Which math method best guesses the real answer from many small studies?
A Comparative Evaluation of Psychometric Meta‐Analysis Methods in Management and Applied Psychology: Toward a Nuanced Understanding of Their Accuracy
Not medical advice. For informational purposes only. Always consult a healthcare professional. Terms
Surprising Findings
The most widely used method (Schmidt and Hunter RE) is not the most accurate, despite its popularity.
It contradicts the assumption that widely adopted scientific methods are also the most optimal — suggests inertia in methodological practice.
Practical Takeaways
Researchers should consider using blended random-effects meta-analysis with random-effects weights, particularly the Schmidt and Hunter (2015)-based version, for slightly more accurate estimates.
Not medical advice. For informational purposes only. Always consult a healthcare professional. Terms
Surprising Findings
The most widely used method (Schmidt and Hunter RE) is not the most accurate, despite its popularity.
It contradicts the assumption that widely adopted scientific methods are also the most optimal — suggests inertia in methodological practice.
Practical Takeaways
Researchers should consider using blended random-effects meta-analysis with random-effects weights, particularly the Schmidt and Hunter (2015)-based version, for slightly more accurate estimates.
Publication
Journal
Personnel Psychology
Year
2025
Authors
In‐Sue Oh, Huy Le, Frank L. Schmidt
Related Content
Claims (5)
Putting together lots of small studies gives a better guess about what's really going on in the whole population than looking at just one small study.
Random-effects meta-analysis is better than fixed-effect at guessing the true average result across different psychology and management studies, especially when fixing common data problems.
This math method for combining study results seems to give slightly more accurate answers than other similar methods when figuring out the true effect across many studies.
Some math methods for combining study results work better than others when trying to understand how much results vary — ones that adjust for more uncertainty give clearer answers, according to computer simulations in psychology research.
An old but popular way of combining study results isn't as good as some newer methods when it comes to accuracy in computer simulations that fix common measurement errors.