Preparación de datos para encuestas de empresas: El manual EDIMBUS

• Autores: Beat Hulliger, Daniel Kilchmann
• Localización: XXX Congreso Nacional de Estadística e Investigación Operativa y de las IV Jornadas de Estadística Pública: actas, 2007, ISBN 978-84-690-7249-3
• Idioma: español
• Texto completo no disponible (Saber más ...)
• Resumen

  • Raw data of business surveys often contains errors and missing values. Their detection and treatment is expensive and long. Many methods and approaches for particular problems exist but few comprehensive manuals are available. The EDIMBUS project, a joint e®ort of the National Statistical Institutes of Italy (ISTAT), Netherlands (CBS) and Switzerland (SFSO) with support from Eu- rostat, developed a Recommended Practices Manual (RPM) which treats the design of the process of editing and imputation, its place in the design of a survey and presents recommended methods. The main objectives of editing and imputation, to inform on the quality of a survey and to prepare the data for analysis, are explained. Advantages and limitations of established methods for the detection and treatment of errors and missing values are discussed and rec- ommendations on their use are given. The methods range from computer aided interactive treatment to automatic procedures and from micro to macro level.

    Testing of the editing and imputation process and monitoring its application by quality indicators as well as the documentation of editing and imputation are important parts of the EDIMBUS-RPM.

    The talk will give an overview over the EDIMBUS manual and its main features. Selective editing, which aims at concentrating e®orts on the most in- °uential errors, and outlier detection and treatment are discussed in more detail.

    The relation between the score functions of selective editing, the in°uence of observations and outliers is explored.