In one sentence
459 Swedish upper-secondary students who played Bad News, used a News Evaluator fact-check workshop, or sat subject literacy lessons did not show a clear three-month gain in judging true vs false posts compared with classmates who skipped those sessions.
What the researchers did
Nygren and Efimova tested delayed classroom interventions in authentic lessons (about half an hour of the digital activity, then a teacher-led discussion). Outcomes were measured three months later — not the next day.
The matched sample: 459 students in 43 classes (median 10 per class). 198 played Bad News (a game that lets students practise misleading tactics — “inoculation” / prebunking); 182 used the News Evaluator workshop (searchable Facebook-style posts plus prompts to check sources); 79 were a control. Assignment was by class and not fully randomised (school timetables constrained the process). Control students were, on average, further along in their programmes.
They rated tweet and Facebook items from 1 (very unreliable) to 7 (very reliable). The News Evaluator group was also asked whether they used search tools. Separate subject-specific workshops (up to five subjects) used 204 non-participants as a comparison.
What they found
- Bad News: no statistically reliable treatment × time effect on any of the four tweet items at three months.
- News Evaluator: no reliable delayed effect on the two Facebook items or on reported tool use. One credible post showed a marginal hint (p = 0.056) — not a confirmed gain.
- Students did not report using digital verification tools more three months later.
- Subject literacy sessions also lacked long-term effects in this analysis.
- Students who said credible information matters were better at spotting accurate news. Those who valued democracy were better at spotting false news.
- Actively open-minded thinking did not improve discernment; it tracked more scepticism toward both true and false items.
- Self-rated skill did not predict actual scores. Students on vocational tracks did worse at identifying trustworthy news.
What this means for learners and educators
- A single polished lesson can feel useful and still fade. Treat media literacy like spaced retrieval: short returns, mixed examples, and checks weeks later — not one assembly.
- Confidence is a poor dashboard. Feeling good at “spotting fakes” did not match performance here.
- Values and track mattered more than the one-off tools. Vocational classes may need more practice time, not a different slogan.
Limitations and what we don't know yet
The study was powered for a larger original plan (~1000 students / 50 classes); attrition and coding losses shrunk the matched set. Randomisation was incomplete. Control students differed at baseline. Outcome items had low internal consistency, so tweets were analysed one by one. Immediate post-tests from other Bad News / News Evaluator studies are not contradicted — this paper is about three-month classroom delay.
FAQ
Does this mean games like Bad News do not work?
No. Earlier studies found short-term effects. This trial asked whether a school slot plus discussion still showed up three months later — and it mostly did not.
Should schools drop fact-checking workshops?
The authors argue for repeated, embedded practice, not for abandoning the topic. One isolated hour was not enough for lasting behaviour change here.
Can I trust my own “I can spot fake news” feeling?
In this sample, self-ratings did not predict scores. Use blind checks on real items instead.