AbstractTriangles provide a valuable means of diagrammatically summarizing and understanding data upon triads. In this paper, various properties of the triangular representation of data are discussed. The paper focuses on data relating to parliamentary elections, and demonstrates that an analysis based on the use of transition matrices can be misleading, especially when based upon what is referred to as a‘proportional loss’model. A triangular presentation of voting data suggests an alternative analysis-namely, a model of uniform change. Finally, there is discussion of some of the consequent implications for the marketing strategy of individual political parties.