What Schmidt & Hunter Taught Us About Selection Method Validity and What HR Still Gets Wrong

"The validity of GMA measures for predicting job performance is now, after over 85 years of research, well established. There is no longer any scientific basis for denying GMA's validity. In a competitive world, organizations using suboptimal selection methods unnecessarily handicap themselves." — Frank L. Schmidt & John E. Hunter, Psychological Bulletin (1998)

Executive Summary

What if the most influential research on hiring validity was published over 25 years ago—and most organizations still haven't implemented it? In 1998, Frank L. Schmidt and John E. Hunter published a landmark meta-analysis synthesizing 85 years of personnel selection research. Their work identified which hiring methods actually predict job performance—and which don't.

Despite being one of the most cited and influential publications in Industrial/Organizational Psychology, their findings remain largely ignored by organizations. Most continue using low-validity methods while overlooking proven predictors.

This analysis explains what Schmidt & Hunter discovered, why their findings challenged conventional hiring practices, and why understanding validity coefficients is critical for building high-performing organizations.

The Problem: Why HR Hiring Practices Diverge from Science

Most organizations continue using hiring methods with minimal predictive validity. Schmidt & Hunter note that many organizations rely on unstructured interviews, resumes, and biographical information—methods with validity coefficients ranging from 0.10 to 0.38. Meanwhile, they avoid or under-utilize general mental ability (GMA) tests and structured interviews—methods with validity coefficients of 0.51 to 0.58 or higher.

This divergence between what research proves works and what organizations actually use creates enormous competitive disadvantages. As Schmidt and Hunter stated: "In a competitive world, organizations using suboptimal selection methods unnecessarily handicap themselves."

The economic stakes are massive: gains from improving selection method validity can amount to literally millions of dollars over time through increased output, reduced turnover, and improved quality. Yet this science remains largely unimplemented.

Understanding Validity: The Foundation of Selection Science

Predictive validity is the correlation between a selection method and future job performance. It's measured using validity coefficients (r) ranging from -1.0 to +1.0, where:

• r = 0.35 or higher is considered "very beneficial" in practical contexts

• r = 0.20-0.34 is considered "somewhat useful"

• r < 0.20 is considered "unlikely to be useful"

The relationship between validity and utility is directly proportional: increasing the predictive validity coefficient increases the utility of the selection process and organizational performance. A selection method with r = 0.65 is approximately 3.8 times more effective at predicting job success than a method with r = 0.17.

The Schmidt & Hunter Meta-Analysis: What 85 Years of Research Reveals

Schmidt & Hunter's meta-analysis systematically reviewed validity research on 19 different selection methods, examining literally thousands of studies spanning nine decades. Their approach—now called the Schmidt-Hunter method of meta-analysis—corrected for statistical artifacts (sampling error, measurement error, range restriction) that had previously obscured true validity relationships.

This correction was revolutionary. Prior to their work, many validity estimates were substantially lower than their true values because studies underestimated relationships by not accounting for these artifacts.

The Validity Hierarchy: Ranked by Predictive Power

Schmidt & Hunter's findings ranked selection methods by their ability to predict job performance:

• Work Sample Tests: r = .54 — Excellent predictor; limited applicability

• General Mental Ability (GMA): r = .51-.67 — Excellent predictor; universally applicable

• Structured Interview: r = .58 — Strong predictor; practical and widely applicable

• Integrity Tests: r = .41 — Good predictor

• Conscientiousness (Big 5): r = .31 — Moderate predictor

• Unstructured Interview: r = .38 — Weak to moderate predictor

• Job Experience (years): r = .18 — Weak predictor

• Education Level: r = .10 — Weak predictor

• Resume/Application Form: r = .13 — Very weak predictor

• Handwriting Analysis: r = .00 — No predictive validity

Key Finding: The Primacy of General Mental Ability

General Mental Ability (GMA)—general cognitive ability or intelligence—is the single strongest predictor of job performance across virtually all job types.

Empirical Evidence: The original Schmidt & Hunter (1998) meta-analysis found GMA had a validity coefficient of r = .51. Later meta-analyses using refined methods found that GMA validity is actually higher when corrected appropriately for range restriction:

• Professional and managerial jobs: r = .74

• Medium-complexity jobs (62% of all U.S. jobs): r = .66

• Unskilled jobs: r = .39

• Average across all jobs: r = .67

To put this in context: literally thousands of studies over nine decades support GMA's validity as a job performance predictor—vastly more evidence than any other selection method.

In contrast, Schmidt & Hunter found that: Education level predicts job performance at r = .10 (essentially useless); Years of experience predicts at r = .18 (weak); Resume information predicts at r = .13 (very weak). The implication is stark: organizations routinely prioritize information (education, experience, resume content) that barely predicts performance while avoiding methods (GMA tests) that powerfully predict it.

The Interview Validity Surprise: Why Structure Matters

One of Schmidt & Hunter's most consequential findings involved employment interviews—a ubiquitous selection method that most organizations use.

Unstructured interviews (the traditional "let's talk about your background" approach) show validity coefficients of r = .38 when corrected appropriately for range restriction artifacts.

Structured interviews (where all candidates are asked the same job-relevant questions in the same sequence, with systematic rating scales) show validity coefficients of r = .58.

This represents a remarkable difference: structured interviews are approximately 1.5 times more predictive than unstructured interviews at predicting actual job performance.

Yet research on what HR managers believe is equally revealing: In one study of 201 HR professionals, when unstructured and structured interview validity coefficients were presented alone, HR managers rated the unstructured interview as more effective despite its lower validity coefficient (d = -0.30). It wasn't until validity coefficients were presented side-by-side that structured interviews were strongly favored. This suggests that intuition drives decisions: unstructured interviews feel better because they allow personal impression. But empirical data shows structured interviews work better.

The Power of Combinations: When Selection Methods Complement Each Other

A key Schmidt & Hunter insight: while individual selection methods vary in validity, combining methods strategically produces even stronger predictions.

Empirical Results: Key combinations and their composite validity:

• GMA + Integrity Test: r = .65 — Highest validity; excellent practical utility

• GMA + Structured Interview: r = .63 — High validity; excellent practical utility

• GMA + Work Sample: r = .63 — High validity; more expensive

• Conscientiousness (Big 5) alone: r = .31 — Moderate; low utility

• Unstructured Interview alone: r = .38 — Weak; widely used but ineffective

The composite validity of .65 (GMA + Integrity) represents a 26% improvement over GMA alone (r = .51), demonstrating that thoughtfully selected combinations capture variance that single methods miss.

What HR Gets Wrong: The Implementation Gap

Despite 85+ years of research and meta-analytic consensus, organizations systematically under-utilize high-validity methods while over-relying on low-validity approaches:

Error 1: Over-Reliance on Unstructured Interviews. Most organizations still conduct unstructured, conversational interviews (validity r = .38) despite structured interviews' demonstrated superiority (r = .58). The reason: unstructured interviews feel natural and comfortable.

Error 2: Dismissing Cognitive Ability Tests. Despite GMA's proven validity (r = .51 to .67), many organizations avoid formal cognitive ability testing because they perceive it as "unfair" or believe "cultural fit" matters more. Research contradicts this.

Error 3: Prioritizing Experience Over Ability. Schmidt & Hunter found that years of job experience predicts performance at only r = .18. Yet organizations often prioritize "5+ years of experience" in job requirements, assuming tenure predicts performance. Research shows it doesn't—at least not reliably.

Error 4: Overweighting Formal Education. Formal education level predicts job performance at only r = .10. Organizations requiring specific degrees exclude high-ability candidates without those credentials.

Conclusion: The Research-Practice Gap in Selection

Eighty-five years of rigorous research has definitively answered the question: Which hiring methods predict employee performance? The answer is clear, consistent, and powerful: General mental ability, structured interviews, work samples, and integrity tests are the strongest predictors.

Yet most organizations continue using weaker methods. Schmidt & Hunter's work reveals not a gap in knowledge (the research has been available for decades) but a gap between what we know works and what we actually do.

Organizations embracing Schmidt & Hunter's findings—using validated GMA assessments and structured interviews—gain a competitive advantage. They hire better, retain better performers longer, and achieve higher productivity. For any organization serious about performance, understanding and implementing selection method validity isn't optional—it's essential.

Ready to Transform Your Selection System with Evidence-Based Methods?

Organization Learning Labs offers validated selection method assessments, structured interview design, and implementation support grounded in decades of I/O Psychology research, including the foundational work of Schmidt & Hunter. Contact us at research@organizationlearninglabs.com.

References

Hunter, J. E., & Hunter, R. F. (1984). Validity and utility of alternative predictors of job performance. Psychological Bulletin, 96(1), 72-98.

McDaniel, M. A., Whetzel, D. L., Schmidt, F. L., & Maurer, S. D. (1994). The validity of employment interviews. Journal of Applied Psychology, 79(4), 599-616.

Sackett, P. R., Zhang, C., Berry, C. M., & Lievens, F. (2021). Revisiting meta-analytic estimates of validity in personnel selection. Journal of Applied Psychology, 107(11), 1-28.

Schmidt, F. L., & Hunter, J. E. (1998). The validity and utility of selection methods in personnel psychology. Psychological Bulletin, 124(2), 262-274.