[xapata]

When I read an analysis paper, I find it more convincing if the dependant variable is continuous rather than binary. It's too easy to find spurious statistical "significance" otherwise. The heuristic holds up here as well.