How the Question Shaped the Answer: A Randomized Field Experiment Examining Elicitation Formats and Inflation Perceptions

ABSTRACT Objective I investigate the impact of changes in the format of survey questions on reported inflation perceptions. In 2022, the Reserve Bank of New Zealand replaced a range‐based elicitation method with an open‐ended format requiring respondents to provide numerical estimates. The consequences of this shift for measured inflation perceptions remain unclear, and this paper is devoted to evaluating the impact of that design change on reported inflation perceptions. Methods Using a randomized field experiment, I compare inflation perceptions elicited through a range‐based question with those obtained from open‐ended numerical responses. I estimate regressions using ordinary least squares to assess effects on the mean and quantile regressions to examine effects on the median. Results The experiment yields three key findings. First, respondents providing open‐ended numerical estimates reported inflation perceptions that were, on average, 2.08 percentage points higher than those selecting from predefined ranges. Second, median perceptions were 2.05 percentage points higher under the open‐ended format. Third, median perceptions showed no systematic association with demographic or behavioral characteristics. Conclusions The results reveal anchoring effects in inflation recall and suggest that the previous range‐based survey design may have led analysts and policymakers to overestimate the alignment between perceived and official inflation. The findings reinforce that elicitation formats can systematically shape reported inflation perceptions.