Registry · Priors

construct.work_engagement predicts construct.turnover_intention

normal · weakly_informative · 3 studies · N = 29,832

Distribution

Storage scale (Fisher z)
Prior PDF · normalnormal distribution. Storage scale (Fisher z). 95% CI ≈ [-0.545, -0.421]; mean ≈ -0.483.-0.609-0.546-0.483-0.419-0.356z0density

mean ≈ -0.483 · 95% CI ≈ [-0.545, -0.421]

Reader scale (r)
Prior PDF · normalnormal distribution. Reader scale (r). 95% CI ≈ [-0.498, -0.399]; mean ≈ -0.448.-0.550-0.499-0.448-0.398-0.347r0density

mean ≈ -0.448 · 95% CI ≈ [-0.498, -0.399]

Weakly informative prior. This prior is weakly informative. It will nudge your posterior but won't overwhelm it; expect data to do most of the work in modest samples.

Intervals

Confidence interval (95%) — uncertainty about the mean ρ
[-0.50, -0.40]
Credibility interval (95%) — distribution of the true effect across settings (the Bayesian prior)
[-0.52, -0.37]
I² (heterogeneity) — share of total variance from between-study differences
96% — the contributing studies disagree almost completely; read the pooled value as a midpoint of conflicting findings, not as one population's effect

The true effect is ~constant across settings — it generalizes.

SD_ρ≈0 — true effect is ~constant across settings; generalizes (observed-score scale until artifact correction, PRN-058)

Evidence provenance

published ρ=-0.45 (k=3, high heterogeneity (I²=0.96)); no primary-deployment evidence yet

Sourceρ (r)Scope
Published literature-0.45k = 3 · N = 29,832

No primary-deployment evidence yet — this prior rests on published literature alone. As anonymized, aggregated effect sizes from real deployments are contributed, they appear here as a distinct, publication-bias-free source, fused with the literature into a posterior estimate.

Code

Drop this prior straight into your model. Snippets generated from the synthesized distribution + parameters.

target += normal_lpdf(beta | -0.482763, 0.0316614);
beta = pm.Normal("beta", mu=-0.482763, sigma=0.0316614)
brms::prior(normal(-0.482763, 0.0316614), class = "b")
# base R sample
rnorm(N, mean = -0.482763, sd = 0.0316614)
np.random.normal(loc=-0.482763, scale=0.0316614, size=N)

Parameters

FamilyParameters
normalI2 = 0.9559, mu = -0.4828, sigma = 0.03166, r_mean = -0.4485, k_studies = 3.000, tau_squared = 0.002296, fisher_z_bias = -0.01039, r_mean_bare_bones = -0.4381

Synthesis

Method
random_effects_meta
Informativeness
weakly_informative
Replication status
meta-analytic
k_studies
3
n_total
29,832
Last updated
2026-07-16T01:10:13.824Z

Quality distribution

GradeCount
A1
B2
C0
D0

Source articles

The research this prior is synthesized from — each is a full dossier (findings, the models it informs, and what the literature says).

Contributing effect sizes

Effect-size detail pages land with a later sub-ticket; for now, ids link to the filtered list. Browse all rows via /registry/effects.