Using an entropy-maximizing procedure to estimate territorial social indicators: An introduction and illustration

The creation of quantitative territorial social indicators for studies of spatial variations in human behavior is often hampered by the lack of spatially disaggregable data, with the consequence that the absence of any such variations much be assumed. An entropy-maximizing procedure has been developed which produces maximum likelihood estimates of spatial variations within known, aggregate, constraints. This procedure is introduced and illustrated using the example of tactical voting in Great Britain.