Voluntary Turnover
Every factor researchers have brought together to explain voluntary turnover, combined from 3 study models into one ranked driver set — the model the priority engine reasons over.
25 drivers · 3 contributing models · Handbook entry →
§1The drivers, by consensus
Ranked by how many independent models include each driver (consensus), then pooled effect. “In N of 3” 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.
| Driver | Consensus | Pooled r | Grade | Heterogeneity |
|---|---|---|---|---|
| Organizational Commitment | 3 / 3 | −0.27 | A | I²≈0.98 |
| Job Satisfaction | 3 / 3 | −0.23 | A | I²≈0.99 |
| Turnover Intention | 2 / 3 | +0.48 | A | I²≈1.00 |
| Job Search | 2 / 3 | +0.40 | A | — |
| Met Expectations | 2 / 3 | −0.16 | A | — |
| Person Organization Fit | 1 / 3 | −0.29 | A | — |
| Job Embeddedness | 1 / 3 | −0.26 | A | — |
| Perceived Job Alternatives | 1 / 3 | +0.23 | A | — |
| Emotional Exhaustion | 1 / 3 | +0.21 | A | — |
| Overall Job Performance | 1 / 3 | −0.21 | A | — |
| Role Clarity | 1 / 3 | −0.21 | A | — |
| Tenure | 1 / 3 | −0.20 | A | — |
| Work Engagement | 1 / 3 | −0.20 | A | — |
| Job Involvement | 1 / 3 | −0.19 | A | — |
| Perceived Organizational Support | 1 / 3 | −0.19 | A | — |
| Work Family Conflict | 1 / 3 | +0.19 | A | — |
| Psychological Contract Breach | 1 / 3 | +0.18 | A | — |
| Compensation | 1 / 3 | −0.17 | A | — |
| Organizational Justice | 1 / 3 | −0.17 | A | — |
| Role Ambiguity | 1 / 3 | +0.15 | A | — |
| Role Conflict | 1 / 3 | +0.15 | A | — |
| Internal Mobility | 1 / 3 | −0.11 | A | — |
| Age | 1 / 3 | −0.09 | A | — |
| Pay Level | 1 / 3 | −0.09 | A | — |
| Pay Satisfaction | 1 / 3 | −0.06 | — | — |
“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 study | Predictors | Grade |
|---|---|---|
| doi:10.1111/peps.12226 | 19 | A |
| doi:10.1177/014920630002600305 | 11 | A |
| doi:10.1111/j.1744-6570.1993.tb00874.x | 2 | A |