The Correlation Double Standard: Why Modest Psychometric Assessment Validity Can Still Matter

The Correlation Double Standard: Why Modest Psychometric Assessment Validity Can Still Matter

Every day, people take medication because it improves the likelihood of a better outcome—not because it guarantees one. Yet organizations often dismiss psychometric assessments for serving a similar decision-making purpose: improving the odds of a better hiring decision.

For HR leaders in the Middle East, asking rigorous questions about assessment evidence is important—and healthy.

Organizations should absolutely question assessment providers. They should ask what an assessment measures, whether it is relevant to the role and what evidence supports its usefulness.

The problem arises when an assessment is dismissed simply because its correlation with job performance appears modest.

An HR leader hears that an assessment correlates with performance at r = .20 or .30 and concludes:

“That sounds low.”

“Does that mean it is only 20% accurate?”

“If the correlation is not stronger, why should we use it?”

These reactions are understandable. But they reflect unrealistic expectations of what a correlation should look like when predicting something as complex as human performance.

What does a correlation actually tell us?

A correlation describes the relationship between two variables.

In recruitment, those variables might be a candidate’s assessment score and their subsequent job performance.

A correlation of r = .20 does not mean the assessment is 20% accurate. It does not mean it works for only one in five people. Nor does it mean every person with a higher score will outperform every person with a lower score.

It means that, across a group, higher assessment scores tend to be associated with stronger performance. The assessment identifies a pattern—it does not guarantee individual outcomes.

Psychometric assessments are not crystal balls. Their purpose is to improve the probability of making better hiring decisions, not eliminate uncertainty.

The benefit of a correlation is not the number itself. The benefit is how that relationship changes the decisions an organization makes.

What medicine can teach us about decision-making

People take medication because it improves the likelihood of a better outcome—not because it guarantees one.

A widely cited review by Gregory Meyer and colleagues expressed findings from medicine, psychology and other fields using the same statistical measure: the correlation coefficient. Among the historical examples were:

• Blood pressure medication and reduced stroke risk: r = .03
• Anti-inflammatory medication (e.g. ibuprofen) and pain reduction: r = .14
• General employee selection procedures and job performance: r = .27

These examples are not intended to suggest that medical treatments and psychometric assessments are equivalent. They simply illustrate an important principle: a relationship can appear statistically modest while still producing meaningful real-world outcomes.

People don't reject medication because it isn't perfect. They judge whether it improves outcomes compared with the alternative.

Psychometric assessments deserve the same standard. The question isn't whether the correlation is perfect—it's whether using the assessment leads to better hiring decisions than not using it.

What can a correlation of .30 mean in actual hiring decisions?

The Taylor–Russell model was developed to translate assessment validity into something more tangible: the expected percentage of successful people among those selected.

Consider a simplified example.

 Imagine an organization receives 1,000 applications and plans to hire 100 salespeople. Under their current hiring process, around half of the people appointed turn out to be successful (i.e. a 50% base rate). This means that without a predictive selection tool—essentially an (r = 0) scenario—the organization can expect roughly 50 successful appointments out of their 100 hires.

Using the Taylor–Russell model, introducing an assessment with a modest validity correlation of (r = .30) changes the math completely. Given the company's tight selection ratio (hiring only 10% of applicants), that assessment pushes the expected success rate from 50% up to approximately 71%.

In practical terms, instead of making 50 successful hires, the organization yields closer to 71. If a successful salesperson generates an average of $100,000 more in annual revenue than a marginal performer, securing 21 additional high-performing hires translates to an incremental $2.1 million in commercial value.

That is a massive organizational impact from a correlation (r = .30) that many might initially dismiss as "low." While exact figures always depend on the applicant pool and selection ratio, the principle remains constant: when selectiveness is high, even modest predictability delivers massive institutional utility.

The exact figures depend on the applicant pool and hiring ratio, but the principle remains the same.

This is an illustration, not a promise that every assessment with a correlation of .30 will produce the same result.

It nevertheless demonstrates the central point: a modest correlation can produce a noticeable improvement in the quality of hiring decisions.

From predictive validity to business value

The Taylor–Russell model demonstrates how a modest correlation can improve hiring outcomes. But HR leaders also need to understand what those improvements could mean for the organization.

Elev8's ROI Calculator takes the next step by estimating the potential business impact of improved predictive validity using your own hiring assumptions, such as recruitment volume, replacement costs and productivity.

It shifts the conversation from:

"Is a correlation of .30 large enough?"

to:

"What could a correlation of .30 be worth to our organization?"

That is the question business leaders should be asking.

Explore Elev8's ROI Calculator using your organization’s own hiring assumptions: Talent Assessment Platform for Recruitment & Development | Elev8 Assessments

Why perfect prediction is unrealistic

Job performance is influenced by many factors, including cognitive ability, personality, leadership, organizational culture, opportunity and team dynamics.

Expecting a single assessment to perfectly predict such a complex outcome is unrealistic.

A valid assessment contributes useful information. It does not explain the whole person.

The alternative is not free from error or bias

When organizations ask whether an assessment predicts performance perfectly, they are often comparing it with an impossible standard.

The real comparison is between an evidence-based assessment process and the process that would otherwise be used.

Without structured evidence, hiring decisions often rely on:

• Informal interviews
• A manager's intuition
• First impressions from a CV
• Personal chemistry between interviewer and candidate
• Different questions and criteria applied by different managers

These methods can feel more trustworthy because an experienced person is making the judgement rather than a tool producing a score.

But their uncertainty is hidden—not absent.

Unlike psychometric assessments, subjective hiring decisions rarely come with a published correlation coefficient or validation study. Their limitations are simply harder to see.

For example, informal interviews often involve different questions for different candidates, making fair comparisons difficult. First impressions, confirmation bias and personal chemistry can all influence decisions without the interviewer even realizing it. Structured methods reduce these inconsistencies by evaluating candidates against the same job-related criteria.

A well-designed psychometric assessment does not remove every source of bias or error.

It introduces a consistent, job-relevant source of evidence that can be validated, challenged and combined with other selection methods.

Not every assessment deserves trust

A more realistic interpretation of correlations should not lead organisations to accept every product described as a psychometric assessment.

Some tools are supported by robust scientific evidence. Others are not. Some were developed specifically for recruitment and prediction, while others were designed for self-awareness or team development and may not be appropriate for high-stakes employment decisions.

Before adopting an assessment, HR leaders should ask whether it is reliable, validated for its intended purpose, relevant to the role, fair, and demonstrably adds value to the existing selection process.

A correlation is one part of the evidence. It is not a universal stamp of approval.

Better decisions come from combining evidence

The strongest hiring decisions rarely rely on a single source of information.

Occupational psychology has consistently shown that combining multiple, job-relevant methods—such as psychometric assessments, structured interviews, work samples and cognitive ability measures—produces stronger prediction than relying on any single method alone. Each contributes different information, creating a more complete picture of the candidate. This can boost correlations from .2 or .3 to .4 or .5!

Psychometric assessments should therefore support professional judgement—not replace it.

Perhaps the real mistake isn't expecting too much from psychometric assessments. It's expecting one assessment to predict everything.

A more mature standard for psychometric assessment validity

HR leaders in the Middle East should continue to question assessment providers.

They should challenge unsupported promises, request relevant validity evidence and expect clarity about what an assessment can and cannot predict.

But rigorous scrutiny must also be statistically realistic.

A correlation of .20 or .30 should not be dismissed simply because it appears modest.

The question isn't whether an assessment predicts every individual perfectly.

The question is whether it helps your organization make better hiring decisions than it would otherwise make.

Medicine has accepted this principle for decades.

HR should too.

References

• Highhouse, S. (2008). Stubborn reliance on intuition and subjectivity in employee selection. Industrial and Organizational Psychology: Perspectives on Science and Practice, 1(3), 333–342. https://doi.org/10.1111/j.1754-9434.2008.00058.x
• Meyer, G. J., Finn, S. E., Eyde, L. D., Kay, G. G., Moreland, K. L., Dies, R. R., Eisman, E. J., Kubiszyn, T. W., & Reed, G. M. (2001). Psychological testing and psychological assessment: A review of evidence and issues. American Psychologist, 56(2), 128–165. https://doi.org/10.1037/0003-066X.56.2.128
• Sackett, P. R., Zhang, C., Berry, C. M., & Lievens, F. (2023). Revisiting the design of selection systems in light of new findings regarding the validity of widely used predictors. Industrial and Organizational Psychology: Perspectives on Science and Practice, 16(3), 283–300. https://doi.org/10.1017/iop.2023.24
• Taylor, H. C., & Russell, J. T. (1939). The relationship of validity coefficients to the practical effectiveness of tests in selection: discussion and tables. Journal of Applied Psychology, 23(5), 565–578. https://doi.org/10.1037/h0057079.

Frequently Asked Questions

Is a correlation of .30 considered good in psychometric assessments?

A correlation of .30 may appear modest, but in occupational psychology it can meaningfully improve hiring decisions—particularly when combined with other evidence-based selection methods.

Can psychometric assessments predict job performance?

Yes. Validated psychometric assessments can predict aspects of future job performance, but they should be used alongside structured interviews, work samples and other relevant selection methods.

Should psychometric assessments be used on their own?

No. Research consistently shows that combining multiple structured assessment methods provides stronger prediction than relying on a single tool.

--------

SEO Title

Psychometric Assessment Validity: Why Correlations Matter

H1 (Article Title)

The Correlation Double Standard: Why Modest Psychometric Assessment Validity Can Still Matter

URL Slug

psychometric-assessment-validity

Meta Description

Does a correlation of .30 really matter? Discover why modest correlations can improve hiring decisions and how HR should evaluate psychometric assessment validity.

Primary Keyword

psychometric assessment validity

Secondary Keywords

• predictive validity in recruitment
• psychometric assessments and job performance
• evidence-based hiring
• hiring decision quality
• talent assessment ROI
• correlation and job performance
• structured hiring
• employee selection validity

Suggested Internal Links

Use these naturally within the article:

• Psychometric Assessments (services page)
• ROI Calculator
• Trait vs Type-Based Psychometric Assessments
• Has AI Really Broken Assessments?

Suggested Feature Image – see embedded in athe article.

Image Alt Text

1. Comparison showing how modest correlations can improve medical and hiring outcomes.
2. Illustrative Taylor–Russell example showing expected successful hires increasing from 50 out of 100 at r = 0 to 71 out of 100 at r = .30.

Social Share Title

Would You Reject a Medical Treatment Because Its Correlation Was Only .30?

This is much stronger than the SEO title for LinkedIn. It creates curiosity and directly challenges assumptions.

Social Share Description

Medical treatments don't have to guarantee success to be valuable. So why do we expect psychometric assessments to predict human performance perfectly? This article explores the double standard in how HR interprets correlations and what decades of research tell us about making better hiring decisions.