Interleaving

Mixing problem types makes a practice session look worse and a later test look better. These entries separate the effect, which is reasonably robust, from the explanation most often given for it, which is tested only indirectly.