Dienstag, 18. März 2014

Multilevel approaches to panel data – an application to human rights violations in established democracies

CO:STA presentation by Dag Tanneberg (WZB)

Panel data - i.e. the observation of multiple variables over multiple time intervals for the same units of analysis - often suffer from various violations of the Gauss-Markov assumptions. For example, in many settings it can hardly be assumed that repeated measurements within units are independent of each other. In contrast to seminal pooled time-series models, multilevel approaches can directly model such dependencies by allowing for varying intercepts and slopes. This enables us to disentangle temporal from cross-sectional variation and to study their cross-level interactions.
Dag Tanneberg exploits such models to study violations of physical integrity in established democracies, distinguishing variation due to political threats over time from cross-sectional variation due to the institutional context of individual democracies.

Resources:

Tanneberg, Dag (2014) 'Caught Red-handed: Human rights violations in established democracies', Unpublished manuscript.

Hox, Joop J. (2010): Multilevel Analysis. Techniques and Applications, 2nd. Ed., Routledge: Hove, pp. 79-11

H. Aguinis, et al. (2013). `Best-Practice Recommendations for Estimating Cross-Level Interaction Effects Using Multilevel Modeling'. Journal of Management 39:1490-1528

Bauer, Daniel J. and Patrick J. Curan (2005) 'Probing Interactions in Fixed and Multilevel Regression: Inferential and Graphical Techniques', Multivariate Behavioral Research 40(3): 373-400