Spurious precision in meta-analysis of observational research

Abstract

Meta-analysis assigns more weight to studies with smaller standard errors to maximize the precision of the overall estimate. In observational settings, however, standard errors are shaped by methodological decisions. These decisions can interact with publication bias and p-hacking, potentially leading to spuriously precise results reported by primary studies. Here we show that such spurious precision undermines standard meta-analytic techniques, including inverse-variance weighting and bias corrections based on the funnel plot. Through simulations and large-scale empirical applications, we find that selection models do not resolve the issue. In some cases, a simple unweighted mean of reported estimates outperforms widely used correction methods. We introduce MAIVE (Meta-Analysis Instrumental Variable Estimator), an approach that reduces bias by using sample size as an instrument for reported precision. MAIVE offers a simple and robust solution for improving the reliability of meta-analyses in the presence of spurious precision.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 10 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.