Stated preference (SP) surveys typically ask respondents to make a choice under a hypothetical situation. However, the choice context is often unrealistic, leading to errors and biases in the response. To overcome this, revealed preference (RP) data has been used to create a more realistic choice context for generating SP questions. For instance, in stated adaptation (SA) surveys, users are asked to answer SP questions based on a specific RP context they actually experienced. One challenge in SA surveys is that it is difficult for the respondents to precisely recall the RP context, especially when there is a longer time gap between RP behavior and SP response (response lag). However, no empirical studies have been conducted to test how elicited preferences vary in response to changes in the response lag. This study empirically examines the impact of response lag on SP responses using real-time SA survey data collected from Kumamoto and Hiroshima, Japan. To accomplish this, we developed a survey tool that enables respondents to answer SP questions in real time, i.e., immediately after their RP behavior. The empirical results confirmed that systematic bias increases with an increase in response lag. Additionally, the results showed that the greater the response lag, the more respondents tended to focus on the SP attributes rather than the RP attributes. These findings indicate that the timing of responses is an important survey design parameter when conducting an SA survey.

陈述偏好（SP）调查通常要求受访者在假设情况下做出选择。然而，选择的背景往往是不切实际的，导致反应中出现错误和偏见。为了克服这个问题，显示偏好 (RP) 数据被用来创建更现实的选择上下文来生成 SP 问题。例如，在陈述适应（SA）调查中，用户被要求根据他们实际经历的特定 RP 上下文回答 SP 问题。SA 调查的一大挑战是受访者很难准确回忆起 RP 情境，尤其是当 RP 行为和 SP 响应（响应滞后）之间存在较长时间差距时。然而，还没有进行实证研究来测试引发的偏好如何随着响应滞后的变化而变化。本研究使用从日本熊本和广岛收集的实时 SA 调查数据，实证检验了响应滞后对 SP 响应的影响。为了实现这一目标，我们开发了一种调查工具，使受访者能够实时回答 SP 问题，即在他们的 RP 行为之后立即回答。实证结果证实，系统偏差随着响应滞后的增加而增加。此外，结果表明，响应滞后越大，更多的受访者倾向于关注 SP 属性而不是 RP 属性。这些发现表明，在进行 SA 调查时，响应时间是一个重要的调查设计参数。