Semiparametric Bayesian approaches to systems factorial technology

• A semiparametric piecewise exponential model is used to estimate response time hazard functions.
• This semiparametric model is fit to data in a Bayesian hierarchy.
• The resulting posterior estimates of the hazard functions can be used to compute the posterior estimates of systems factorial functionals.

Systems factorial technology (e.g., Townsend and Nozawa, 1995) is a toolbox of methods to test characteristics of multichannel processing systems. In this paper, we present a semiparametric Bayesian model of response times based on a piecewise-exponential model. This model allows us to compute the posterior estimates of the response time distribution hazard function and the systems factorial technology functionals that can be computed from it, including the survivor function interaction contrast and the capacity coefficient. We then compute posterior estimates of statistics that reflect different behaviors of the functionals, and so arrive at a fully Bayesian and theoretically-motivated approach to systems factorial technology. Simulation studies demonstrate the effectiveness of the approach, and we apply it to a published data set.