M-estymacja w badaniu małych przedsiębiorstw
M-estimation in a Small Business Survey

Summary/Abstract: In many business surveys, sample sizes are large enough to compensate for the presence of outliers, which have a relatively small impact on estimates. However, at low levels of aggregation, the impact of outliers might be significant. Therefore, in the case of a population such as the population of enterprises, the classical approach should be accompanied by methods that resist the occurrence of outliers. To deal with this problem, several alternative technique of estimation, less sensitive to outliers, have been proposed in the statistics literature. In this paper we look at one of them – M-estimation, and compare its usefulness in the small businesses survey.

• Details
• Contents