Modeling dual-task performance: do individualized models predict dual-task performance better than average models?

Abstract

Understanding multitasking can be a complicated venture. The goal of this paper is to see whether using individual parameters for modeling dual-task will lead to better predictions of individual performance compared to using the global total average of all participants. It is expected that modeling individual skill will lead to more accurate models because individual parameters will lead to closer fits between individual data points and their models. Therefore, using individual parameters might provide a better explanation of the adaption of different strategies among participants. Data from a study involving 12 participants performing a phone-driving task has been used. The model consists of a driving and dialing model. The results of individualized performance show that using individualized parameters don't necessarily provide more accurate model predictions than using global parameters. This implies that factoring individual skill might not be very useful when modeling dual-tasking performance, however it does tell us an interesting story about the whether there are other individual factors we should take into account. The implication is that it's still too simplistic to look at just average performance to explain multitasking behavior, so it could be interesting to take a closer look at other individual factors in future research.