American National Election Studies, Time Series Study, 2008

Data Archive > U.S. Surveys > General Population > National > National Election Studies > Summary


The ANES 2008 Time Series Study is the 28th study in a series of biennial election studies conducted since 1948 (the "ANES Time Series"). The main goal of the study is to allow a broad cross-section of scholars and citizens analyze survey data pertinent to important questions about vote choice, turnout and related matters in the context of the 2008 federal election. In addition to content on electoral participation, voting behavior, and public opinion, the 2008 ANES Time Series Study contains questions in other areas such as media exposure, cognitive style, and values and predispositions. Special-interest and topical content provided significant coverage of foreign policy, including the "war on terrorism" and the war in Iraq. In addition, the study carried expanded instrumentation on organizational membership, unemployment, the federal budget, modern sexism, and race and gender politics. The post-election interview also included Module 3 from the Comparative Study of Electoral Systems (CSES).

The ANES 2008 Time Series Study contains several questions on religion, including religious affiliation, views of the Bible, church attendance, frequency of prayer and spouse's religious affiliation. Detailed information on all the questions that were asked can be found in the report "Background Information on the 2008 ANES Time Series Questionnaires."

Data File
Cases: 2,323
Variables: 1,956
Weight Variable: V080101, V080102, V080101A, V080103
Data Collection
Date Collected: Pre-election survey: September 2 - November 3, 2008 Post-election survey: November 5 - December 30, 2008
Funded By
This material is based upon work supported by the National Science Foundation under Grants SES-0535334, SES-0720428 and SES-0840550. Additional funding from NSF was made available in conjunction with the Department of Homeland Security as part of grant SES-0651271. ANES is also supported by the University of Michigan and Stanford University.
Collection Procedures
Field operations were conducted by RTI International (www.rti.org). All interviews were administered face-to-face using Computer-Assisted Personal Interviewing (CAPI) technology, which also incorporated, for the first time in an ANES study, an interview segment in each wave that was self-administered. The selfadministered module in the pre-election survey was administered by Audio Computer-Assisted Self-Interviewing (ACASI), for which respondents used headphones and looked directly at the laptop computer screen out of view of the interviewer so they could answer audio-played questions confidentially.

A total of 2,323 pre-election and 2,102 post-election interviews were successfully completed during the field period, including 512 Latino interviews and 577 interviews by African American respondents. These counts are computed from the racial/ethnicity identifications in the Household Screener, which was used to roster eligible household members and randomly select a survey respondent. In the Household Screener application, "Latino" can designate persons of any race and "African American" designates those respondents not identified in the Household Screener as Latino who identified a single race of black/African American in response to the Screener race question.
Sampling Procedures
The target population for the ANES 2008 Time Series Study constitutes English-speaking or Spanish-speaking U.S. citizens of voting age residing in the 48 coterminous United States and the District of Columbia. For the purposes of this study, voting age was defined to be 18 years or older as of October 31, 2008. Since the eligible-age cutoff date was necessarily conveyed to one or more household members as part of the roster/selection process, this October 31st date was used in place of the actual November election date in order to avoid the suggestion of overtly political or election-related survey content, so that potential respondents without strong interest in politics and elections might be more motivated to participate.

The ANES 2008 Time Series Study was designed with a target of 2,470 total pre-election interviews, including a base target of 1,810 interviews plus 350 supplemental Latino interviews ("Latino oversample") and 310 supplemental African-American interviews ("African-American oversample"). Completion of 507 total Latino interviews and 527 total African-American interviews in the pre-election wave, including both base and supplemental interviews, were additional sample objectives. Differential sampling rates among race/ethnicity groups were needed to achieve the target distribution of survey participants.
Principal Investigators
Jon A. Krosnick, Arthur Lupia, and Vincent Hutchings
Note on Weight Variables
Unlike oversamples present in previous ANES Time Series studies (1964, 1968 and 1970), the oversamples in the ANES 2008 Time Series Study are integral to the cross-section, which can only be represented with the use of the sample weights provided in the dataset. The inclusion of the "oversample" cases when representing the 2008 Time Series cross-section provides improved estimates for the Latino and African American populations of eligible voters.

There are two sets of sample weights. The first set of weights is centered at a mean of 1.0; these are variable V080101 (pre-election) and V080102 (post-election). The second set of weights represent population V080101a (pre-election) and V080102 (post-election). The pre-election sample weights are the product of the household non-response adjustment factor by age and education. The post-election sample weights are adjusted for attrition. The household weight (V080103) used in creation of the sample weight is also available. Analyses intended to generalize to the target population should be weighted. The unweighted data are not representative of the target population, so unweighted estimates of population percentages and means are wrong. Also, due to the complex sample design of the ANES, sampling errors and related statistics (including confidence intervals, p-values, t-tests, and all other tests of statistical significance) should not be calculated using methods intended for simple random samples.