Statistics Experts, CBS Director of Surveys to Discuss Accuracy of Election Polls at Joint Statistical Meetings in Denver

Share Article

Just a few weeks before the Democratic National Convention and in the same venue, the 168th annual Joint Statistical Meetings (JSM) will host several sessions focusing on topics related to electoral process, the American Statistical Association (ASA) announced today. Kathleen Frankovic, Director of Surveys and producer for CBS News, will participate in a discussion of election poll accuracy in one of the conference's two late-breaking sessions. Some 5,500 statisticians will convene at the Colorado Convention Center August 3--7 to participate in the world's largest gathering of statisticians.

Just a few weeks before the Democratic National Convention and in the same venue, the 168th annual Joint Statistical Meetings (JSM) will host several sessions focusing on topics related to electoral process, the American Statistical Association (ASA) announced today. Kathleen Frankovic, Director of Surveys and producer for CBS News, will participate in a discussion of election poll accuracy in one of the conference's two late-breaking sessions. Some 5,500 statisticians will convene at the Colorado Convention Center August 3-7, 2008 to participate in the world's largest gathering of statisticians.

Frankovic will present a present a paper illustrating examples of problems in accurately measuring preference in election polls and addressing whether journalists give polls more precision than they deserve. With major responsibility for the design, analysis and broadcasting of results from CBS News and CBS News/New York Times polls, Frankovic speaks and writes extensively about elections and public opinion, as well as the development and use of polling by newspapers and television. She is president of the World Association for Public Opinion Research.

The Accuracy of Election Polls (Activity #164), Monday, August 4, 2 pm
The late-breaking session includes the following presentations. The discussant for this session is Clyde Tucker of the Bureau of Labor Statistics, who consults with CNN on calling elections.

  •     Understanding and Communicating Sources of Measurement and Operational Error in Opinion Polls; Beyond Sampling Error (Presenter: Dr. Kathleen Frankovic, CBS News): This paper will present examples of problems in accurately measuring preference in election polls, focusing on poll timing, measuring intensity, and the interaction between respondent and interviewer. Have interviewer characteristics like race, gender and age affected poll accuracy? When they report polls, are journalists giving polls more precision than they deserve?
  •     Evaluating the Performance of the 2008 Pre-Election Polls in the Primaries (Presenter: Michael W. Traugott, University of Michigan): The performance of the polls in election years is a marker for the quality of work in the profession generally. The accuracy of published polls in general elections has been very good, but this year there were some estimation problems in the early primary and caucus polls. This paper will discuss range of sources of such error and whether they might present problems for the fall election.
  •     Sources of Variation in Pre-Election Polling (Presenter: Cliff Zukin, Rutgers University): This paper presents a variety of answers to the question of "Why Do Election Polls Differ?" It examines issues of timing, sampling, questionnaire wording, weighting and the identification of "likely voters" among others.

Several other sessions that will focus on electoral issues are listed below. Complete information can be obtained from the online JSM program, searching by activity number or keyword, at http://www.amstat.org/meetings/jsm/2008/onlineprogram/index.cfm?fuseaction=main.

Accurate Elections: The Role of Statistics (Activity #217), Tuesday, August 5, 8:30 am

  •     Designing an Audit System to Increase Voter Confidence in Elections (Presenter: Michael W. Traugott, University of Michigan): This paper will report on preliminary results from a research project to develop empirical information to assist Michigan in the development of an election audit system. One purpose of an audit is to ensure that the electoral system is functioning in a way that provides accurate counts. Another purpose is to increase and maintain public confidence in the electoral system, suggesting that the audit should be regular and ongoing rather than episodic.
  •     Engaging the Unengaged Voter: Vote Centers and Voter Turnout

(Presenters: Robert M. Stein and Greg Vonnahme, Rice University): Previous election reforms designed to increase turnout have often made voting more convenient for frequent voters without significantly increasing turnout among infrequent voters. Election Day vote centers, which are non-precinct based locations for voting, provide an alternative means of motivating electoral participation among infrequent voters.

  •     Measuring Voting System Failures: Survey Evidence of the Frequency of Voting Problems in the 2006 and 2008 U.S. Election

(Presenters: Stephen Ansolabehere and Charles H. Stewart III, MIT): This session presents the results of two nationwide surveys in the United States, one from the 2006 general election and one from Super Tuesday 2008, which involved large national samples to allow measurement of what are low-probability events - problems relating to voter registrations, voting equipment difficulties, lines, polling place operations, and poll workers, as well as measures of the reasons for nonvoting.

  • Predicting Electoral Vote Totals for the Presidential Election (Activity #312; Abstracts 301839 and 302446), Tuesday, August 5, 2 pm (Presenters: Christopher J. Rigdon, Steven Rigdon, Edward Sewell, Sheldon Jacobsen, Arizona State University, Southern Illinois University Edwardsville, and University of Illinois at Urbana, Champaign; Steven Rigdon, Michael Siekerka (Southern Illinois), Christopher J. Rigdon): During the 2008 presidential race, hundreds of polls will be conducted in each state. The objective of this research is to create a model using polling data to determine the number of electoral votes each candidate will receive by considering the impact of undecided voters. In the 2004 election, the model missed the correct number of electoral votes for each candidate by only two.
  • Statistical Measures Can Help Restore Confidence in U.S. Elections (Abstract #460), Thursday, August 7, 8:30 am
  • Voter Confidence and the Election Day Voting Experience (Presenters: J. Quin Monson, Kelly Patterson, and David Magleby, Brigham Young University; Ryan Claassen, Kent State University): Based on findings of 2006 exit poll data collected by the Center for the Study of Elections and Democracy at Brigham Young University, the findings reveal good reason for boards of elections to improve service at polling places. The quality of service at the polling place affects voters' confidence that ballots will be counted accurately, as do wait times, sense of privacy and voter partisanship.
  •     How ASA Members Are Helping States Improve Elections (Presenter: Arlene Ash, Boston University): ASA members have been actively working with voting rights activists, computer scientists, state legislators, county supervisors of elections and secretaries of state to ensure the use of credible audit procedures for elections. Issues range from highly pragmatic (e.g., how to randomly select precincts in a way that is credible to public observers, how to efficiently and accurately tally votes from multiple races) to more theoretical (e.g., initial sampling designs, efficient sequential procedures for when findings from the initial sample are equivocal, post-hoc power calculations).
  •     National Election Scorecard (Presenters: Fritz Scheuren, The University of Chicago; Edward Mulrow, National Opinion Research Center): In recent election years, we have been inundated with polling results, the focus of which has been on the potential winning candidate and who will or did vote for him or her. Seldom have there been surveys that asked voters about the voting experience, itself. This paper discusses the first national attempt at such a poll. Issues of design are covered, with sampling and especially non-sampling issues given prominence. The focus will be on what might be called a national customer satisfaction survey. (Key here is completing interviews before winners are announced.)
  • Election Statistics Results (Activity #162, Roundtable with Lunch), Monday, August 4, 12:30 pm (Presenter: Wendy Rotz, Ernst & Young LLP): Voter confidence and transparent elections are essential to modern democracies. States are legislating precinct-level, sample-based audits. A statistical foundation in the development of election auditing procedures aids transparency and improves voter confidence in election results and the democratic process in both the United States and abroad. Statistical issues may include random recounts in all elections, 100% recounts in close elections, paper trails, Six Sigma applications, process control, exit polling, and more.
  • Issues Matter: Social Attributes, Predispositions, and Issues Affect Voting Choice (Activity 24), Sunday, August 3, 2pm (Presenter: Robert B. Smith, Cytel Inc.): Clarifying the relative impacts of economic and social issues, this study analyzes a 1992 election night survey assessing how social attributes, philosophical self-designation (liberal, centrist, conservative), party identification (Democrat, Independent, Republican), and the issues influenced the voters' choices.

About JSM:
JSM, the world's largest gathering of statisticians, is held jointly with the American Statistical Association (ASA), the International Biometric Society (ENAR and WNAR), the Institute of Mathematical Statistics (IMS), and the Statistical Society of Canada. The theme for this year's conference is Communicating Statistics: Speaking Out and Reaching Out.

About the American Statistical Association:
The American Statistical Association (ASA), a scientific and educational society founded in Boston in 1839, is the second oldest continuously operating professional society in the United States. For more than 160 years, ASA has been providing its 18,000 members serving in academia, government, and industry and the public with up-to-date, useful information about statistics. The ASA has a proud tradition of service to statisticians, quantitative scientists, and users of statistics across a wealth of academic areas and applications. For additional information about the American Statistical Association, please visit the association's web site at http://www.amstat.org or call 703.684.1221.

Related JSM sessions will address voter confidence, predicting electoral votes, ASA involvement with voting rights activists, residual voting, and Election Day vote centers.

Note to editors: ASA will be pleased to try to set up interviews with any of the presenters; contact rosanne @ amstat.org Media can register to attend any of conference at http://www.amstat.org/meetings/jsm/2008/index.cfm?fuseaction=pressregistration

###

Share article on social media or email:

View article via:

Pdf Print

Contact Author

Visit website