Working Paper: NBER ID: w29686
Authors: Beatrice Ferrario; Stefanie Stantcheva
Abstract: This paper illustrates the design and use of open-ended survey questions as a way of eliciting people's first-order concerns on policies. Multiple choice questions are the backbone of most surveys, but they may prime respondents to select answer options that they would not naturally have thought about, and they may omit relevant options. Open-ended questions that do not constrain respondents with specific answer choices are a valuable tool for eliciting first-order thinking. We discuss three text analysis methods to analyze open-ended questions' answers. To illustrate how to apply these methods, we provide evidence from large-scale surveys on income and estate taxation. We show the that key concerns relate mostly to distribution issues, fairness, and government, rather than to efficiency concerns. There are large partisan gaps in the first-order concerns on policies.
Keywords: open-ended survey questions; text analysis; taxation; first-order concerns; public policy
JEL Codes: D72; D91; H1; H2
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| type of survey question (open-ended vs. closed-ended) (C83) | depth and relevance of the responses (C83) |
| political affiliation (D72) | nature of respondents' first-order considerations (D91) |