Robust Estimation and Inference for Categorical Data

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autor principal: Welz, Max
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866913608731983872
author Welz, Max
author_facet Welz, Max
contents While there is a rich literature on robust methodologies for contamination in continuously distributed data, contamination in categorical data is largely overlooked. This is regrettable because many datasets are categorical and oftentimes suffer from contamination. Examples include inattentive responding and bot responses in questionnaires or zero-inflated count data. We propose a novel class of contamination-robust estimators of models for categorical data, coined $C$-estimators (``$C$'' for categorical). We show that the countable and possibly finite sample space of categorical data results in non-standard theoretical properties. Notably, in contrast to classic robustness theory, $C$-estimators can be simultaneously robust \textit{and} fully efficient at the postulated model. In addition, a certain particularly robust specification fails to be asymptotically Gaussian at the postulated model, but is asymptotically Gaussian in the presence of contamination. We furthermore propose a diagnostic test to identify categorical outliers and demonstrate the enhanced robustness of $C$-estimators in a simulation study.
format Preprint
id arxiv_https___arxiv_org_abs_2403_11954
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Robust Estimation and Inference for Categorical Data
Welz, Max
Methodology
Econometrics
Statistics Theory
While there is a rich literature on robust methodologies for contamination in continuously distributed data, contamination in categorical data is largely overlooked. This is regrettable because many datasets are categorical and oftentimes suffer from contamination. Examples include inattentive responding and bot responses in questionnaires or zero-inflated count data. We propose a novel class of contamination-robust estimators of models for categorical data, coined $C$-estimators (``$C$'' for categorical). We show that the countable and possibly finite sample space of categorical data results in non-standard theoretical properties. Notably, in contrast to classic robustness theory, $C$-estimators can be simultaneously robust \textit{and} fully efficient at the postulated model. In addition, a certain particularly robust specification fails to be asymptotically Gaussian at the postulated model, but is asymptotically Gaussian in the presence of contamination. We furthermore propose a diagnostic test to identify categorical outliers and demonstrate the enhanced robustness of $C$-estimators in a simulation study.
title Robust Estimation and Inference for Categorical Data
topic Methodology
Econometrics
Statistics Theory
url https://arxiv.org/abs/2403.11954