus barely met the cut-off: slower on timed tests, awkward in video interviews, flagged for “unusual communication patterns.”

Six months later, performance reviews told a different story.

Zia had missed several critical security vulnerabilities. Marcus caught more in his first month than his more experienced colleagues. The assessment system had predicted exactly the opposite of the actual performance.

This isn’t hypothetical. It’s happening everywhere – the supposedly sophisticated hiring technologies fail the core test of validity: predicting who will excel at the job.

The Double Validity Problem

Problem 1: Measuring the Wrong Things

Most hiring failures stem from fundamental selection validity errors, such as measuring traits that don’t predict job performance. Before any assessment can be valid, organizations must understand what the job actually requires through systematic job analysis. But instead, many just use traditional selection methods that often emphasize charisma and fast responses, even when these are irrelevant to job performance.

For example, attention to detail is essential for cybersecurity jobs. And yet, in the Zia and Marcus example, it was not as prominent in the assessments used as were irrelevant personality and communication characteristics and speed of processing. No wonder a better candidate barely passed – and it is likely that many candidates better qualified than Zia were rejected.

Here are some other examples of irrelevant assessments that may influence hiring outcomes in situations similar to Zia outscoring Marcus:

• Video interviews evaluating eye contact for cybersecurity roles where social interaction isn’t job-critical.
• Timed cognitive tests rewarding speed over accuracy when the job requires thorough problem-solving and attention to detail.
• Personality assessment predicting generic “ideal personality” defined as extraversion or “charisma” instead of measuring specific work capabilities.

The pattern of the problem is clear: extensively assessing general traits while ignoring specific core functions and job skills. The solution is also clear: be specific. If small talk isn’t part of the job, don’t measure it. If fact-checking is essential, test it directly.

Problem 2: Measurement Instruments that Don’t Work for all Populations 

Beyond measuring irrelevant traits, there is an additional issue: not all measurement instruments are valid for all populations – they should be evaluated for validity for specific groups. Currently, however, instruments are often designed and validated based on the neurotypical population, who could be very different from neurodivergent applicants.

Even job-relevant assessments can fail neurodivergent candidates through:

• Format bias that contaminates otherwise valid measures
• Environmental factors (sensory distractions, testing anxiety)
• Communication style variations influencing instruction interpretation

For example, it matters HOW intelligence is measured. Research shows that autistic children scored on average 30 percentile points higher on Raven’s Progressive Matrices than on Wechsler IQ tests, with some cases showing more than 70 percentile point differences. Raven’s measures intelligence with the focus on visual pattern understanding, while Wechsler relies heavily on verbal reasoning under time pressure.

This means that a hiring assessment might be perfectly valid for measuring analytical skills in neurotypical candidates, but completely miss the strengths of neurodivergent candidates who may process information differently. This calls for validating assessments for autistic, ADHD, dyslexic, and other neurodivergent populations. Industry research found negligible score differences between neurotypical and neurodivergent individuals on properly designed cognitive assessments, challenging negative assumptions about neurodivergent capabilities. Research with Specialisterne – placed employees in Canada – also shows that careful selection and support result in successful autism employment.

The Solution Framework

Step 1: Start with Job Analysis

Document critical tasks, essential competencies, and performance standards before designing selection assessments. Having a clear record of core job-relevant knowledge, skills, abilities, and other characteristics provides the foundation for valid selection. Focus on demonstrable competencies rather than personality proxies. Can the candidate perform the required tasks to the required standards?

Step 2: Audit Current Tools

• Compare assessments against job analysis findings
• Identify components measuring irrelevant traits
• Examine whether validation studies included neurodivergent participants. If not, look for tools that have been validated with neurodivergent populations.
• Test for differential performance across populations

Step 3: Prioritize Work-Relevant Assessment

• Work Samples as Gold Standard: The highest validity comes from candidates performing actual job tasks. Software engineers should code, teachers should demonstrate teaching skills, researchers should evaluate sources, and customer service representatives must demonstrate how they would engage with customers.
• Structured Interviews: Sometimes, skill demonstration is all that is needed. But if you must use interviews, then use structured interviews with understanding that humans may process information differently, yet be equally capable of doing the job. Focus on examples of past performance relevant to essential job functions rather than hypothetical scenarios, and avoid vague or intentionally tricky questions unless dealing with “trickiness” is part of the job description.
• Eliminate Non-Job-Related Criteria: For example, remove extraversion assessments for precision manufacturing and analyst roles, timed tasks for jobs that require attention to detail, and cultural fit measures that correlate with neurotypical communication patterns.

Step 4: Continue Validating Your Selection System Across Populations

Even job-relevant assessments require population-specific and context-specific validation:

• Verify equivalent predictive validity across neurological profiles
• Test for environmental or procedural bias
• Offer alternative formats when needed and document accommodation effectiveness, such as taking tests in quiet environments and with instructions offered in different formats

Continue evaluating performance, retention, satisfaction, and other outcomes as time goes on, and be prepared to adjust the system as necessary.

Moving Forward

The most effective inclusive hiring addresses both validity challenges – overall relevance and validity for specific groups – simultaneously. This will lead to increased fairness and superior business results via:

• Enhanced Predictive Power: Job-relevant assessments validated across populations provide the most accurate hiring predictions.
• Productivity: When organizations focus on what actually predicts performance rather than irrelevant proxies, they make better hires regardless of neurological profile.

In other words, companies are much more likely to hire employees with an excellent fit for specific positions who become top performers, like Marcus.

• Reduced Legal Risk: Assessments that are both job-relevant and population-validated meet the highest standards for legal defensibility.

The promise of “scientific” hiring can only be fulfilled when science itself is inclusive and valid. Valid assessment is not about lowering standards or making exceptions. It’s about measuring what actually matters: can candidates do the job? When assessments focus on job-relevant capabilities and work across populations, everyone benefits.