Regresar

Random-effect energy regression models for clustered sub-city area fuel poverty data

Abstract:

The research presented herein is part of an ongoing project as to provide an evidenced based and geographically targeted energy/carbon route map for the city of Newcastle upon Tyne, UK. Area-based case scenarios are used in evidence-informed energy efficiency and/or fuel poverty policies/practice which attempt to identify appropriate energy/carbon reduction targets in aggregated building stocks. This study demonstrates how multilevel modelling builds on traditional statistical methods for the comparison of groups, where the groups are fuel poverty (FP) dwellings in different Lower Level Super Output Areas (LLSOA) in England. We begin with the standard regression methods for comparing the means of two or more groups, using 'fixed effects' modelling (commonly called analysis of variance ANOVA). Then, we contrast this approach with multilevel or 'random effect' modelling. We used the methods for single-level statistical inference, including Normal tests for comparing means and likelihood ratio tests, which are also used in multilevel mode. We conclude that being in fuel poverty and having an Energy Performance Certificate (EPC) affect the energy consumption of individual dwellings.