RT ANALYSIS WITH THE WEIBULL-GAUSSIAN CONVOLUTION MODEL

  • Soyogu Matsushita

Abstract

The author analyzed the reaction time (RT) distribution using the convolution of Weibull and Gaussian distributions. RT analyses have often been conducted with the ex-Gaussian model, whose hazard function converges to a constant. However, for the visual search task, the hazard is not always stable; there is a possibility that the hazard directs upward. Hence, the author replaced the exponential of the ex-Gaussian with the Weibull and suggested the Weibull-Gaussian model. The Weibull-Gaussian model is a general-purpose model that has backward compatibilities to the ex-Gaussian and Weibull, because each of them is a special case of the Weibull-Gaussian. The author conducted the visual search experiment and then fitted the model and estimated the parameters. The results revealed that both the ex-Gaussian and Weibull were not sufficient for RT distributions, and the author concluded that the flexibility of the Weibull-Gaussian was of benefit for the analysis of RT distribution.

Section
Full Articles