Updated Analysis of Exit Poll Controversies Released by SSRC

[dzika]

(sorry if this has already been posted)

The National Research Commission on Elections and Voting brings scholarly research, knowledge, and perspective to bear on efforts to improve the integrity of the American electoral process. Organized by the Social Science Research Council, the Commission website serves as a national, nonpartisan clearinghouse of data, independent research, and other resources to support the efforts of scholars and organizations studying voting and election issues. In providing this public service, the SSRC intends to help deepen both public and scholarly awareness of electoral process concerns that must be addressed in order to ensure that our elections and voting system remains legitimate and fair.

---

Updated Analysis of Exit Poll Controversies Released

Three Commission members have completed a working paper (pdf) providing further analysis of controversies surrounding exit polls during the 2004 Presidential Election.

From the PDF report:

A Review of Recent Controversies Concerning the 2004 Presidential Election Exit Polls

10 March 2005

Michael Traugott, University of Michigan
Benjamin Highton, University of California (Davis)
Henry E. Brady, University of California (Berkeley)

THE NATIONAL RESEARCH COMMISSION ON ELECTIONS AND VOTING
A Project of the Social Science Research Council
810 Seventh Avenue New York, NY 10019 USA
http://elections.ssrc.org

I. Introduction

One feature of the end of the 2004 presidential campaign was the leaking of preliminary exit poll data from the National Election Pool (NEP) on several websites during the afternoon of November 2.1 The leaks were not completely unexpected because a similar incident occurred during the 2000 presidential campaign. Several websites announced in advance their intention to locate and release early data, and the activity and visibility of several websites and blogs had increased significantly during this campaign. The difference in 2004 was that the early exit poll data were incorrect in that they suggested that John Kerry was ahead in the national poll and leading in several key states that would have been sufficient to give him an Electoral College majority. Furthermore, even the final data had a Kerry bias in their estimate of the outcome.2

What was also unusual about the leaks was the fact that the campaigns took the information seriously. There was a brief period during the afternoon on Election Day when the Kerry campaign went into seclusion to think about an acceptance speech and an even briefer period when the Bush campaign team contemplated an unexpected loss.3 And the stock market took a brief plunge at the close of trading on November 2, apparently based on this information as well.4 When the Republicans quickly turned to their on-the-ground intelligence in the key states of Florida, Ohio, and Virginia and it gave them a different picture of the likely outcome there, their confidence in victory returned.

By the end of the evening, when the full set of NEP data and the output of their statistical models were available, none of the network partners had made any incorrect calls of individual states, and it was clear at a reasonable hour that President Bush had been re-elected.5 However, many bloggers and websites were concerned that the early exit poll results did in fact accurately represent the voters' preferences, and the differences in actual vote outcomes from the exit poll results may have stemmed from fraud, problems with new voting technology, or administrative malfeasance in specific locations.

As a result, there was sufficient public discussion and consternation that a committee of the U.S. House headed by Rep. John Conyers (D-Michigan) held open sessions about election administration and the exit polls to which several of the principles were invited for presentations. NEP and its partners remained relatively silent through all of this, and it was not until January 19, 2005 that a report was released by Edison Research/Mitofsky International about what happened with the exit polls on Election Day.6 That report has now become a topic of extended discussion on the Web.

Link to the report in PDF format:
http://elections.ssrc.org/research/ExitPollReport031005.pdf


COMMENT: My initial impression of this report is that they agree with Mitofsky and NEP on the "reluctant responder" theory.