Benefits and challenges of using logistic regression to assess neuropsychological performance validity: Evidence from a simulation study