C28 Learning outside the classroom
narrowed 1 attack: 1 narrowed · serves Q1 · last change 8 November 2026
Curriculum-linked learning outside the classroom (field trips, museum visits, outdoor teaching) raises school achievement compared with teaching the same content only in the classroom, by about 0.3 standard deviations in randomised studies (about 0.5 across all studies, inflated by non-randomised ones), and more in primary school than later.
Go deeper
Where it holds
All limits (6)
Inside the circle: the core of the claim. Outside: every bound and carve-out in the claim file, numbered, with the attack that forced or added it where the record shows one.
forced by a narrowing attackin the original wordingwritten at creation, or added without a narrowing
Warm cards in the list were added by later attacks; plain cards were in the original draft or written at creation.
Structured, curriculum-linked activities outside the classroom; not homework (C27) and not unstructured leisure.
Written when the claim was created (29 October 2026); not forced by an attack
Evidence on file: Delen et al. 2026 (Sustainability 18(3), 1437, DOI 10.3390/su18031437; full text read 2026-11-08, open access): 42 studies (1970–2022), 54 effect sizes; overall g = 0.529 (95% CI 0.392 to 0.665; Q = 744.8, df 53); by design (Table 3), randomised studies g = 0.307 (0.167 to 0.447, k = 29) vs non-randomised g = 1.020 (0.722 to 1.318, k = 16), and the authors say stronger claims should rest mainly on randomised or well-controlled studies; by region, North America g = 0.149 (−0.010 to 0.308, k = 15) and Scandinavia g = 0.180 (−0.093 to 0.454, k = 4), against Türkiye 1.023 (k = 11); by subject, science 0.632, social science 0.757, maths 0.295, reading 0.166; by level, elementary 0.805, middle 0.617, secondary 0.500 (grade slope −0.136, p = .002; subgroups not adjusted for design); exposures of 2–6 hours g = 0.034 (−0.163 to 0.230, k = 7).
Written when the claim was created (29 October 2026); not forced by an attack
Wording changed later: Forced by attack #2 · Delen, Özüdoğru, Şen & Biasutti 2026 · 8 November 2026.
The meta-analysis does not code or report test delay, so durability is not addressed; its assessment-item-type figures differ between Table 3 and the text.
Forced by attack #2 · Delen, Özüdoğru, Şen & Biasutti 2026 · 8 November 2026
Holmes 2011 (Learning Environments Research, DOI 10.1007/s10984-011-9094-y; abstract only): 228 grade-6 pupils randomly assigned to control, museum exhibit, lesson or both; the exhibit group gained, but no lasting differences between groups were shown — a warning on durability.
Written when the claim was created (29 October 2026); not forced by an attack
A bold claim: the size rests on one meta-analysis, now read in full; the half-SD average is not the randomised estimate.
Written when the claim was created (29 October 2026); not forced by an attack
Wording changed later: Forced by attack #2 · Delen, Özüdoğru, Şen & Biasutti 2026 · 8 November 2026.
Regular weekly outdoor teaching (udeskole-type), full text of Becker et al. 2017 (IJERPH 14(5) 485, DOI 10.3390/ijerph14050485; §3, Tables 1–3): a systematic review of 13 studies found no randomised trial, one controlled study with standardised tests (a one-year out-of-school science programme for fifth graders; reading and writing higher than controls, p = 0.03, no effect size given, maths not reported) and no delayed achievement tests.
Added with attack #1 (corrected; it did not narrow the claim) · Becker, Lauterbach, Spengler, Dettweiler & Mess 2017 · 2 November 2026
1 of 6 bounds and carve-outs were forced by a specific attack.
How it changed
The ledger records 1 attack (and 1 correction to our notes), but no earlier wording of C28 is on record, so the text cannot be replayed step by step. Only the current wording is shown.
Curriculum-linked learning outside the classroom (field trips, museum visits, outdoor teaching) raises school achievement compared with teaching the same content only in the classroom, by about 0.3 standard deviations in randomised studies (about 0.5 across all studies, inflated by non-randomised ones), and more in primary school than later.
What would change this
What would kill it: A meta-analysis or large randomised trial finds that curriculum-linked out-of-school learning does not raise achievement over classroom teaching of the same content, or that the gain is gone on a delayed test. Added as a claim on 29 October 2026, promoted from a candidate belief in the evidence library.
Attack record
1 attack: 1 narrowed. Counts of attacks, not a score.
Full attack list (1)
The full text of Delen et al. 2026, the only source for C28's "about half a standard deviation", shows that the half-SD average is carried by non-randomised studies: randomised studies give g = 0.307 (95% CI 0.167 to 0.447, k = 29) against g = 1.020 for non-randomised ones (k = 16), and studies from North America (0.149, CI −0.010 to 0.308) and Scandinavia (0.180, CI −0.093 to 0.454) do not clearly differ from zero.
Strongest reading and verdict reasoning
Steelmans to: C28 states the size as "about half a standard deviation on average", but the authors themselves say stronger claims should rely mainly on randomised or well-controlled studies, and in their own Table 3 the randomised estimate is about 0.3; the half-SD figure is the pooled value across designs (Q_B = 17.998, p < 0.001 for the design split).
Why: Nightly-attacks 2026-11-08 Europe/Stockholm. Delen et al. had been on file as abstract only; the publisher page opened tonight and the full text was read. Not a restatement of the existing Bounds ("high heterogeneity; depends on how well the activity is tied to teaching" did not split by design or region) and not a duplicate of the 2026-11-02 Becker correction. "More in primary school than later" holds in the full text (elementary 0.805, middle 0.617, secondary 0.500; grade slope −0.136, p = .002), though those subgroups are not adjusted for design. Not a preprint. Claim stays under ~600 characters, so no rewrite to shorten the claim was needed. Status alive → weakened.
Corrections to our notes
These fix a number or a fact in our notes. They are not attacks and do not count toward the floor of 3 survivals.
- — C28's limits do not record what the full text of Becker et al. 2017 shows for regular weekly outdoor teaching: no randomised trial, one controlled standardised-test study, no delayed achievement tests (a closer reading of the study).
Details
What was checked: Bounds (not a kill-condition test).
Why: Nightly-attacks run of 2026-11-02 (re-run in the day after the scheduled run stopped for lack of usage) Europe/Stockholm. Checked tonight in the stored full text: Ernst et al. one-year out-of-school science programme, fifth grade, reading and writing higher than controls (p = 0.03), maths not reported; review calls for randomised trials (Discussion). A correction of the evidence picture, not a test of the claim (no kill-condition result).
Evidence
7 linked sources: 3 supports, 3 limits, 1 challenges (first-pass draft, pending human review). supports: Delen et al. 2026, Larsen et al. 2026, Pulido et al. 2025 · limits: Holmes 2011, Becker et al. 2017, Otte et al. 2019 · challenges: Otte et al. 2019