Do data from Mechanical Turk subjects replicate accuracy, response time, and diffusion modeling results?