| Sumario: | Part of a special issue on communication and the political process in the 1992 U.S. presidential election campaign. A graphical multivariate approach to the analysis and reporting of political poll results when more than two candidates command a significant percentage of the vote is presented. The graphical display of confidence ellipses derived from the multivariate normal distribution is a meaningful, informative, and accurate means of documenting sampling errors in such polls. The application of this methodology to the results of four major national polls reported in the week prior to the presidential election revealed that the advantages of the approach are statistical correctness; a clear visual indication of whether either, or only one, of the candidates could win, given a specific level of confidence based on poll results; and further information, at a glance, on the size of the portions of ellipse contained in each winning area.
|
|---|