Brown and Heathcote (2008) proposed the LBA as the simplest model of choice and response time data. This claim was, in part, based on the LBA requiring fewer parameters to fit most data sets than the leading alternative, the Ratcliff diffusion model (Ratcliff & Tuerlinckx, 2002). However, parameter counts fail to take into account functional form complexity, or how the parameters interact in the model when being estimated from data. We used pD, or the “effective number of parameters”, calculated from Markov Chain Monte Carlo samples, to take these factors into account. We found that in a relatively simple, simulated, data set and on average in a complex, real, data set that the diffusion had fewer effective parameters than the LBA.