Considerations for multiple imputation. Case of study with panel data

Missing data is a challenge for statistical analysis. Imputation, as the process of replacing missing data with an estimated value, is a regular problem in any research project. There are many imputation models and packages that make this process. Nevertheless, the election of the adequate imputatio...

Descripción completa

Detalles Bibliográficos
Autores principales: Del Callejo Canal, Diana, Canal-Martínez, Margarita Edith, Vernazza, Elena, Urruticoechea, Alar, Álvarez-Vaz, Ramón
Formato: Artículo publishedVersion
Lenguaje:Español
Publicado: CIMBAGE - IADCOM - Facultad de Ciencias Económicas - Universidad de Buenos Aires 2022
Materias:
IMPUTACIÓN
DATOS FALTANTES
SERIES DE TIEMPO
DATOS PANEL
IMPUTACIÓN MÁšLTIPLE.
IMPUTATION
MISSING DATA
TIME SERIES
PANEL DATA
MULTIPLE IMPUTATION
Acceso en línea:https://ojs.economicas.uba.ar/CIMBAGE/article/view/2295
https://repositoriouba.sisbi.uba.ar/gsdl/cgi-bin/library.cgi?a=d&c=cimbage&d=2295_oai
Aporte de:
Repositorio Digital de la Universidad de Buenos Aires (UBA) de Universidad de Buenos Aires
id I28-R145-2295_oai
record_format dspace
spelling I28-R145-2295_oai2025-02-11 Del Callejo Canal, Diana Canal-Martínez, Margarita Edith Vernazza, Elena Urruticoechea, Alar Álvarez-Vaz, Ramón 2022-06-09 Missing data is a challenge for statistical analysis. Imputation, as the process of replacing missing data with an estimated value, is a regular problem in any research project. There are many imputation models and packages that make this process. Nevertheless, the election of the adequate imputation model is transcendental for the results reliability. In this study we work with a Time-Series Cross-Section dataset (TSCS) and 24% of missing data. We used a multiple imputation model and aggregated some prior information to  the system. The principal contribution to this exercise is to show that a good imputation requires (beside the software) a problem diagnosis, the configurations of the model imputation, and finally, the diagnostic of the quality of the data imputation. Los datos faltantes son todo un reto en los análisis estadísticos. La imputación, entendida como el proceso de reemplazar los datos faltantes con un valor estimado, es un problema regular en los proyectos de investigación. Existen muchos modelos y subrutinas de diversos software destinadas para este proceso, sin embargo, la selección del modelo de imputación adecuado al tipo de datos disponibles es trascendental para la fiabilidad del resultado. En este estudio se trabaja con una tabla de datos cruzada que involucran series de tiempo (datos panel) con un 24% de datos faltantes. Con el objetivo de imputar estos datos, se utilizó un modelo de imputación múltiple y se agregaron algunas restricciones al sistema. El principal aporte de este ejercicio es mostrar que un buen proceso de imputación requiere del diagnóstico del problema, de la configuración del modelo de imputación y, finalmente, de la verificación de la calidad de los datos imputados. application/pdf text/html https://ojs.economicas.uba.ar/CIMBAGE/article/view/2295 10.56503/CIMBAGE/Vol.1/Nro.24(2022)p.33-47 spa CIMBAGE - IADCOM - Facultad de Ciencias Económicas - Universidad de Buenos Aires https://ojs.economicas.uba.ar/CIMBAGE/article/view/2295/3070 https://ojs.economicas.uba.ar/CIMBAGE/article/view/2295/3075 Derechos de autor 2022 Cuadernos del CIMBAGE Cuadernos del CIMBAGE; Vol. 1 No. 24 (2022): Cuadernos del CIMBAGE N°24 (Junio 2022); 33-47 Cuadernos del CIMBAGE; Vol. 1 Núm. 24 (2022): Cuadernos del CIMBAGE N°24 (Junio 2022); 33-47 1669-1830 1666-5112 IMPUTACIÓN DATOS FALTANTES SERIES DE TIEMPO DATOS PANEL IMPUTACIÓN MÁšLTIPLE. IMPUTATION MISSING DATA TIME SERIES PANEL DATA MULTIPLE IMPUTATION Considerations for multiple imputation. Case of study with panel data Consideraciones a la imputación múltiple. Un caso de estudio con datos panel info:eu-repo/semantics/article info:eu-repo/semantics/publishedVersion https://repositoriouba.sisbi.uba.ar/gsdl/cgi-bin/library.cgi?a=d&c=cimbage&d=2295_oai
institution Universidad de Buenos Aires
institution_str I-28
repository_str R-145
collection Repositorio Digital de la Universidad de Buenos Aires (UBA)
language Español
orig_language_str_mv spa
topic IMPUTACIÓN
DATOS FALTANTES
SERIES DE TIEMPO
DATOS PANEL
IMPUTACIÓN MÁšLTIPLE.
IMPUTATION
MISSING DATA
TIME SERIES
PANEL DATA
MULTIPLE IMPUTATION
spellingShingle IMPUTACIÓN
DATOS FALTANTES
SERIES DE TIEMPO
DATOS PANEL
IMPUTACIÓN MÁšLTIPLE.
IMPUTATION
MISSING DATA
TIME SERIES
PANEL DATA
MULTIPLE IMPUTATION
Del Callejo Canal, Diana
Canal-Martínez, Margarita Edith
Vernazza, Elena
Urruticoechea, Alar
Álvarez-Vaz, Ramón
Considerations for multiple imputation. Case of study with panel data
topic_facet IMPUTACIÓN
DATOS FALTANTES
SERIES DE TIEMPO
DATOS PANEL
IMPUTACIÓN MÁšLTIPLE.
IMPUTATION
MISSING DATA
TIME SERIES
PANEL DATA
MULTIPLE IMPUTATION
description Missing data is a challenge for statistical analysis. Imputation, as the process of replacing missing data with an estimated value, is a regular problem in any research project. There are many imputation models and packages that make this process. Nevertheless, the election of the adequate imputation model is transcendental for the results reliability. In this study we work with a Time-Series Cross-Section dataset (TSCS) and 24% of missing data. We used a multiple imputation model and aggregated some prior information to  the system. The principal contribution to this exercise is to show that a good imputation requires (beside the software) a problem diagnosis, the configurations of the model imputation, and finally, the diagnostic of the quality of the data imputation.
format Artículo
publishedVersion
author Del Callejo Canal, Diana
Canal-Martínez, Margarita Edith
Vernazza, Elena
Urruticoechea, Alar
Álvarez-Vaz, Ramón
author_facet Del Callejo Canal, Diana
Canal-Martínez, Margarita Edith
Vernazza, Elena
Urruticoechea, Alar
Álvarez-Vaz, Ramón
author_sort Del Callejo Canal, Diana
title Considerations for multiple imputation. Case of study with panel data
title_short Considerations for multiple imputation. Case of study with panel data
title_full Considerations for multiple imputation. Case of study with panel data
title_fullStr Considerations for multiple imputation. Case of study with panel data
title_full_unstemmed Considerations for multiple imputation. Case of study with panel data
title_sort considerations for multiple imputation. case of study with panel data
publisher CIMBAGE - IADCOM - Facultad de Ciencias Económicas - Universidad de Buenos Aires
publishDate 2022
url https://ojs.economicas.uba.ar/CIMBAGE/article/view/2295
https://repositoriouba.sisbi.uba.ar/gsdl/cgi-bin/library.cgi?a=d&c=cimbage&d=2295_oai
work_keys_str_mv AT delcallejocanaldiana considerationsformultipleimputationcaseofstudywithpaneldata
AT canalmartinezmargaritaedith considerationsformultipleimputationcaseofstudywithpaneldata
AT vernazzaelena considerationsformultipleimputationcaseofstudywithpaneldata
AT urruticoecheaalar considerationsformultipleimputationcaseofstudywithpaneldata
AT alvarezvazramon considerationsformultipleimputationcaseofstudywithpaneldata
AT delcallejocanaldiana consideracionesalaimputacionmultipleuncasodeestudiocondatospanel
AT canalmartinezmargaritaedith consideracionesalaimputacionmultipleuncasodeestudiocondatospanel
AT vernazzaelena consideracionesalaimputacionmultipleuncasodeestudiocondatospanel
AT urruticoecheaalar consideracionesalaimputacionmultipleuncasodeestudiocondatospanel
AT alvarezvazramon consideracionesalaimputacionmultipleuncasodeestudiocondatospanel
_version_ 1825551040688160768

Ejemplares similares