In the last decade, web surveys have replaced traditional methods, such as in-person interviews, offering cost reduction and paradata collection but presenting complexities due to various question formats and dynamic features. To test the proper implementation of web surveys, researchers rely on costly and time-consuming pretests by humans. Our tool SurveyBot automates this pretesting with two rule-based bots (Basic and Medium-I) and two AI-based bots (Medium-II and Advanced) that use Large Language Models. Tested on two web surveys, SurveyBot effectively handles diverse question types and dynamic features, reducing pretesting costs and human effort. |
*** Title, author list and abstract as submitted during Camera-Ready version delivery. Small changes that may have occurred during processing by Springer may not appear in this window.