NES logo


Memo
To: Members of the ANES Research Community
From: Nancy Burns and Donald Kinder
Re: ANES Announcement: MPSA Public Meeting and NES2002 Information
Date: March 25, 2003

- ANES Public Meeting -

We'll be holding our regular public meeting at the Midwest. Thursday night, April 3rd, at 6 p.m., in Parlor A on the 4th Floor of the Palmer House Hilton. We hope to see you there.

- NES 2002 -

We hope you've had a chance to use the 2002 ANES Advance Release already!

In the 2002 ANES, we were fortunate to be able to re-interview the respondents to the 2000 election study. In addition, we incorporated a small fresh cross-sectional sample to address panel attrition. The consequence is a study uniquely situated to gauge the effects of two big events - the 2000 election and the attacks of September 11th, 2001 - on Americans.

...Funding and Content

Thanks to the support of the Carnegie Foundation, the Center for Information and Research on Civic Learning and Engagement, the Institute for Social Research at the University of Michigan, the Office of Vice President for Research at the University of Michigan, the Russell Sage Foundation, and the University of Michigan Provost, the 2002 ANES incorporates both a core component to allow time-series comparisons with ANES studies of the past and several innovative new modules. Among the innovations are focused batteries of questions on the meaning and consequences of the 2000 election and the events of September 11th and questions on Americans' views on economic inequality. In addition, the study offers scholars a range of repeated measures from the 2000 study. These repeated measures will allow scholars to say, in the details, just what has changed, what hasn't, and why.

We hope scholars put the study to use for what it can say about continuities and disjunctions with the past. The time series design and the inclusion of core content on attitudes toward democratic procedure and the political system, engagement and participation in politics, identification with social groups, opinions on political issues, and more, accomplish that goal. We hope scholars will use the study to build accounts of individual change. The panel design accomplishes that with re-interviews of 1187 of the ANES 2000 respondents. And we hope scholars will use some of the new instrumentation to build cross-sectional accounts. The 2002 ANES carried multiple measures of concepts to give scholars the ability to work deeply in the data to build these cross-sectional accounts.

The final release will offer a range of new research opportunities, in addition to those we just mentioned. We are working to add some one hundred variables that build links to Members of Congress and to Congress as an institution to the data set.

...Technical Aspects of the Study

The 2002 American National Election Study interviewed 1513 respondents for approximately 30 minutes between September 18, 2002, and November 4, 2002, and re-interviewed 1347 of these respondents for another 30 minutes immediately after the 2002 election, from November 6, 2002, through December 6, 2002. The interviews were carried out over the telephone at the Indiana University Center for Survey Research.

The pre-election/post-election design is unusual for a mid-term ANES study. Why did the 2002 ANES use the pre-election/post-election design? For the past 25 years, our mid-term studies have carried instrumentation tied to the respondent's congressional district. In order to make the pre-loaded instrumentation in 2002 comparable with the past, we needed respondent addresses to place respondents in the proper congressional district. While we could get many of those via the tracking we do of panel respondents and while some portion of the others are available via reverse matching, we would have been left without addresses for a systematic group of respondents. Thus, we implemented a pre-election study to enable the pre-loaded instrumentation. This pre-election/post-election design also allows more distance between the measurement of things scholars use to predict vote choice and the measurement of vote choice itself.

For reasons of cost, the study was carried out over the telephone, and that has implications for measurement. We drew lessons from the mode comparison in 2000 to make sure that we implemented best practices on the telephone. As you know, ANES questions are generally designed for face-to-face interviewing. Because the 2002 ANES was carried out over the telephone, we had to modify a number of time-series questions to employ the best practices for collecting valid, reliable data on the telephone. Thus, there are several new question formats - branched questions instead of seven-point scales, altered response options to address recall issues over the telephone, and the like. Seven-point scales often produce unreliable data when the data are taken over the telephone. Long lists of response categories pose memory problems for respondents on the telephone. And so on. When there are question format differences with the past, they are a consequence of telephone interviewing.

...What Is an Advance Release?

Later this spring, we'll release the final version of the 2002 data. A number of members of the NES community have requested an advance release, one with basic cleaning, but without the range of constructed and contextual variables that will be on the final release and without full documentation. With this advance release, scholars who want to put in a bit of additional work can have access to the data at the earliest possible date. So, beginning with the 2000 ANES, we have been releasing the data in an advance release several months before the final data set is ready for release.

The 2002 ANES contains both panel respondents and respondents from a small fresh cross-section. When we expected variables to remain stable between 2000 and 2002 - for example, basic demographic variables - we did not spend precious minutes on the 2002 study asking those questions again of the panel respondents. Thus, scholars who use the Advance Release will want to combine the variables measuring basic demographics from the two separate variables measuring those demographics for the panel respondents and for the fresh cross-section. Some of the necessary accommodations to the telephone - the branched versions of the seven-point scales, for example - require recoding to build a consolidated variable. Scholars who use the Advance Release will want to build those scales. In the final release, we will have completed these tasks - and others like them - but we wanted to get the data to you as soon as it was usable.

...Access to the Instruments

Survey instruments for both the pre-election and post-election studies are available on our website: http://www.electionstudies.org/studypages/2002prepost/2002prepost.htm