Ignoring a level in a multilevel model: evidence from UK census data

Because of the inherent multilevel nature of census data, it is often appro priate to use multilevel models to investigate relationships between census variables. For a local population, the data available from the census allo w a three-level nested model to be assumed, with an individual level (level 1), an enumeration district (ED) level (level 2), and a ward level (level 3). The consequences of ignoring one of the three levels in this model are assessed here theoretically. Empirical results, based on 1991 UK Census dat a, are also provided, comparing the variance components estimated from the three-level model with analyses based on models where the ED or ward level are ignored. The results show how the variation that occurs at the level no t included in the models is redistributed to the other levels that the mode ls do include.