A Note on the Estimation of Voter Transition Probabilities

SUMMARY This paper utilizes the results of a panel survey, conducted after the General Elections of February and October 1974, to obtain the estimated voter transition probabilities for constituencies contested on each occasion by candidates representing the Conservative, Labour and Liberal parties. The panel estimates are compared with those derived by McCarthy and Ryan (1977) who used a quadratic programming method. Although the panel results underestimate the proportion of the electorate who did not vote, they provide clear evidence that the quadratic programming oversimplifies the pattern of voting change. The panel results also show clearly that the pattern of voting change between two elections is affected by the relative party strengths in the first election.