Assessing the bias due to non-coverage of residential movers in the German microcensus panel

Auteurs

• Ulrich Rendtel

Paramètres

• Affaires et économie

En savoir plus sur le livre

The German Microcensus (MC) is a large-scale rotating panel survey conducted over three years, featuring mandatory participation and approximately 200,000 respondents, making it valuable for longitudinal analysis. However, its area sampling method results in a lack of data on residential mobility, leading to potential biases in longitudinal analyses, particularly regarding transitions between labor market states. This study investigates whether these biases exist and evaluates methods to mitigate them. Using data from the German SocioEconomic Panel (SOEP), which includes residential mobility, we assessed the impact of missing data from residential movers on labor force flow estimations. By comparing results from the complete SOEP sample with those restricted to nonmovers, we found that the omission of residential movers significantly affects the findings. We explored correction methods, including weighting by inverse mobility scores and loglinear models for partially observed contingency tables. Our findings show that weighting by inverse mobility scores reduces bias by about 60%, while official longitudinal weights obtained through calibration achieve an 80% reduction. However, loglinear models for nonignorable nonresponse produced unstable results. This research highlights the importance of addressing noncoverage bias in panel surveys, particularly in labor market analysis.

Achat du livre

• Année de publication: 2007