selectMeta is a package implementing parametric and nonparametric weight functions to model publication bias in meta-analyses. The methods include a non-increasing variant of the nonparametric weight function from Dear & Begg (1992) and a differential evolution algorithm for optimization. It was authored by Kaspar Rufibach and includes functions for computing confidence intervals and simulation-based p-values for effect sizes.
Use Cases
- Modeling publication bias in meta-analyses based on selection models and weighted probability distributions.
- Estimating a non-increasing weight function for selection processes based on the method from Rufibach (2011).
- Computing a confidence interval for an overall effect size adjusted for selection bias.
- Assessing the null hypothesis of no selection bias using a simulation-based p-value.
Strengths
- Implements several parametric and nonparametric weight functions for modeling selection bias.
- Includes a novel non-increasing variant of a nonparametric weight function proposed in 2011.
- Provides methods for confidence interval calculation and hypothesis testing for effect sizes.
Limitations
- Description metadata is limited; actual data quality requires manual inspection after download.
- Column-level documentation is absent; field semantics must be inferred after download.
- Row count is unknown, which may limit suitability assessment.
Provenance
- Source
- Kaspar Rufibach
- Collection Method
- Implementation of statistical methods for meta-analysis, including algorithms from referenced papers.