Rank Presumption of Strength of Conviction Level Based on Respondent's Mouse-log in Web-query with Image Choices

Abstract

A growing number of studies clarify user's behavior on the Web because of decreasing costs of data collection using the Web. The primary purpose of the present study is analysis of PC-mouse movements as behavioral information to reveal an answered rank to image choices from 9 choices. We analyzed a web-query with 300 people, and clarified the ability to presume the order of giving priority by PC-mouse movements. The agreement rate of the presumption order is 31%. There is a tolerance also in the noise caused by an illegal respondent and the age. When the sample size is 240 or more, the inferential accuracy becomes steady.