The complete model

Overall Job Performance

Every factor researchers have brought together to explain overall job performance, combined from 4 study models into one ranked driver set — the model the priority engine reasons over.

18 drivers · 4 contributing models · Handbook entry →

§1The drivers, by consensus

Ranked by how many independent models include each driver (consensus), then pooled effect. “In N of 4” is a robustness signal — a driver in many models is one many researchers agreed matters; a driver in one is a single lead worth following.

DriverConsensusPooled rGradeHeterogeneity
Conscientiousness3 / 4+0.22A
Structured Employment Interviews2 / 4+0.44A
General Mental Ability GMA2 / 4+0.43AI²≈1.00
Structured Interviews2 / 4+0.42A
Job Knowledge1 / 4+0.48A
Biographical Data Measures Biodata1 / 4+0.38A
Work Sample1 / 4+0.33A
Integrity1 / 4+0.31A
Assessment Centers1 / 4+0.28A
Reference Checks1 / 4+0.26A
Situational Judgment Tests1 / 4+0.26A
Job Experience Years1 / 4+0.18A
Neuroticism1 / 4−0.14A
Extraversion1 / 4+0.10A
Emotional Stability1 / 4+0.09A
Financial Incentives1 / 4Astudy-reported
Job Knowledge Test1 / 4Astudy-reported
Pay For Performance1 / 4Astudy-reported

“Pooled r” is the meta-analytic prior where one exists; drivers marked study-reported have a model but not yet a pooled prior — a concrete acquisition target. This is the value-of-information view in miniature: the priority engine can rank which driver to measure first by consensus, strength, and how much is still unknown.

§2The models it’s built from

Each contributing study model — the researcher’s theory of this outcome — with its size and grade, so the meta-model is auditable back to its sources.

Source studyPredictorsGrade
doi:10.1037/apl000099410A
doi:10.1037/0033-2909.124.2.2628A
doi:10.1037/0021-9010.85.6.8693A
doi:10.1111/joop.120392A
Composed live from the registry. The complete model is the union of all study-level models of this outcome; study-level models are derived from the effect corpus and re-derived every loop cycle.