Inference instruction and reading comprehension

This meta-analysis looked at results from 2nd–9th grade inference interventions (with most on 3rd–8th-graders). The authors identified seven primary types of inference instruction, and describe each. The two most common (over 2/3 of studies) were:

  • using text clues (“…use clues in the text to construct a coherent representation”) &
  • background knowledge (“…use relevant background knowledge to fill in gaps in the text”)

he authors state, “Many of the studies focused on teaching students to locate relevant information in text to generate an inference, to integrate information across text, to provide evidence in the text of their answers to inferential questions, or to activate and integrate background knowledge with information in the text.”
Overall, they found moderate to large effects of inference instruction on reading comprehension for both skilled and unskilled readers. Interestingly, the less skilled readers also, “…benefitted substantially on literal measures of comprehension after receiving inference instruction.” That’s right! Inference instruction not only improved the students’ ability to “read between the lines”, but read, well, the lines as well. The authors predict it may be because: “Many of the studies in this review provided explicit instruction in finding pertinent information in a text and integrating it with prior knowledge to answer inferential questions. It seems that this type of instruction would be especially beneficial to less skilled readers, as it would… require them to attend to important details to which they may not otherwise have noticed.”
They also found that students in small groups achieved better learning outcomes than those in whole-class instruction, and that it didn’t take a ton of instruction to show effects: “Although higher order skills are considered difficult to teach, surprisingly, most of the studies showed positive results in relatively short periods of time (i.e., less than 10 hr).”
Of the various types methods of inference instruction, however, it was not possible to determine which was best. The authors explain that this is because so many of the studies used several types of inference instruction within one intervention, plus many methods of teaching one type (e.g. there are many ways to get a child to draw upon background knowledge), so results were impossible to tease apart.

Elleman, A. (2017). Examining the impact of inference instruction on the literal and inferential comprehension of skilled and less skilled readers: a meta-analytic review. Journal of Educational Psychology. Advance online publication.