Does Screen Reading Cost Comprehension

Why We Engage · 24 October 2026 · Reviewed by Faisal Zaidi

Screens are blamed for shallow reading, and the largest meta-analysis does find paper ahead. The margin is small, it shrinks for tablets, it depends on the text, and some of it sits in how well readers judge their own understanding. The evidence supports a narrower claim than the popular one.

Answer to the question

Does screen reading cost comprehension?

Slightly, and only under some conditions. Meta-analyses of screen reading find a small paper advantage in comprehension for informational text, no reliable gap for narrative text, and a smaller gap for tablets than the 2018 pooled estimate. They scored comprehension tests and did not measure depth of reading directly.

Key takeaways

Delgado and colleagues (2018) pooled 54 studies and 171,055 readers and found a small paper advantage, larger under time limits and absent in studies using narrative texts alone.
Clinton (2019) found paper ahead for expository texts, no difference for narrative texts, no reliable difference in reading time, and more accurate self-judgement after reading on paper.
Salmerón and colleagues (2024) found roughly half the 2018 margin for tablets and e-readers, larger for undergraduates and one-to-one sessions, with time limits, genre and publication year no longer predicting the gap.
Ackerman and Goldsmith (2011) found 61.0 per cent on screen against 60.7 per cent on paper with fixed study time, but screen readers predicted 71.0 per cent against 65.6 per cent; when readers set their own pace the scores were 63.2 and 72.3 per cent.
Carr's measured evidence on reading is a 2007 review of 38 hypertext experiments (p. 129), a question about links rather than screens.

Fifty-four studies and more than a hundred and seventy thousand readers found paper ahead of screens on comprehension, by a small margin. It is the largest meta-analysis of screen reading, and it is what people reach for when they say screens make us read less deeply.

Pablo Delgado and colleagues ran that analysis. A later analysis of tablets found about half that margin, and an older experiment on how readers judge themselves placed much of the cost in confidence rather than comprehension. Together they describe a modest penalty under particular conditions, with the sharpest problem in readers' sense of what they understood.

A small edge for paper, and only under some conditions

In the 2018 pooled data screen reading trailed paper by the same small margin whether the studies compared different readers on each medium or the same readers on both. It is the kind of difference that shows up across thousands of people and would be invisible in any one of them. Three things moved it. The paper advantage was larger when reading was timed than when readers set their own pace. It appeared for informational texts and for mixes of informational and narrative texts, and not for studies using narrative texts alone. And it grew across the period the studies covered, from 2000 to 2017.

A separate meta-analysis by Virginia Clinton agreed on genre: screen reading was worse for expository texts and showed no difference for narrative ones. That analysis found no reliable difference in reading time, so screen readers were not simply rushing through the page. It also found that readers judged their own performance more accurately after reading on paper.

Handheld devices narrow the gap without closing it. A 2024 analysis by Ladislao Salmerón and colleagues covered mostly tablets with some e-readers, and screen reading on them still trailed paper by a small margin in both kinds of study design, roughly half the size of the 2018 figure. The gap was larger for undergraduates than for school students, for whom it was absent and leaned slightly towards screens. It appeared in one-to-one sessions and not in group sessions.

The moderators that mattered in 2018 did not carry over. Time limits, text genre and publication year did not predict the gap in the 2024 data, so the finding that the penalty was growing did not reappear. The studies in the 2024 analysis also disagreed sharply with one another, which is why a single average says less than it seems to.

Cognitive load is one proposed mechanism for the penalty, and how it works at a till is described in [What Cognitive Load Costs at Checkout](https://thewhyweproject.com/articles/what-cognitive-load-costs-at-checkout).

Do screen readers overrate how much they understood?

A different kind of cost sits in how readers monitor themselves. Rakefet Ackerman and Morris Goldsmith ran experiments in 2011 in which participants read expository texts of 1,000 to 1,200 words, either on screen or on paper, and predicted their score on a later multiple-choice test. When study time was fixed, 70 participants scored the same on both media, 61.0 per cent on screen and 60.7 per cent on paper, but the screen readers predicted higher scores, 71.0 per cent against 65.6 per cent.

In a second experiment, 74 participants set their own study time, and the screen readers scored 63.2 per cent and the paper readers 72.3 per cent. The screen readers overestimated their scores by more, and they regulated their study time more erratically. The paper readers' average error was too small to distinguish from zero. The authors attribute the gap mainly to how readers monitor and control their own learning, rather than to anything the display did to comprehension itself. Clinton's finding that paper readers judged their performance more accurately points the same way, although the two bodies of work used different methods.

Later work from Ackerman's group found a related pattern. In three experiments with brief but difficult problems, the screen-related drops in performance and in the accuracy of confidence appeared under time pressure and not under a loose time frame, and they returned when the task was framed as a low-importance preliminary. The authors read the screen as a cue for shallower processing.

Self-monitoring is the part of this literature that fits the idea of shallow reading best, because it concerns whether the reader notices that a text has not landed.

It also changes what the cost of screen reading means in practice. If the penalty is partly a failure to slow down when a text has not landed, it is a matter of self-regulation as much as of the display, which is the authors' reading rather than a settled result. Kahneman makes the general version of the point in Thinking, Fast and Slow, that the confidence people feel in a belief follows how coherent the story is and not how much evidence stands behind it (p. 87), though he says nothing about screens.

If the penalty is partly a failure to slow down when a text has not landed, it is a matter of self-regulation as much as of the display.

For narrative reading the pooled evidence shows no reliable gap, so a novel on a tablet is not where the trouble sits. For informational text read against a deadline, paper has a small and fairly steady edge, and the edge is larger for undergraduates working alone. Give screen reading of expository text the time that paper would get, treat a feeling of having grasped it as a weaker signal than it seems, and keep paper for the contract, the report or the study guide where even a small and steady cost would matter to the person paying it. What screens displace more broadly is the subject of [What a Screen Replaces](https://thewhyweproject.com/articles/what-a-screen-replaces).

Carr's case, measured against the later evidence

Nicholas Carr's The Shallows made the claim familiar. Carr argues that a medium shapes the mind that uses it, and that one built around links and interruptions trains its readers to skim, so that screen reading becomes a different activity from reading a book. His opening chapters lean on testimony from readers who say they can no longer manage long texts, which is vivid and is offered as illustration rather than measurement. The measured evidence in the book is thinner and better.

Carr reports a 2007 review of 38 hypertext experiments that found "very little support" for the idea that links enrich reading, and found that the added demands of decisions and visual processing impaired performance. He adds that "not all the studies showed that hypertext diminished comprehension" (p. 129).

That review was about links inside a text, a narrower question than whether screen reading as such is worse. Another part of the book has aged less well. The Google effect, a memory claim the book cites, failed a preregistered multi-lab replication in 2018. The reading claim sits among the parts that later research has partly supported, and the support comes with conditions. The wider screen-time argument is weighed in [Both Sides of the Screen-Time Argument](https://thewhyweproject.com/articles/screen-time-both-sides).

What the comprehension tests did not measure

None of the analyses above measured depth in the sense Carr means. Each scored tests taken afterwards, and a test score is a narrower thing than a grasp of an argument. Whether screens change how deeply people think about a text, as distinct from how much they retain, is a question these designs were not built to answer, and settling it would take a different kind of study.

Common questions

Does screen reading cost comprehension?

Is the screen penalty getting bigger over time?

The 2018 meta-analysis found the paper advantage grew across studies published between 2000 and 2017. The 2024 analysis of tablets and e-readers found that publication year did not predict the gap, so the growth did not reappear in that data.

Do people overestimate how well they understood on screen?

In Ackerman and Goldsmith's 2011 experiments they did. Screen readers predicted higher scores than paper readers when study time was fixed, and overestimated by more when they set their own pace. Clinton's 2019 meta-analysis also found that readers judged their performance more accurately after reading on paper.

Does narrative text read worse on a screen?

The pooled evidence says no, and screen reading of stories is not where the penalty sits. The 2018 meta-analysis found the paper advantage in informational and mixed texts but not in studies using only narrative texts, and Clinton's 2019 analysis agreed. The 2024 handheld analysis found that text genre did not significantly change the gap.

Sources

Carr, N. (2010; updated edition 2020). The Shallows: How the Internet Is Changing the Way We Think, Read and Remember. London: Atlantic Books. ISBN 978-1-83895-258-7. p. 129 for the hypertext review.

Kahneman, D. (2011). Thinking, Fast and Slow. New York: Farrar, Straus and Giroux. p. 87 for confidence following the coherence of the story.

Delgado, P., Vargas, C., Ackerman, R., & Salmerón, L. (2018). 'Don't throw away your printed books: A meta-analysis on the effects of reading media on reading comprehension.' Educational Research Review, 25, 23-38. doi:10.1016/j.edurev.2018.09.003

Clinton, V. (2019). 'Reading from paper compared to screens: A systematic review and meta-analysis.' Journal of Research in Reading, 42(2), 288-325. doi:10.1111/1467-9817.12269

Salmerón, L., Altamura, L., Delgado, P., Karagiorgi, A., & Vargas, C. (2024). 'Reading comprehension on handheld devices versus on paper: A narrative review and meta-analysis of the medium effect and its moderators.' Journal of Educational Psychology, 116(2), 153-172. doi:10.1037/edu0000830

Ackerman, R., & Goldsmith, M. (2011). 'Metacognitive regulation of text learning: On screen versus on paper.' Journal of Experimental Psychology: Applied, 17(1), 18-32. doi:10.1037/a0022086

Sidi, Y., Shpigelman, M., Zalmanov, H., & Ackerman, R. (2017). 'Understanding metacognitive inferiority on screen by exposing cues for depth of processing.' Learning and Instruction, 51, 61-73. doi:10.1016/j.learninstruc.2017.01.002

DeStefano, D., & LeFevre, J.-A. (2007). 'Cognitive load in hypertext reading: A review.' Computers in Human Behavior, 23(3), 1616-1641. doi:10.1016/j.chb.2005.08.012

Camerer, C. F., et al. (2018). 'Evaluating the replicability of social science experiments in Nature and Science between 2010 and 2015.' Nature Human Behaviour, 2, 637-644. doi:10.1038/s41562-018-0399-z

Sparrow, B., Liu, J., & Wegner, D. M. (2011). 'Google effects on memory: Cognitive consequences of having information at our fingertips.' Science, 333(6043), 776-778. doi:10.1126/science.1207745