摘要
Commonly used models for predicting movement time, such as Fitts' law and the ballistic movement model, are generally restricted to one-dimensional movement. To address this issue, this study developed a model applicable for the time prediction of two-dimensional movement. In Experiment 1, start points and targets were randomly generated to collect simple movement and clicking behavior. This diverse dataset enabled model selection for identifying the more appropriate prediction formula to be used while minimizing potential confounds among factors encountered in prior studies. The accuracy of the established model was subsequently verified using data collected when interacting with ordinary interfaces in Experiment 2. Results of the correlation analysis and prediction interval revealed that the model effectively captured overall trends of movement time and exhibited strong predictive reliability despite significant prediction errors. Moreover, prediction errors can be further reduced by considering the task types and individual differences among users when necessary.