Robust point and variance estimation for meta‐analyses with selective reporting and dependent effect sizes
Bibliographic record
Abstract
Abstract Meta‐analysis produces a quantitative synthesis of evidence‐based knowledge, shaping not only research trends but also policies and practices in biology. However, two statistical issues, selective reporting and statistical dependence, can severely distort meta‐analytic parameter estimation and inference. Here, we re‐analyse 448 meta‐analyses to demonstrate a new two‐step procedure to deal with two common challenges in biological meta‐analyses that often occur simultaneously: publication bias and non‐independence. First, we employ bias‐robust weighting schemes under the generalized least square estimator to obtain average effect sizes that are more robust to selective reporting. We then use cluster‐robust variance estimation to account for statistical dependence, reducing bias in estimating standard errors and ensuring valid statistical inference. The first step of our approach demonstrates comparable performance in estimating average effect sizes to the existing publication‐bias adjustment methods in the presence of selective reporting. This equivalence holds across two publication bias selection processes. The second step achieves estimates of standard errors consistent with the multilevel meta‐analytic model, a benchmark method with adequate control of Type I error rates for multiple, statistically dependent effect sizes. Re‐analyses of 448 meta‐analyses show that ignoring these two issues tends to overestimate effect sizes by an average of 110% and underestimate standard errors by 120%. To facilitate implementation, we have developed a website including a step‐by‐step tutorial. Complementing current meta‐analytic workflows with the proposed method as a sensitivity analysis can facilitate a transition to a more robust approach in quantitative evidence synthesis.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.190 | 0.524 |
| Meta-epidemiology (narrow) | 0.004 | 0.004 |
| Meta-epidemiology (broad) | 0.007 | 0.018 |
| Bibliometrics | 0.010 | 0.011 |
| Science and technology studies | 0.001 | 0.002 |
| Scholarly communication | 0.006 | 0.004 |
| Open science | 0.007 | 0.006 |
| Research integrity | 0.004 | 0.008 |
| Insufficient payload (model declined to judge) | 0.011 | 0.004 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; the direct Gemma label and the distilled Codex classifier agree on what is shown here.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".