Does allowing survey takers to switch devices improve responses?