Search

Candidate selection in representative democracies is usually either by party elites or by mass primaries, an American innovation lately spreading o the rest of the democratic world. The choice between the two poses a dilemma for reform. Primaries allow mass participation, but by voters who are rarely well informed and or deliberative. Elites are generally well informed and may even deliberative, but decision making confined to them violates political equality.How then to provide for both deliberation and equality? A solution has lain in the dust of history for centuries—the deliberative microcosm chosen by lot. An ancient version was common for many political processes in Athens. A modern adaptation is embodied in Deliberative Polling.This paper reports on the first use of Deliberative Polling for the official selection of candidates for office by a major political party anywhere in the world. In June 2006, Deliberative Polling was used to select a mayoral candidate in Marousi, one of the principal municipalities of the Athens area. PASOK, the Greek Socialist party, used a recruited a random sample hundreds of Marousi citizens to discuss the policy issues facing the city and the choice of a mayoral candidate to deal with them. At the end of the process, two rounds of secret balloting selected one of six candidates as PASOK's official candidate for the mayorship of Marousi.This paper analyzes the changes in both policy attitude changes and candidate preferences. We consider the sample's knowledge gains, the extent to which they drove the changes in candidate preference, and the extent to which the final secret ballot voting can be generated by the questionnaire results. The paper also considers the viability of Deliberative Polling for candidate selection based on this first effort. ..PAT.-Unpublished Manuscript [ABSTRACT FROM AUTHOR]

This paper examines the recommendation that knowledge questions be phrased, framed, and administered so as to minimize the frequency of "don't know" (DK) responses. We experimentally manipulate knowledge items put to a national random sample. [ABSTRACT FROM AUTHOR]

Highlights the results of an online poll on the deliberation of participants on politicians running in the U.S. primaries. Experimental design of the online intervention; Reasons for studying online deliberation; Population of the participants.

It is by now widely if still not quite universally accepted that political sophistication has important effects on many of the ways in which ordinary citizens relate to politics, and the variable has become a regular player in analyses of policy preferences, vote choice, and participation, among other things. There is also a still larger consensus that the best way to measure sophistication is through knowledge, using either straightforwardly factual items or placements of parties or political figures on policy or ideological scales. Willy-nilly, every study deploying such a measure has made a variety of nuts-and-bolts decisions in constructing it. Yet, despite some recent attention to the treatment of don’t-know (DK) responses, the issues behind these decisions remain surprisingly little examined. What exactly makes for the best knowledge based measurement? This paper considers the merits of different types of knowledge items and different ways of doing them up. Are placement- or fact-based items superior? What kinds of placements are best, what kinds of facts? What about the choice of absolute versus relative placements? Viz., are the Republican and Democratic placements on a liberal-conservative scale two distinct items, requiring that the Democrats be placed left of center and the Republicans right of center, or do they boil down to a single item, requiring only that the Democrats be placed to the left of the Republicans? How to treat placements at the midpoint (for absolute measures) or at the same point (for relative ones)? Are they correct or incorrect? Does it help to gauge degrees of correctness, rather than simply scoring items as right or wrong? We give special attention to the treatment of DKs and, relatedly, incorrect answers. Should the former be randomly assigned to response categories, as Mondak suggests? Should the latter be subject to a correction for guessing? We consider these questions by examining the correlations with six criterion variables; a mix of alternative measures, causes, and consequences; and the results from a regression model of the sort popularized by Bartels, where sophistication is the key conditioning variable. [ABSTRACT FROM AUTHOR]