You open a long report on your phone and drag your thumb down the page. The text slides past, and no page ever holds still long enough to remember where a point sat.
Virginia Clinton-Lisell and Christine Litzinger of the University of North Dakota compared paper with four devices, computers, tablets, e-readers and smartphones, in a network meta-analysis published in Education and Information Technologies in January 2026. The analysis pooled 56 studies with independent samples and 79 comparisons. The studies used informational text and assigned readers at random or counterbalanced the conditions.
Across all comparisons, paper beat computers and smartphones by reliable margins. Tablets showed no reliable gap, and e-readers reached only marginal significance. Paper took the top ranking with a P-score of 0.97, ahead of tablets at 0.67 and smartphones at 0.13.
Scrolling sorted the results. In 25 studies where readers had to scroll, paper beat both computers and tablets by reliable margins. The smartphone gap did not reach significance in those comparisons, and no scrolling study used an e-reader. In 31 studies where readers did not scroll, because the text fit on the screen or the device turned pages, no digital device showed a reliable gap to paper. The abstract puts the effect sizes at 0.35 to 0.48 with scrolling and 0.03 to 0.12 without. With scrolling, paper beat computers and tablets by reliable margins. Without it, no device showed a reliable gap to paper.
The authors infer that scrolling interferes with comprehension. Readers store the location of information as part of what they remember, a concept the paper calls “place on the page,” and moving text erases that cue. The authors also review research showing that readers feel more confident about their understanding on screens than their scores justify. Paper still ranked first in the no-scrolling comparisons, and the authors point to eye strain and screen size as possible reasons.
The paper reports that an earlier meta-analysis found no medium difference for texts under 1,000 words. Short texts showed no gap in that analysis.
The authors state their limits. Most studies involved college students reading science texts in English, and 28 of the 56 did not report the reliability of their comprehension measure. The authors also coded scrolling as a yes or no, so the paper cannot say how much scrolling matters.
For a long text you need to learn from, pick paper or a paged view. Switch your reading app to page-turn mode if it offers one. Afterward, write three sentences from memory to check what stayed. These steps are suggestions, and the meta-analysis did not test them.
Tomorrow, print the one long article you plan to learn from.
Reference
Clinton-Lisell, V., Litzinger, C. Decoding digital reading: a network meta-analysis of comprehension across devices. Educ Inf Technol 31, 1611–1643 (2026). https://doi.org/10.1007/s10639-025-13843-8
Top comments (0)