Applications of multilevel models with country as a unit of analysis

5:00-7:30 pm, 11 December 2012, The Royal Statistical Society 12 Errol Street, London EC1Y 8LX

The Social Consequences of Unemployment in Europe: a Two-Stage Multilevel Analysis: Vanessa Gash (City University) In this paper we examine the relationship between unemployment and social participation and aim to identify the role of national policies and attitudes as possible mediators. We use the 2006 EU-SILC module on social participation – a dataset that provides rich information on social participation for 24 EU countries. We adopt a two-stage multilevel design, allowing us to directly examine the impact of national policies and norms on individual outcome. The paper reveals clear evidence that the negative impact of unemployment on participation levels can be alleviated by macro-level factors. Societies where egalitarian ideals are held high have higher social participation rates amongst their unemployed

Exploiting Space and Time in Multilevel Models: Applications to Climate Change and Religiosity: Malcolm Fairbrother (University of Bristol) Multilevel/mixed models are routinely fitted to longitudinal data (with observations nested within units) and also to units nested within groups at a given point in time (such as respondents to a survey nested within their countries of residence). Increasing numbers of datasets in many fields of research are characterised by both types of clustering simultaneously. I present two useful techniques for analysing data with these characteristics. These techniques allow change over time in y to be a function of change over time in x and/or the time-invariant level of x. Simulation studies show that these techniques are generally robust even to the presence of complications that may arise with real-world data. I then demonstrate the kinds of insights they can provide, using applications to two very different topics: the relationships between climate and economic growth, and between social inequality and religiosity.

Ian Plewis (University of Manchester) will act as a discussant for the papers. The meeting will be preceded by the AGM of the Social Statistics section.