NES 101: An Introduction to the National Election Studies
|
NES 101: An Introduction to the National Election Studies
|
An Introduction to the
American National Election Studies (ANES)
Jeremy C. Pope, Political Science (August 21, 2002)
The American National Election Study was born out of the research of Angus Campbell, Philip Converse, Warren Miller and Donald Stokes. They published The American Voter in 1960 based on their analysis of the 1950s elections. Their research institute (The Survey Research Center and the Center for Political Studies of the Institute for Social Research at the University of Michigan) studied the 1952 through 1976 elections. This was essentially a private research effort, but was widely influential in political science, and even outside the discipline. Their research developed the basis for behavioral political science and such core concepts as party identification and the basic models of vote choice. It was also extraordinarily influential with regard to the survey questions that measure demographics and other relevant political variables.
In 1977, the National Science Foundation authorized funding to create the National Election Studies. The change in management and funding precipitated some changes to the survey instrument (see the note on the Continuity Guide below). But the core mission remains the same: “to generate data pertaining to the citizen’s social background, political predispositions, underlying social and political values, contemporary perceptions and evaluations of relevant groups and would-be leaders, opinions on questions of public policy, and participation in political life (NES Website).”
This broad mandate means the NES studies have a wealth of data, not always consistent or connected by theory. But there is no better source for understanding American public opinion—particularly for time-series comparisons of political beliefs and evaluations.
The National Election Study conducts and compiles
surveys of the American electorate in most presidential and midterm election
years. As of summer 2002, 23 biennial election studies from 1948 through 2000
were archived in the resource (currently loaded into DEWI). Additionally, there
are three panel studies spanning the 1956 – 60, 1972 – 76, and 1988 – 92
elections. Finally there are several pilot studies generally testing questions
and experimenting with the survey instrument.
The sample is a stratified random sample of American adults (over 18, and therefore eligible to vote). Sample size is typically between 1800 and 2700. In recent years it is lower as non-response rates have risen. There are no oversamples for specific populations (e.g. minorities), thus the NES is probably somewhat limited in its applicability to those populations.
Typically (though not always so there are some missing values) each person is interviewed twice: once before the election and once after the election. In the documentation responses are designated as either pre- or post- answers. The codebooks generally indicate this, but researchers should be careful to note whether they are using variables drawn from the pre- or post- survey. Obviously certain questions are only asked on the post-election survey (e.g. vote choice), but some questions have been asked on both portions. Researchers should simply be aware.
The NES covers a great deal of ground. The
following outline is typical of the question topics on recent surveys. Because
of the pre/post format researchers should always consult the documentation (see
below) to learn more about specific variables. The following section parallels
descriptions that can be found on the NES website. While it is not
comprehensive, it does provide a flavor of the data that can be searched via
the DEWI system.
1. Social And Religious Characteristics of the Electorate: Age, Gender, Race, Level of Education, Occupation, Union membership, region, self-identified social class, religion, attitudes toward the Bible, religious practice, and church attendance. Additionally, recent surveys have introduced some factual knowledge questions about American government to measure the respondent’s political knowledge.
2. Partisanship and Evaluation of the Political Parties: self-identification of partisanship is measured through a seven-point scale (1 = Strong Democrat, 7 = Strong Republican), a very common measure on the NES. Additionally, there are several party evaluation questions including questions about which party is more conservative, capable, and better able to keep the U.S. out of war. Additionally the NES allows open-ended responses about respondent’s feelings about the parties.
3. Ideological self-identification: This is also measured on a seven-point scale (1 = Very Liberal, 7 = Very Conservative). This section of the survey also includes questions about feelings toward conservatives/liberals, etc.
4. General Public Opinion: A host of questions about any number of issues is included in each electoral cycle. While many are unique to that election, some are consistent across elections. These questions are often but not exclusively measured on seven-point scales.
a. Social Welfare and the Role of Government: power of the federal government; nationalized health care/national health insurance; government support for guaranteed jobs/standard of living; government spending
b. Race: change in the situation of blacks; school busing/school desegregation, federal aid to blacks; open housing
c. Social Issues: women’s role; abortion, school prayer
d. Military/Foreign Policy: likelihood of war; levels of isolationism; military spending
e. Economy: Condition of nation’s economy; condition of respondent’s situation; various government policy effects
5. Support for the Political System: This section of the survey includes measures of trust in federal and state governments; measures of respondent’s perceived political efficacy; measures of perceptions of government responsiveness; and approval ratings of Congress and the President. Each of these areas have index variables that simply add up positive and negative responses to specific questions. But there are also specific questions like “are government officials crooked,” “how much does the government listen to people,” or “do elections make the government pay attention.
6. Political Involvement and Participation in Politics: This section includes numerous variables on the respondent’s registration, turnout in previous elections and campaign participation. The campaign participation variables include such items as whether or not the respondent worked for a candidate, attended meetings, or gave money. There are also questions about the parties and political mobilization to measure how much contact respondents receive from political parties and that contact’s effectiveness. Finally there are a number of questions about engagement in the campaign, including such things as television news consumed, interest, and expectations about the campaign.
7. Evaluation of the Presidential Candidates: Evaluations of the candidates (including a number of open-ended items) is available. Typical measures include approval ratings (for incumbents), feeling thermometers, and the candidate “affect” and “trait” variables. These latter variables are designed to measure the emotional reaction evoked by candidates and the candidate’s reputation. An example of an affect question is “has [candidate] ever made you feel angry?” Other affect measures include afraid, hopeful and proud. Traits associated with a candidate include: inspiring, knowledgeable, moral and strong leader.
8. Evaluation of Congressional Candidates: Evaluations of congressional candidates are far more limited. There are typically feeling thermometer questions and questions about contact with their members of Congress.
9. Vote Choice: Respondent’s are asked about their vote and time of decision for both presidential and congressional races.
The National Election Study (NES) web site is an excellent place to find information and documentation on the National Election Study. Data can be downloaded directly from that site in an ASCII, SAS, or SPSS format. Documentation can be downloaded as ASCII (text), PDF, or other formatted files. You will need to register the first time you download any files. Reading through the documentation on the site is essential for a serious analysis of NES data. Some examples include:
The original questionnaires for each survey are invaluable for analysis. Typically the questions are asked in roughly the same order and so once a student has skimmed one questionnaire he or she should find other studies to be very similar. Questionnaires can be downloaded with the data on the National Election Study (NES) web site.
Each election’s survey codebook clarifies whether the question was asked on the pre- or post-election survey. Additionally it clarifies the question of format. Some of the questions are asked in a branching format, rather than as a simple question. For researchers who merely want a list of the questions the introduction provides limited documentation. But for specific questions of format, the main body of the codebook will be required. Appendices tend to deal with questions of weight and corrections to the data. Codebookscan be downloaded with the data on the National Election Study (NES) web site.
Obviously, one of the most important features of this survey (similar to the GSS) is the ability to analyze the same question or set of questions across a number of years—roughly the post-War era. Continuity of questions is thus a significant issue for the board of governors that oversees the project and researchers using the data. The Continuity Guide assists the researcher in discerning which variables are asked when, and in what format.
Researchers interested in the pilot studies, or in experiments to verify the validity of the variables used in the NES, should consult the pilot study reports. These reports—prepared by prominent political scientists—verify such measures as the seven-point scale, measures of political knowledge and other similar variables.
The NES compiles and maintains a bibliography of works that use the NES data. While the NES Bibiliography is not comprehensive, it is a good resource of scholarly work on important issues in political science related to public opinion and vote choice.