An analysis of voting patterns for the Liberal Party with initial reference to the General Election of 1966

The Liberal Party is conventionally regarded as drawing its support relatively evenly, in comparison with the Labour and Conservative Parties, from most social groups. The first section of this study examines the empirical basis for this assumption using data largely derived from public opinion surveys. Although this evidence tends to confirm the impression that Liberal support is drawn more evenly from the various socio-structural groups than is the case with the other parties, alternative bases of cleavage, which might be more relevant to Liberal support than to major party support, are proposed, though these theories cannot generally be tested because data in suitable categories is unavailable.The second section of the study examines the relationships between the most apparently relevant socio-structural, and other, variables and the Liberal vote at an aggregate level using multivariate techniques of analysis. The Liberal vote is less susceptible to accurate prediction using these methods than is the support for the major parties owing to the greater spread of support for Liberal candidates amongst the various social groups. However, up to two thirds of the variation in Liberal support in 1966 at constituency level can be predicted using a relatively economical set of predictor variables. Having derived equations for 1966 which combine relative statistical precision with meaningful variables these models are, then weighted in order to determine the extent to which Liberal performance at elections other than 1966 differ with respect to their predictability. Examination of regression and correlation coefficients between 1959 and 1970 reveals little systematic change,in relationships between predictor variables and Liberal support. Finally, the topological form of the cusp catastrophe is proposed as providing an appropriate means of looking at Liberal performance over time at constituency level given both the regularities and irregularities of behaviour which had previously been delineated.