First release of the NMAR package for estimating nonignorable nonresponse (NMAR) bias in survey data.
Methods
Empirical Likelihood (EL): Added el_engine() implementing the estimator of Qin, Leung, and Shao (2002). This method uses empirical likelihood weights satisfying response mechanism equations and auxiliary moment constraints.
Exponential Tilting (Parametric & Nonparametric): Included robust implementations for both microdata (exptilt_engine) and aggregated contingency tables (exptilt_nonparam_engine) based on Riddles, Kim, and Im (2016).
Key Features
Unified API: All estimators are accessible via a single, consistent nmar() interface supporting standard formula syntax (e.g., Y ~ X | Z).
Complex Survey Support: Seamless integration with the survey package. nmar() accepts survey.design objects, automatically handling weights and stratification.
Variance Estimation: Robust bootstrapping (S3) implementation for standard errors and confidence intervals across all engines.
Diagnostics: Rich return objects including convergence statistics, Jacobian condition numbers, and detailed weight summaries.
Major Changes
Refactored Architecture: The exptilt and el engines share a unified structural design, ensuring consistent behavior for controls, standardization, and error handling.
Standardization: Added standardize = TRUE argument to engines to improve numerical stability during optimization.
