Replication package for "Mortality-based estimation of early pandemic infections benchmarked against seroprevalence"

This Zenodo deposit contains the complete analytical dataset, Stata source code, raw upstream inputs, and generated figures and tables needed to fully reproduce all results in: Villalobos, C. (2026). Mortality-based estimation of early pandemic infections benchmarked against seroprevalence (manuscript, 2026, under review). The paper benchmarks a parsimonious mortality-based estimator of cumulative SARS-CoV-2 infections, V = D(t+23)/IFR, against 276 seroprevalence surveys from 42 countries conducted before February 2021, comparing its performance with confirmed case counts using Lin's concordance correlation, Wilcoxon signed-rank tests, and Bland-Altman analysis. Principal results: the mortality-based estimator approximated seroprevalence-implied infections more closely than confirmed cases in 40 of 42 countries and 92.0% of observations. Lin's CCC on log10-transformed counts was 0.71 for V versus 0.37 for C. Robustness is confirmed through random-intercept mixed models (ICC = 0.70), cluster-bootstrap confidence intervals, permutation tests, and seven alternative IFR specifications. Contents: code/ (Stata .do files with master_zenodo.do orchestrator), data/ (analytical dataset, 276 observations, plus intermediate files), data_raw/ (SeroTracker export and UN World Population Prospects), figures/ (PDF and PNG), tables/ (CSV), and the manuscript and supplementary appendix with LaTeX sources. Licensing: code under MIT; data and figures under CC-BY-4.0. Upstream inputs retain their own terms (Our World in Data CC-BY-4.0; UN WPP CC BY 3.0 IGO; the SeroTracker snapshot, dated 14 March 2026, is included with the citation requested by the providers: Arora et al. 2021, Lancet Infect Dis, doi:10.1016/S1473-3099(20)30631-9).