The National Annenberg Election Survey 2004: The Largest Academic Survey Conducted on American Political Attitudes & Behavior

by Kate Kenski

During the 2004 presidential campaign, the Annenberg Public Policy Center (APPC) under the direction of Kathleen Hall Jamieson conducted the National Annenberg Election Survey (NAES), the largest academic survey on American attitudes toward the presidential candidates. This is the second time that the NAES has been conducted. The first NAES in 2000 contained 100,626 interviews with 79,458 adults living in the United States. Exact counts for the 2004 NAES are not yet available as the survey is still in the field, but the overall number of interviews will be comparable to 2000.

The 2004 NAES contains three types of survey design: (1) the national rolling cross-section or RCS, (2) panel studies around key events, such as the major party conventions, the presidential debates, and the general election, and (3) independent cross-sections, including a study of military families. The 2004 NAES differs from 2000 in that panel studies were not conducted during the primary season and the RCS sample size was increased by 23,049 cases.

In this article, I outline the ways in which the NAES is unique and highlight a few findings from the survey. The NAES is distinctive for four reasons. First, the survey was in the field thirteen months prior to the general election. Second, the survey contains a large RCS component. Third, the content of the survey includes a comprehensive set of news media exposure items, political knowledge items about the presidential candidates, and an absentee/early voting battery. Fourth, the survey sample size is large, which permits analysis of proportionally small subgroups.

Time Period

The NAES 2004 RCS survey began October 7, 2003 and ended November 16, 2004. During this period, 81,422 adults in the United States were interviewed for the RCS. Households throughout the national were randomly selected using random-digit dialing. With the exception of eight holidays, interviews took place daily. Just over 200 interviews were conducted each day on average. Sample sizes did change from month to month, with larger sample releases occurring before key events. Over 60 individuals were interviewed per day on average during October 2003, while over 360 individuals were interviewed per day during October 2004.

Because the survey was in the field over a year before the general election, the emergence of Senator John Kerry as the Democratic nominee can be tracked. Given that campaigns are taking place earlier than in years past, it is important for researchers to be in the field as public opinion begins to take shape. Of particular importance to political communication scholars is tracking the impact of campaign advertisements on public opinion toward the candidates. While many Americans contend that they do not trust television commercials, NAES data show that a great many people absorb and believe what they see, no matter how tangential its relation to the truth. From April 15 through May 2, 1,026 adults were interviewed in 18 battleground states where the George W. Bush and John Kerry campaigns had been showing commercials since March. In those states, 61 percent of the public believed Bush “favors sending American jobs overseas,” 56 percent believed Kerry “voted for higher taxes 350 times,” and 72 percent said 3 million jobs have been lost in Bush’s presidency. The margin of sampling error was plus or minus three percentage points (see NAES press release May 12).

The RCS Design

The RCS design distinguishes the NAES from other surveys. An RCS is a series of repeated cross-sections collected with a rigorous sampling plan. This sampling plan works to ensure that each of the repeated cross-sections is composed of randomly selected members from the population under study. Researchers thus can treat the date of interview as a chance event.  Because the date of interview can be treated as a chance event, researchers can analyze the data as a single cross-section or as a time series. With this design, communication scholars can track the dynamic effects of campaign events across the presidential campaign. The 2004 NAES uses the same RCS procedures created and refined by political scientist Richard Johnston in 2000 when the NAES was first fully implemented.

In his November 19 press release on party identification, NAES political director Adam Clymer utilized the survey’s RCS design to show the ebbs and flows of party allegiance throughout the year. The variation in party identification suggests that party allegiance is not a demographic constant. Among registered citizens, Republican party identification was lowest in late June but rebounded by November as shown in Graph 1. This is just one of many important political variables that can be tracked across time with the NAES.

Survey Content

In addition to candidate favorability ratings, trait ratings, vote intention, issue positions, party identification, ideology, and a detailed set of demographic items, the survey contains comprehensive sets of news media exposure and attention items, candidate knowledge items, an absentee/early voting battery, and much more.

The survey’s panel studies were able to track the electorate’s learning about the presidential candidates in response to campaign events. In an August 29 press release, Kathleen Hall Jamieson said, “Despite signals from pundits that conventions are staged substanceless, media events and the broadcast networks’ assumption that they are worthy of minimal prime time, party conventions remain an important source of issue information for those who watch as well as those whose interest in them is expressed in a heightened attention to news.” The Democratic National Convention increased the public’s knowledge of the positions Kerry and Bush held on a range of issues. A sample of 847 adults was interviewed before the convention and again after it, and their answers changed on many questions. Before the convention 33 percent of the panel correctly identified the income group ($200,000 or more) whose taxes Kerry wanted to increase; afterwards 50 percent could. That was the biggest change in awareness on an issue. Another message Kerry was trying to get out – that he was a former prosecutor – also showed a sharp increase, from 48 percent knowing before the convention to 63 percent after it was over.

A unique content feature of the NAES is an absentee and early voting battery that I designed in 1998 (the pilot year of the NAES RCS design). In many states, citizens are given the opportunity to vote prior to Election Day by either balloting by mail or voting early at polling stations. During the final weeks of the campaign, the NAES asks respondents whether or not they have already cast their ballots rather than simply giving them a vote intention question. Since many people have the opportunity to cast ballots before Election Day, it is important to ask questions that distinguish between vote intention and behavior. The absentee and early voting battery was implemented in 2000 and again in 2004. The percentage of voters who cast their ballots before Election Day thus can be tracked as Election Day approaches. In my October 29 press release on early voting, I reported that in the six to ten day period before Election Day, 14 percent of registered voters told our interviewers that they had already voted, 3 percentage points higher than during the comparable time period in 2000. The rise in absentee and early voting rates is consequential for campaign analysts as the campaigns must work to target voters earlier than before and citizens who have already cast their ballots may interact with campaign messages differently than those who have yet to cast their ballots.

Large Sample Size

For researchers less interested in the dynamic effects of campaign, the NAES is nevertheless useful because it possesses an incredibly large sample size coupled with a detailed demographic battery, therefore providing the potential to study small subgroups.

Analysis conducted by NAES team member Dannagal Goldthwaite Young demonstrates the power of the survey to detect differences between small subgroups. In her September 21 press release on viewers of late-night comedy programs, Young analyzed data collected between July 15 and September 19 from 19,013 individuals to compare people who watched David Letterman, Jay Leno, and Jon Stewart. The large sample size allowed Young to find that “People who watch The Daily Show are more interested in the presidential campaign, more educated, younger, and more liberal than the average American or than Leno or Letterman viewers,” a finding later reported by Jon Stewart on his show. Because the Leno, Letterman, and Stewart audiences make up a relatively small percentage of the overall population, large samples are needed to capture these subgroups.

Final Remarks

Over eighty NAES reports were released by the APPC between September 2003 and November 2004; these reports can be found at the survey’s website. Those press releases and the examples presented in this article barely scratch the surface of what is possible to learn about the 2004 presidential campaign from these data. The NAES team continues to analyze the data to gain insight into the campaign process. Data are still the process of being collected. As with the 2000 NAES (full 2000 dataset available with the book Capturing Campaign Dynamics: The National Annenberg Election Survey, 2004, Oxford University Press), the APPC will release the 2004 NAES at some point in the future.

Kathleen Hall Jamieson, Director of the Annenberg Public Policy Center, leads the NAES 2004 team. Ensuring that campaign events were appropriately reflected in the survey, Adam Clymer, former chief Washington Correspondent of the New York Times joined the Washington office of the Annenberg Public Policy Center of the University of Pennsylvania as Political Director. Based at the APPC’s offices in Philadelphia, Ken Winneg, former Vice President for Penn, Schoen and Berland, serves as the Managing Director of the Survey. Schulman, Ronca, Bucuvalas, Inc conducts the interviewing.

Kate Kenski is Senior Analyst at the Annenberg Public Policy Center.

Return to the Roundtable