Bayesian sensitivity analysis for unmeasured confounding in causal mediation analysis
Bibliographic record
Abstract
Causal mediation analysis techniques enable investigators to examine whether the effect of the exposure on an outcome is mediated by some intermediate variable. Motivated by a data example from epidemiology, we consider estimation of natural direct and indirect effects on a survival outcome. An important concern is bias from confounders that may be unmeasured. Estimating natural direct and indirect effects requires an elaborate series of assumptions in order to identify the target quantities. The analyst must carefully measure and adjust for important predictors of the exposure, mediator and outcome. Omitting important confounders may bias the results in a way that is difficult to predict. In recent years, several methods have been proposed to explore sensitivity to unmeasured confounding in mediation analysis. However, many of these methods limit complexity by relying on a handful of sensitivity parameters that are difficult to interpret, or alternatively, by assuming that specific patterns of unmeasured confounding are absent. Instead, we propose a simple Bayesian sensitivity analysis technique that is indexed by four bias parameters. Our method has the unique advantage that it is able to simultaneously assess unmeasured confounding in the mediator-outcome, exposure-outcome and exposure-mediator relationships. It is a natural Bayesian extension of the sensitivity analysis methodologies of VanderWeele, which have been widely used in the epidemiology literature. We present simulation findings, and additionally, we illustrate the method in an epidemiological study of mortality rates in criminal offenders from British Columbia.
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 | 0.131 | 0.353 |
| Meta-epidemiology (narrow) | 0.002 | 0.002 |
| Meta-epidemiology (broad) | 0.004 | 0.005 |
| Bibliometrics | 0.006 | 0.004 |
| Science and technology studies | 0.002 | 0.004 |
| Scholarly communication | 0.003 | 0.006 |
| Open science | 0.004 | 0.006 |
| Research integrity | 0.004 | 0.007 |
| Insufficient payload (model declined to judge) | 0.010 | 0.001 |
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; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
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".