HSW-0274 · How Studying Works
A learner opens the same chapter in two forms.
One is printed. The other is on a laptop, tablet, e-reader or phone.
The words may be identical, yet the reading task is not necessarily identical. One version may keep the whole page spatially stable. Another may require scrolling. One device may carry messages, tabs and habitual quick-check behaviour. Another may be dedicated almost entirely to reading. One digital text may add useful definitions or navigational support. Another may simply move the same prose onto a smaller surface.
Reading medium can affect comprehension, but “paper good, screens bad” is too crude to be useful.
This article owns the study decision: when should a learner care about medium, which features are doing the work, and how can digital reading be designed so the device does not quietly become part of the difficulty? It does not own the broader moral argument about screen exposure; use Is “Screen Time” the Wrong Unit for Judging Digital Learning? for that. It also does not replace comprehension monitoring; use Comprehension Monitoring and Repair when the issue is detecting and fixing misunderstanding.
Quick Answer
A 2026 network meta-analysis of 56 independent samples and 79 effect sizes compared paper with computers, tablets, e-readers and smartphones for expository reading. Paper ranked highest overall. The most important practical moderator was scrolling: when scrolling was required, paper showed a notable comprehension advantage over digital devices; when scrolling was not required, differences between paper and digital devices were small and not reliably different. See Clinton-Lisell and Litzinger, 2026.
A 2024 systematic review of 88 articles on children and adolescents was more cautious. Across ages, languages and reading outcomes, most studies found no medium difference. The review reported only weak evidence for screen inferiority in typical older readers and informational texts, while digital formats sometimes benefited engagement and other outcomes, particularly for younger children, learners with learning differences and enhanced digital materials. See Hare and colleagues, 2024.
The useful conclusion is conditional: medium matters most when it changes the reading operation—navigation, visible context, scrolling, distraction, pacing, annotation, accessibility or the learner’s monitoring of their own understanding.
Do not ask, “Is paper better than screens?” Ask, “What extra work does this medium add to this reading task, and does the learner have to pay for it with comprehension?”
1. A Medium Effect Is Not a Property of the Material Alone
Paper and screens are not single treatments.
A printed A4 article, a paginated e-reader, a tablet displaying one stable page, a laptop with a long vertical scroll and a phone showing six lines at a time are different reading environments. Grouping them all into “print versus digital” can hide the mechanism.
For studying, the unit of analysis should be the complete reading arrangement: text, device, navigation, interaction, distractions, time pressure and learner.
2. Why Spatial Stability May Matter
Long texts ask readers to connect statements that are separated in space. A definition introduced earlier may constrain a claim several paragraphs later. Evidence may need to be compared with an earlier table. A pronoun may refer back across a page boundary.
Paper naturally supplies stable physical locations. Digital pagination can do something similar. Continuous scrolling can make earlier information move relative to the screen, reducing stable “place on the page” cues.
The 2026 network meta-analysis treats this as one plausible explanation for why scrolling comparisons showed larger paper advantages than non-scrolling comparisons. The mechanism remains a research question rather than a settled single cause, but the study decision is immediate: if a digital text is hard because the learner keeps losing where information lives, change navigation before blaming reading ability.
3. Scrolling Is Not Just a Finger Movement
Scrolling can require the learner to move the text, relocate the reading position and reconstruct local context. On a short page, that cost may be negligible. Across a dense chapter, the repeated cost can accumulate.
The 2026 meta-analysis found paper advantages of roughly g = 0.35 to 0.48 in comparisons where scrolling was necessary, but small, unreliable paper–digital differences when it was not. This does not prove scrolling alone caused every difference: scrolling co-varies with device and layout in many studies. It does tell us that “screen” is too broad a design category.
4. Text Length Changes the Decision
A six-sentence passage and a forty-page chapter place different demands on navigation and integration.
The 2026 review discusses prior meta-analytic work finding little medium difference for texts under about 1,000 words, with larger paper advantages for longer material. The likely interpretation is not that a thousand words is a magical threshold. Longer material creates more opportunities for navigation, memory integration and spatial disorientation to matter.
For a short factual lookup, a phone may be perfectly adequate. For a dense argument whose structure must remain visible across many pages, medium deserves more attention.
5. Device Type Is Part of the Treatment
The 2026 network meta-analysis ranked paper highest overall, followed by tablets, e-readers, computers and smartphones. Those rankings are averages across included studies, not universal device laws.
A smartphone can be excellent for a flashcard prompt and poor for a complex comparison table. A tablet can display a full page comfortably. An e-reader can reduce some visual and notification burden. A desktop can display multiple sources side by side even though a single long scrolling text may be less comfortable.
Choose the device around the representation the task needs.
6. Digital Reading Can Add Useful Capabilities
Digital is not merely paper behind glass.
- text size can be enlarged;
- contrast can be changed;
- text-to-speech may support access;
- definitions can be opened quickly;
- search can locate recurring terms;
- notes can be linked across documents;
- interactive diagrams can reveal relationships;
- translation and accessibility tools can reduce barriers.
The 2024 systematic review found that enhanced digital formats can benefit some outcomes and learner groups. A study system that removes these supports simply to imitate paper may make learning less accessible without gaining any meaningful comprehension advantage.
7. Accessibility Is Not a Confound to Be Removed
If a learner needs enlargement, text-to-speech, switch access, colour adjustment or another legitimate accommodation, independence does not mean stripping those supports away.
The target capability is comprehension of the intended material under the learner’s appropriate access conditions. The medium decision must preserve access before it optimises convenience.
8. Familiarity With a Device Can Change Reading Behaviour
Devices carry learned expectations.
A phone is often used for fast scanning, messaging and short interactions. A printed textbook signals a different pace. These associations can influence how the learner approaches the text before any property of pixels versus ink becomes relevant.
One hypothesis in the literature is that readers can become more overconfident on screens, allocating less time or monitoring less carefully. This is plausible and supported in some prior work, but it should not be converted into a claim that every learner reads shallowly on every screen.
9. The Medium Can Change Metacognition
A student may finish a screen passage quickly and feel fluent because navigation was fast and the words looked familiar. That subjective ease can be mistaken for comprehension.
The repair is not to distrust screens automatically. It is to separate feeling from evidence. After a substantial section, close or minimise the text and reconstruct the argument, causal chain or main evidence without looking.
If the model survives, the medium worked. If it collapses, adjust the reading process or medium before simply increasing time.
10. Younger Readers Do Not Show One Uniform Screen Penalty
The 2024 systematic review found mixed effects in children and adolescents, with most studies showing no difference across a wide range of outcomes. A 2025 study of 58 first graders comparing paper, laptop and tablet reading also did not detect a general screen-inferiority effect; comprehension monitoring supported comprehension regardless of medium. See Florit and colleagues, 2025.
This matters because adults should not infer that a medium effect demonstrated in older readers, long informational texts or particular devices must generalise unchanged to beginning readers.
11. Informational and Narrative Reading Can Behave Differently
Informational texts often require readers to track definitions, causal relations, evidence and cross-paragraph structures. Narrative texts may provide different coherence cues and reading goals.
The 2026 network meta-analysis deliberately focused on expository text because prior work suggests medium effects are more consistently observed there. Therefore, do not apply its numeric results directly to novels, poetry, chat messages or every other reading form.
12. Time Pressure Can Magnify Weak Reading Habits
When students feel rushed, they may scroll quickly, search for keywords or skip backward checking. Paper can sometimes slow the interaction simply because its navigation is less frictionless.
But slowing is useful only if it produces better processing. Artificially making digital reading cumbersome is not a learning strategy. The aim is deliberate pace, not inconvenience.
13. Annotation Does Not Automatically Favour Paper
Writing in margins can support selection and external memory. Digital annotation can do the same, sometimes with search and linking advantages.
The important distinction is whether annotation changes cognition. Highlighting half the page on paper and highlighting half the PDF are equally weak if neither requires the learner to decide what matters.
Use annotations as questions, relationship marks and repair signals—not as proof that reading occurred deeply.
14. The Split-Screen Trap
Digital reading sometimes places the source on one side and notes, questions, chat or a second document on the other. This can be excellent when the learner must compare representations. It can also create constant attention switching.
Ask whether the second pane is required for the current operation. If the job is initial comprehension, a cleaner single-text view may be better. If the job is comparison, side-by-side visibility can reduce search and memory load.
15. Mathematics: Medium Follows Representation
A long worked solution may be easier to inspect when the whole derivation remains visible. If a phone shows only two algebraic lines at once, the learner must mentally reconstruct the missing steps. A large tablet, paper or large monitor may reduce that burden.
Conversely, a dynamic graphing environment can make parameter changes visible in a way static paper cannot. The correct medium depends on whether the learner is trying to inspect a stable derivation or explore a changing relationship.
16. Science: Keep Diagrams and Explanations Together
If a diagram sits three scroll lengths away from the paragraph that explains it, digital layout can create avoidable split attention. If a tablet displays both together with readable labels, the screen may be excellent.
Do not ask whether the lesson is digital. Ask whether the relevant representations can be coordinated without excessive navigation.
17. English and Humanities: Long Argument Tracking
For a dense argumentative essay, learners often need to return to earlier claims, compare evidence and inspect paragraph function. Stable pagination and visible headings can help.
If reading digitally, create landmarks: page numbers, section headings, bookmarks and a brief argument map. These tools restore some of the spatial and structural cues that continuous scrolling can weaken.
18. The Medium-Selection Matrix
| Reader job | Useful medium features | Main risk |
|---|---|---|
| Quick factual lookup | Searchable digital text | Stopping at a snippet |
| Dense long-form comprehension | Stable pages, low distraction, easy backtracking | Scrolling and shallow pacing |
| Cross-source comparison | Large screen or printed spread with simultaneous visibility | Constant tab switching |
| Accessible reading | Adjustable text, audio, contrast or assistive tools | Removing needed support for the sake of format purity |
| Dynamic system exploration | Interactive digital representation | Clicking without explanation |
| Exam simulation | Match the actual assessment interface where possible | Practising navigation unlike the test |
19. A Digital Reading Repair Protocol
- Remove unrelated notifications and tabs.
- Increase the text size until the page remains readable without excessive line wrapping.
- Prefer pagination or stable section landmarks for long texts when available.
- Pause after each meaningful section and state the claim or mechanism without looking.
- Use search only after trying to remember where the information belonged.
- Bookmark genuine breakdown points rather than highlighting everything.
- At the end, reconstruct the whole argument from a blank page.
20. A Paper Reading Repair Protocol
Paper can also be used badly.
- Set a reading question before beginning.
- Mark only relationships or genuine uncertainties.
- Do not let easy page turning become endless rereading.
- Close the text periodically and retrieve.
- Use an external note only when it compresses or reorganises meaning.
- Test the model after delay.
Paper is not a cognitive strategy. It is a medium that can support or hinder strategies depending on how it is used.
21. Parent and Tutor Guide: Watch the Reading Process, Not the Device Label
Instead of asking “Why are you always on a screen?”, ask:
- Can you show me where the main claim is?
- How do you return to an earlier definition?
- Are you scrolling because the text demands it or because you are searching for an answer?
- Can you explain this section after minimising the source?
- Would a larger or paginated view make the relationship easier to see?
The device becomes educationally relevant when it changes the quality of these operations.
22. When Switching to Paper Is a Reasonable Experiment
- The learner repeatedly loses their place in long digital texts.
- Comprehension is weaker than confidence suggests.
- Scrolling is heavy and cross-paragraph comparison matters.
- Notifications or habitual app checking repeatedly interrupt reading.
- The examination or assessment will itself be paper-based and format familiarity matters.
Treat the switch as a test. Use comparable material and compare delayed comprehension, not just comfort.
23. When Staying Digital Is the Better Choice
- The learner needs accessibility features.
- The task depends on interactive or dynamic representations.
- The text is short and navigation cost is negligible.
- The learner needs to compare several documents on a large display.
- The real future task is digital and navigation itself must be practised.
- A well-designed paginated digital version performs as well as paper for the learner.
24. The Changed-Medium Test
To find out whether the medium is carrying part of the performance, change it.
Read one matched passage on paper and another digitally under similar time conditions. Predict performance before answering. Test main idea, inference and relation questions. Repeat on another day with the media reversed. Then add a delayed explanation from memory.
One passage proves little. A repeated within-person pattern is more useful. If the gap disappears after digital navigation is redesigned, the issue may have been interface rather than “screen reading” broadly.
25. Evidence Boundary
Meta-analyses aggregate heterogeneous studies. Devices, age groups, languages, texts, time limits and interfaces differ. The 2026 network meta-analysis focused on controlled expository reading in a native language and excluded digital conditions with interactive features unavailable in print; its estimates should not be treated as the effect of every real-world digital text.
The 2024 review of younger readers found a much more mixed picture, including potential digital benefits for some outcomes and learners. Small qualitative studies can reveal mechanisms but should not be used to estimate universal effect sizes. No single mechanism—scrolling, cognitive load, metacognitive overconfidence, spatial cues or ergonomics—has been proven to explain every medium difference.
The defensible conclusion is conditional: when digital design adds navigation or monitoring costs, comprehension can suffer; when those costs are controlled or digital features genuinely support the learner, the gap can shrink or reverse.
26. Return: Choose the Reading Environment That Preserves the Model
The aim of reading is not loyalty to paper or loyalty to technology.
The aim is to build a coherent model that survives after the page, tab or device is gone.
Use paper when its stable layout and low interaction cost protect difficult comprehension. Use digital when its accessibility, search, comparison or interactive strengths genuinely serve the task. Redesign scrolling, distractions and landmarks when they are the problem. Then verify the choice with delayed, independent understanding rather than preference alone.
Continue through Comprehension Monitoring and Repair, Is “Screen Time” the Wrong Unit for Judging Digital Learning?, the How Studying Works Numbered Series Reading Index and the How X Works Hub.