Anderson, R. C., & Biddle, W. B. (1975). On Asking People Questions about What They are Reading. In Psychology of Learning and Motivation (Vol. 9, pp. 89–132). Elsevier

The authors review the effects of inserting “adjunct questions” into the reading experience (i.e. Adjunct questions improve comprehension of related but untested content): under what conditions is performance improved? What might be the mechanism?

Conditions under which adjunct questions are found to help:

• Repeated vs. new post-test items: Repeated post-test questions get a bigger boost than new questions (no surprise)
• Position: Asking questions after each section seems consistently better than before
• Timing: The closer to the material it’s about, the better, it seems (How often and where should mnemonic medium review areas be inserted?)
• Response mode: The impact on short answer questions seems to be greater than the impact on multiple choice questions.
• Feedback: Providing feedback usually (but not always) produces better results on new items.
• Overt vs. covert response: Overt responses seem to produce more facilitation.
• Motivation: There’s some light evidence that effect may diminish or disappear with motivation.
• Nature of the adjunct questions: “comprehension”-style questions seem to produce bigger effects than “verbatim” questions (though this paper found the opposite in its experiments)
• Questions vs. review statements: questions perform better than mere statements expressing the most important things to remember (related: Mnemonic medium prompts signal what the author finds important)
• Relevancy: the effect is only produced when the questions are relevant to the text

The author tries valiantly to experimentally validate a hypothesis that adjunct questions increase depth of processing, but they broadly fail.