Article: Narrative Structure in Digital Content
Reviewed: November 2026 · Method version 2 · How these reviews work
Each factual claim in the article was checked against the source it cites, then challenged by reviewers whose job was to find the strongest evidence against it: replication attempts, meta-analyses, later papers that cite the original work, and research from other fields that study the same mechanism. A separate reviewer weighed both sides, and a script re-checked every quoted source. Where a claim was overstated or misattributed, the article was corrected. The research below supports the article as it now reads.
Deborah Small, George Loewenstein and Paul Slovic gave 159 people five one-dollar bills for completing an unrelated survey, then handed each a Save the Children appeal.
The story alone raised an average of $2.38 a person, the statistics alone $1.14, and the story with statistics $1.43.
Figures confirmed against Small et al. (Fig. 3) as reprinted and described in Slovic (2007).
In one well-known experiment it cost about 40% of the donations, though later tests of the same setup have not always found the drop.
A 2024 preregistered replication (Maier, Wong and Feldman, Collabra: Psychology) with hypothetical donations did not reproduce the drop from adding statistics. A 2010 field study (Friedrich and McGuire, Social Influence) found the drop only among less analytic donors, and three later appeal experiments (Hart, Lane and Chinn 2018; Behavioral Sciences 2022; Frontiers in Conservation Science 2025) found no drop.
The authors suggest the numbers may have pulled readers out of their response to one child and into a more calculating frame of mind.
The authors’ interpretation is reported accurately and with appropriate hedging.
Order matters for attention because a story raises questions the reader wants answered. Curiosity research suggests each unresolved turn is a reason to keep going, whether or not the page is useful.
The mechanism matches Loewenstein’s information-gap account of curiosity (1994); evidence on web reading completion was not found.
Green and Brock called the experience transportation: absorption into a story, made up of imagery, feeling and focused attention.
Across their experiments, the more transported readers were, the more their beliefs moved toward the story and the more favorably they judged its protagonists.
Small wording fix: the belief and protagonist findings come from three of the four experiments.
In Braddock and Dillard’s meta-analysis, exposure to a narrative correlated with story-consistent beliefs, attitudes and intentions at around .17 to .19, and with behavior at .23 in the five studies that measured it.
Stories move people a little, on average, though the size of the effect varies with the setting and the format.
Braddock and Dillard note unidentified moderators; effects vary by medium.
Kurt Vonnegut proposed in his rejected master’s thesis that stories have shapes, traced between ill fortune and good fortune from beginning to end.
The most common shapes were not the most downloaded. Ranked by Project Gutenberg downloads, which the authors call a rough proxy for success, the leaders were Icarus, Oedipus, and two “Man in a hole” arcs back to back.
Reagan et al. caution that downloads are a rough proxy and that a few high-download books drive the mode rankings.
For content teams the takeaway is modest: the most popular stories changed direction at least once, so a story with a turn may give readers more to follow than one that only climbs.
Film data (Del Vecchio et al., 2020) also link the Man in a hole shape to box office success, though text studies are mixed.
| **Will Storr | Abrams Press | 2020 | 304 Pages** |
Edition details aligned with the linked publisher page.
Paul Slovic: ‘Numerical representations of human lives do not necessarily convey the importance of those lives.’
Real projects have a setback, and the setback is often the most readable part: the data migration that stalled, the first rollout drivers ignored. It turns a climb into a “Man in a hole” shape, and it can make the eventual result more believable.
Related evidence: the blemishing effect (Ein-Gar, Shiv and Tormala, 2012) and underdog brand biographies (Paharia et al., 2011).
These edits were made on November 12, 2026 after the review.
Claim reviewed: Adding the statistics to the story cost about 40% of the donations in one well-known experiment.
Now reads: In one well-known experiment it cost about 40% of the donations, though later tests of the same setup have not always found the drop.
Why: The 40% figure is correct arithmetic for the Rokia study, and the sentence limits itself to that one experiment. Later tests of the same setup are split: a field study found the drop only among less analytic donors and a charity-appeal experiment found that adding information reduced empathy, while a preregistered replication with hypothetical money, a 924-person experiment, a volunteer-tourism experiment and a wildlife appeal found no drop. The number stands as reported, but readers should not take it as a stable effect size. Now cites Maier, Wong & Feldman, Replication and Extension of Small, Loewenstein, and Slovic (2007), Collabra: Psychology (2024).
Claim reviewed: Adding evidence can break the thing that was holding the reader.
Now reads: The uncomfortable possibility for anyone who writes for screens: adding evidence can weaken the thing that was holding the reader. Later tests of the same setup have been mixed, and persuasion studies outside charity often find a story and its numbers work better together, so treat this as a risk to test rather than a rule.
Why: This turns one donation result into a lesson for all screen writing. Direct tests of adding statistics or information to one person’s story are split: in charity appeals two studies found a drop (one only among less analytic donors) and four found none, and outside charity anecdotes keep their pull even when statistics sit beside them, while combining a story with arguments or statistics usually helps or does no harm. Whether added evidence hurts depends on the setting, the reader and where the evidence sits, so the article presents a contested point as settled. Now cites Allen et al., Testing the persuasiveness of evidence: Combining narrative and statistical forms, Communication Research Reports (2000).
Claim reviewed: A story raises questions the reader wants answered; each unresolved turn is a reason to keep reading.
Now reads: Order matters for attention because a story raises questions the reader wants answered. Curiosity research suggests each unresolved turn is a reason to keep going, whether or not the page is useful.
Why: Curiosity research supports the idea that an open question motivates people to seek the answer, and narrative-tension studies find curiosity organizes suspense. The claim is uncited, and no study found here shows unresolved turns increase how far people read a web page. Now cites Loewenstein, The psychology of curiosity, Psychological Bulletin (1994).
Claim reviewed: Across four experiments, the more transported readers were, the more their beliefs moved toward the story and the more favorably they judged its protagonists.
Now reads: Across their experiments, the more transported readers were, the more their beliefs moved toward the story and the more favorably they judged its protagonists.
Why: Three of the four experiments tested beliefs and evaluations of protagonists; Experiment 2 measured false notes instead. Most of the evidence is correlational, since transportation was measured rather than manipulated in three studies. A later study found transported viewers argued back more, not less.
Claim reviewed: Stories move people a little, and they do it reliably.
Now reads: Stories move people a little, on average, though the size of the effect varies with the setting and the format.
Why: ‘A little’ is fair, but ‘reliably’ overstates. The same meta-analysis found unidentified moderators, a health meta-analysis found print narratives had no significant effect, and the behavior estimate rests on five studies.
Claim reviewed: The most common shapes were not the most read; ranked by downloads, the most successful were Icarus, Oedipus, and two ‘Man in a hole’ arcs back to back.
Now reads: The most common shapes were not the most downloaded. Ranked by Project Gutenberg downloads, which the authors call a rough proxy for success, the leaders were Icarus, Oedipus, and two “Man in a hole” arcs back to back.
Why: The paper does name Icarus, Oedipus and two Man in a hole arcs as most downloaded. But the authors call downloads a rough proxy for success and say a few popular books drive the rankings, and a study of 40,000 narratives found no link between story structure and popularity.
Claim reviewed: A story that changes direction gives readers more to follow than one that only climbs.
Now reads: For content teams the takeaway is modest: the most popular stories changed direction at least once, so a story with a turn may give readers more to follow than one that only climbs.
Why: The article rightly calls this takeaway modest. The download data do not measure whether readers follow a story better, and results across media are mixed: Man in a hole films earned more at the box office, but another large text study found no link between structure and popularity. Now cites Del Vecchio et al., Journal of the Operational Research Society (2020).
Claim reviewed: Will Storr’s The Science of Storytelling (William Collins, 2019, 291 pages) is built around the idea that a story is about a character who has to change, with chapters on the dramatic question, plot and endings.
| Now reads: Will Storr | Abrams Press | 2020 | 304 Pages |
Why: The book’s sections on the dramatic question, plot and endings match the publisher page. The link, however, is the Abrams US edition (2020, 304 pages), while the article names the William Collins 2019 edition and 291 pages, which was not verified.
Claim reviewed: A real setback is the most readable part of a project story and makes the eventual result believable.
Now reads: Real projects have a setback, and the setback is often the most readable part: the data migration that stalled, the first rollout drivers ignored. It turns a climb into a “Man in a hole” shape, and it can make the eventual result more believable.
Why: This is uncited. Consumer research supports it in part: a small negative detail can raise evaluations, and underdog stories raise purchase intentions. But the blemishing effect appeared only when the negative came after the positive and readers were not processing closely, so the claim is stated more firmly than the evidence allows. Now cites Paharia, Keinan, Avery & Schor, The Underdog Effect, Journal of Consumer Research (2011).
Claim reviewed: The safer order is the story first and the numbers after the turn, carried by the people in it.
Now reads: The Rokia study did not test the order, and the few studies that have are small and split. Still, the version to test first is the story, then the numbers after the turn, carried by the people in it: “By March the morning reroute took 20 minutes.”
Why: The Rokia study never varied order: statistics were simply added to the story, so it cannot show that numbers after the turn are safer. The two studies that come closest point different ways: an earthquake-warning study found a case-based impression held when statistics came after it, while a skin-cancer message experiment found strong arguments worked better before the story and only weak ones gained from being embedded in it. The recommendation is a reasonable bet to test, not a finding from the cited study. Now cites When Arguments Meet a Story: An Experiment Testing Message Design Strategies for Skin Cancer Prevention and Detection, Health Communication (2024).