This data dictionary is your guide to understanding Survey Response data — the substantive voter intelligence in the Exchange, and the data most members use most.
What is a Survey Response?
A Survey Response is the specific answer a voter gave to a question during a contact, like a support ID for a candidate, a position on healthcare, a vote plan, a volunteer interest. Submitted by member organizations across the ecosystem and normalized by DDx, this data is what lets a member organization see what voters have told the entire progressive movement, not just their own program.
How normalization works
Member organizations ask thousands of differently worded questions. DDx’s normalization process maps each member question into a structured, attribute-based taxonomy — categories like Candidate, Issue, Demographic, Voting, and Party ID & Engagement are first-class categories, each with defined sub-types and filter logic. This preserves the core intent of the question while obscuring the contributing organization’s proprietary script language. The same structure drives consistency across the wide variations in question wording, response scales, and field conventions submitted by hundreds of member organizations.
This is the third component of blinding (see “Our Data Blinding and Privacy Policies”), and it’s also what makes attribute-based subscriptions in Nexus possible: you subscribe to a structured category, and the platform matches and includes new questions automatically as data arrives — including questions that don’t yet exist when you subscribe.
How Survey Response data is organized
In Nexus, Survey Questions are organized into six categories — the same categories you filter by in Step 3 of the workflow:
- Candidate — Support IDs, favorability, and approval for named candidates and generic Democratic support scales.
- Issue — Positions and interest on policy topics like abortion, healthcare, the economy, gun safety, climate, immigration, and more.
- Ballot Initiative — Support and petition data for specific ballot measures.
- Voting — Vote plans, VBM status, ballot cure indicators, voting history.
- Demographic — Ethnicity, language preference, and race — observed, self-identified, preferred, or spoken.
- Party — Party affiliation, ideology, and favorability.
- Engagement — Organizing signals, volunteer interest, relational organizing responses, yard sign requests
Within each category, Question Type identifies what the question measures (e.g., Support vs. Favorability vs. Approval within Candidate), and Candidate/Issue Name identifies the specific subject where one exists.
Key fields in your data
Survey Questions and Responses are delivered in a single unified table, fully aligned with the updated Survey Question data model, with structured election metadata included in every export. The fields you’ll work with include the normalized question and response identifiers, the category and question type, the subject of the question, the voter’s response value, the blinding-period date, and enriched candidate and election context. For the current field-level schema, contact support@demexchange.com for documentation matching your delivery configuration.
Common response sets
- The Support Scale — Candidate Support responses use a numbered scale running from strong support through undecided to strong opposition. Lower numbers indicate support (1 — Strong Support, 2 — Lean Support), 3 indicates Undecided, and the highest values indicate opposition. Use Select Responses in Nexus to see the exact scale values present for your geography.
- Yes / No / Maybe — Common on engagement and commitment questions (volunteer interest, commit-to-vote).
Because contributing organizations’ original questions are normalized, response values are comparable across the ecosystem — a 1 on a support scale means strong support regardless of which organization’s script produced it.