Abstract
Entropy-maximising methods have been introduced in recent papers for estimating spatial variations in voting behaviour. The estimates can be used to test structural effect hypotheses. This paper illustrates such usage and discusses a major technical issue relating to it.