The fundamentals of the ANES dataset — what it measures, how probability surveys work, and how to read a variable codebook. Students pick their first variable and describe what it captures.
The American National Election Study (ANES) is one of the longest-running political surveys in the world. Since 1948, the ANES has interviewed nationally representative samples of U.S. adults before and after every presidential and midterm election, asking about their political opinions, values, candidate evaluations, and voting behavior. The result is a dataset spanning more than 70 years and tens of thousands of respondents — not a simple spreadsheet, but a living archive of American public opinion across generations.
Every ANES question has a unique identifier called a VCF code (short for Variable Center Format). Examples: VCF0004 is the respondent's age, VCF0803 is a thermometer rating for the Democratic Party, VCF0837 is racial resentment. The VCF system lets you refer to any specific question precisely — it is the equivalent of a variable name in a dataset, but designed to stay consistent across decades of surveys even when question wording changed slightly.
Each VCF code points to exactly one survey question across all ANES survey years it appeared in. A single code like VCF0803 links to the same conceptual measurement -- even if the exact wording shifted slightly in 1990 or 2000. This consistency is what makes 70-year trend analysis possible.
The question wording, exactly as it appeared on the interview form.
The complete list of answer choices (e.g., 1=Extremely liberal, 7=Extremely conservative).
The non-response codes (e.g., 9=Do not know, 0=Refused).
Which survey years the question was asked.
The variable type (scale, thermometer, binary, etc.).
The question wording is not a label — it is the exact language an interviewer read aloud to a respondent. Subtle differences in wording can produce different answers. "Do you favor or oppose cuts to social spending?" and "Do you favor or oppose reductions in government welfare programs?" ask about similar things, but respondents answer them differently. When you look up a VCF code in Pulse, always read the actual question wording before drawing any conclusion about what the number measures.
The answer choices tell you what each numeric value means. A response of 4 on VCF0803 (party thermometer) means a neutral rating of 50 on a 0–100 scale. A response of 4 on VCF0809 (7-point ideology scale) means "moderate." The number alone is meaningless without the codebook.
Not everyone answers every question. Respondents say "don't know," refuse to answer, or simply weren't asked a question in a given year. ANES marks these cases with non-response codes — typically values like 0, 8, or 9, depending on the variable. These are not real opinions. Treating a code of 9 (Don't know) as if it meant "strongly agree" would corrupt your analysis. A critical first step with any ANES variable is identifying its non-response codes and excluding those cases before computing any statistics.
Try it now: Use the Pulse Variable Browser to look up any VCF code that interests you. Read the exact question wording, note the answer scale, and identify the non-response codes. The interactive tool at Layer 1: Variable Anatomy → walks you through this process with real ANES variables.