Textbooks and courses usually teach one kind of problem, then give a block of practice on exactly that kind, then move on. It feels efficient: by the tenth problem you are fast and accurate. Interleaving, mixing different kinds of problems together, feels worse. It is often much better.
The experiment
Rohrer and Taylor (2007) taught students how to find the volume of several kinds of geometric solids. Some practised in blocks, every problem of one kind before the next. Others practised the same problems shuffled together. During practice, the blocked group did better. On a test one week later the result flipped: the mixed group scored 63 percent, the blocked group 20 percent.