Dear Statalist members,
I am pleased to share that metaLong is now available on the SSC archive:
ssc install metalong
metaLong provides a workflow for longitudinal meta-analysis, where studies report outcomes at multiple follow-up time points. The package integrates:
- Random-effects pooling with cluster-robust variance estimation
- Time-varying sensitivity analysis using ITCV
- Benchmark calibration against observed covariates
- Leave-k-out fragility analysis
- Restricted cubic spline trends over time
- Automated publication-ready figures
The commands are designed to handle dependent effect sizes across time and to provide integrated robustness diagnostics within a single pipeline.
A simple example:
sim_longmeta, k(20) times(0 6 12 24) seed(42) clear
ml_meta yi vi, study(study) time(time) saving(meta_res) replace
ml_sens yi vi, study(study) time(time) ///
metafile(meta_res) saving(sens_res) replace
ml_benchmark yi vi, study(study) time(time) ///
metafile(meta_res) sensfile(sens_res) ///
covariates(pub_year quality n) saving(bench_res) replace
ml_fragility yi vi, study(study) time(time) ///
metafile(meta_res) saving(frag_res) replace
ml_spline, metafile(meta_res) df(3) saving(spline_res) replace
metalong_plot, metafile(meta_res) sensfile(sens_res) ///
splinefile(spline_res) fragfile(frag_res) ///
saving(figure.gph) replace
RePEc ID: RePEc:boc:bocode:s459668
Documentation and examples:
https://github.com/causalfragility-lab/metaLong-Stata
I would welcome any feedback or suggestions.
Best regards,
Subir Hait
PhD Candidate, Michigan State University
I am pleased to share that metaLong is now available on the SSC archive:
ssc install metalong
metaLong provides a workflow for longitudinal meta-analysis, where studies report outcomes at multiple follow-up time points. The package integrates:
- Random-effects pooling with cluster-robust variance estimation
- Time-varying sensitivity analysis using ITCV
- Benchmark calibration against observed covariates
- Leave-k-out fragility analysis
- Restricted cubic spline trends over time
- Automated publication-ready figures
The commands are designed to handle dependent effect sizes across time and to provide integrated robustness diagnostics within a single pipeline.
A simple example:
sim_longmeta, k(20) times(0 6 12 24) seed(42) clear
ml_meta yi vi, study(study) time(time) saving(meta_res) replace
ml_sens yi vi, study(study) time(time) ///
metafile(meta_res) saving(sens_res) replace
ml_benchmark yi vi, study(study) time(time) ///
metafile(meta_res) sensfile(sens_res) ///
covariates(pub_year quality n) saving(bench_res) replace
ml_fragility yi vi, study(study) time(time) ///
metafile(meta_res) saving(frag_res) replace
ml_spline, metafile(meta_res) df(3) saving(spline_res) replace
metalong_plot, metafile(meta_res) sensfile(sens_res) ///
splinefile(spline_res) fragfile(frag_res) ///
saving(figure.gph) replace
RePEc ID: RePEc:boc:bocode:s459668
Documentation and examples:
https://github.com/causalfragility-lab/metaLong-Stata
I would welcome any feedback or suggestions.
Best regards,
Subir Hait
PhD Candidate, Michigan State University
