How to measure time preferences in children: a comparison of two methods