Small area estimation of social cohesion indicators
In this paper we develop multilevel time series models for small area estimation of social cohesion indicators based on the Dutch Social Cohesion and Well-being Survey.
This paper develops and compares area-level time series models and unit-level time series models to produce small area estimates for seven social cohesion indicators for different regional classifications. All required output tables are derived from one model. This results in output tables that are numerically consistent by definition. The models include random cross-sectional and time components at these different levels to avoid model predictions becoming too synthetic. It is found that the unit level time series model provides the most flexibility for producing small area estimates for different regional classifications which are not necessarily hierarchical.
