Blood lead exposure, wealth and place in Bangladeshi children

What is included: Five R scripts that build the analysis dataset from the raw SPSS files and produce every result: an installer for the required R packages, a pre-flight check of R/packages/input files (with size and MD5 checksums against the authors' own data release), the main statistical analysis (prevalence, logistic regression, growth-faltering associations with population-attributable fractions), the spatial analysis (Moran's I, LISA, Getis-Ord Gi* hotspots, spatial lag/error regression, kernel-smoothed prevalence maps), and a results-verification script that checks the run's output against both the published manuscript text and the authors' reference results, then reports PASS or CHECK.

A single RUN_ALL.R driver that runs the whole pipeline end to end (roughly 3–5 minutes once the data are in place) and writes a full log. reference_output/: the authors' own tables (plain-numeric and journal-formatted CSVs, plus two consolidated Excel workbooks), 300 dpi TIFF figures with PNG previews, and R session information, so a reviewer's run can be compared cell-by-cell without having to trust a screenshot. A detailed README covering data acquisition, folder layout, package installation, troubleshooting, and methodological notes (survey weighting, missing-data handling, spatial displacement of MICS cluster coordinates, and other choices worth carrying into peer review).

Reproducibility notes: estimation uses the survey package's Taylor-linearised design-based variance estimator with the survey's primary sampling units and strata; Monte Carlo permutation tests are seeded for exact reproducibility; local spatial statistics are FDR-adjusted; and all software versions used for the reference run are recorded.