POSTELECTION SURVEYS CONSISTENTLY OVERESTIMATE VOTER TURNOUT BY A SUBSTANTIAL MARGIN. THIS PAPER COMPARES, VIA DISCRIMINANT ANALYSIS OF DATA FROM THE 1978 NATIONAL ELECTION STUDY, THE CHARACTERISTICS OF RESPONDENTS WHO FALSELY CLAIM TO HAVE VOTED WITH THE CHARACTERISTICS OF ACTUAL VOTERS AND ADMITTED NONVOTERS.
"In March 1960, businessmen and social scientists met to examine research on public voting behavior which has implications for political activity by businessmen. This ia a summary report of their meeting [sponsored by the Foundation for Research on Human Behavior and written by Donald E. Stokes]" ; Bibliography: p. [29] ; Mode of access: Internet.
The volume and variety of published works on voting in Canada suggest to us it is time to take stock of where we have been and where we should go. It is our belief that developments have not matched the rosy prognosis implicit in Donald Smiley's 1967 statement that "undoubtedly the most extensive developments in Canadian political science in the past two decades have been in the field of studies of voting behaviour and political parties." To be sure, the body of literature has expanded considerably, but we are struck by major gaps in its coverage and by certain conceptual and technical shortcomings.Traditionally review articles have treated a limited number of works in detail or have attempted an overview of an entire body of literature. We have eschewed the former approach since it presupposes the existence of a few outstanding works, comprehensive in nature, which can be fruitfully compared because of the diversity of approaches or interpretations they present. Such a situation does not yet obtain in the field of Canadian voting behaviour. The latter approach is less necessary since the recent appearance of a propositional inventory by John Terry and Richard Schultz, and a long summary chapter by Mildred Schwartz.Therefore, we propose to outline major trends in the last decade or so and to suggest new directions.
I am very grateful to Dean Burnham for his labor in checking out a number of "plausible rival hypotheses" that I proposed some years ago might account for the bundle of changes in American voting statistics he had first described in his seminal 1965 paper. I had pointed out these possibilities with some vigor, since if they could be shown to have empirical merit, they would seem to put a rather different light on those historical changes, relative to the interpretations Burnham originally provided.I now read the current paper as indicating that for the areas covered, changes in registration practices do appear to have affected turnout adversely, following one of our original rival hypotheses. It also suggests, however, that this registration change does not begin to account for all of the decline in turnout that occurred around 1900, particularly in states like Pennsylvania. It goes on to imply that some other familiar factors affecting turnout, such as party competition, also contributed to that change. In effect, the turnout decline not accounted for by structural change seems greatest where the realignment of 1896 moved states from competitive to rather noncompetitive status (the more frequent case); but states moved by the same events from noncompetitive into competitive status, like Missouri and, in fact, numerous other upper midwestern states, do not show the same pattern as Pennsylvania.All this is well and good, particularly as Burnham in his penultimate paragraph sets aside the 1965 conspiratorial interpretation of the change that had originally prompted our own brief inquiries as to probable causes.
Philip E. Converse has challenged the findings of a 1965 article, "The Changing Shape of the American Political Universe," and other work by Burnham. Converse asserts that most of the very high voter participation which occurred before 1900 can be explained by a combination of electoral corruption, the absence of personal-registration requirements and other "undramatic" factors, and thus that the anomalies which Burnham reported are largely spurious. Issues of major importance for social-science explanation are joined. The present article attempts to demonstrate that intervening structural variables cannot come close to explaining all the post-1900 decline in voting participation and that the genuine existence of universal nineteenth-century rural corruption has yet to be demonstrated. These efforts to explain away anomaly are held to be unpersuasive. The weight of evidence supports the objective reality of the phenomena originally reported. This in turn means that more adequate conceptualizations are needed to integrate empirical findings than those which have hitherto dominated the voting-behavior research community.
A simple prospective model of voting & party identification is proposed that subsumes much of the social-psychological & retrospective voting literatures. Its eleven major propositions are validated here, & its implications for conventional statistical specifications in voting research are assessed, including: (1) linear models would be discarded in favor of a particular nonlinear specifications, & (2) demographics should be dropped from the list of independent variables. Adapted from the source document.
Several recent studies of voter choice in multiparty elections point to the advantages of multinomial probit (MNP) relative to multinomial/conditional logit (MNL). We compare the MNP & MNL models & argue that the simpler logit is often preferable to the more complex probit for the study of voter choice in multiparty elections. Our argument rests on three areas of comparison between MNP & MNL. First, within the limits of typical data -- a small sample of revealed voter choices among a few candidates or parties -- neither model will clearly appear to have generated the observed data. Second, MNP is susceptible to a number of estimation problems, the most serious of which is that the MNP is often weakly identified in application. Weak identification is difficult to diagnose & may lead to plausible, yet arbitrary or misleading inferences. Finally, the logit model is criticized because it imposes the independence of irrelevant alternatives (IIA) property on voter choice. For most applications, the IIA property is neither relevant nor particularly restrictive. We illustrate our arguments using data from recent US & French presidential elections. 2 Tables, 34 References. Adapted from the source document.