A missing level in the analysis of British voting behaviour: the household as context as shown by analyses of a 1992-1997 longitudinal survey – -not to be quoted without permission of authors-

Abstract

One potential contextual effect which is missing from virtually all survey-based analyses of British voting behaviour is the individual household: indeed, the surveys most commonly deployed - those of the British Election Study - have no data which allow this effect to be explored. Using BHPS data, this paper presents an initial exploration of the degree of within-household consensus of voting at the 1992 and 1997 British general elections. Initial analyses using the multiplication rule of
probabilities indicate much greater within-household agreement that anticipated from a model assuming independence of individuals, as well as much greater within-household agreement on changes in voting between the two elections. The importance of this within-household consensus, holding constant individual voter characteristics, is confirmed by a multi-level model regression of the 1992 data. The findings raise important issues not only for further research but also for the
design of voter surveys