Recently, SWIFT (Engbert et al., 2005,Psych Review, 112, pp. 777-813) has been proposed and implemented in a Bayesian inference framework (Seelig et al., 2009, arXiv:1901.11110). For simulated and empiricaldata representing a broad range of reading behavior, the model captures a substantial amount of variability in reading measures within and between subjects. Model parameters were fitted on training subsets foreach participant. Subsequently, data were predicted based on parameter point estimates and statistically compared against each respective participant’s complementary test subset. For the majority of testedsummary statistics, high correlations were found between predicted and empirical quantities, demonstrating high goodness-of-fit. Point estimates for model parameters were submitted to a regression analysis toexplain subject-level variability between experimental conditions such as normal, mirrored and scrambled reading. The results demonstrate that SWIFT can reliably explain, describe and predict reading behaviorunder a wide range of experimental conditions. (Email: Maximilian Rabe, email@example.com)
OSF does not support the use of Internet Explorer. For optimal performance, please switch to another browser.