Before You Get Too Excited About That Trigger Warning Study…
Read the original on Slate Star Codex →
Summary
A sharp, fair methodological takedown of a study widely reported as proving 'trigger warnings are harmful to college students.' Scott itemizes the problems: the participants averaged 37 (not college students); the study EXCLUDED people with trauma (the intended beneficiaries) and only tested for nonzero harm, not net effect; results are statistically weak (p~0.05, no multiple-testing correction). But the fundamental flaw is that it measures nothing we care about -- participants reported no extra anxiety, just slightly more endorsement of the abstract belief that trauma is harmful (his 'TRIGGER WARNING: SOME SNAKES BEAR LIVE YOUNG' analogy: telling people a thing makes them likelier to endorse it). For real harm, two further things would need to hold: the split-second belief-shift would have to persist (his own AI-essay experiment showed such effects vanish within two years) AND the belief would have to actually cause traumatizability ('it can't be voodoo all the way down'). He names the general failure mode -- 'streetlight psychology,' measuring an easy proxy for a superficially-similar but different question (cf. testing video-game violence minutes later) -- and, scrupulously even-handed, refuses to single out the right-wing misreporting (noting the left is 'light-years ahead' at weaponizing studies), landing on a sensible trigger-warning compromise (a matter-of-fact content list on the 'Boring Legalese Page,' no 'AND THEREFORE BE ANXIOUS' embellishment). NOTE: this critiques a shaky social-psych study rather than relying on one, so no replication caveat applies.
Why this score
Quality 73 · Strong. Strong (73): an excellent, clear, and fair methodological critique that both demolishes a widely-misreported study and articulates a transferable failure mode ('streetlight psychology' / measuring the wrong proxy), capped by a reasonable practical proposal. Upper-Strong rather than Excellent because it is a focused single-study takedown built on his prior 'streetlight psychology' coinage rather than a broad framework.
Claude’s paradigm shift 46 · Moderate. Moderate (46): the 'streetlight psychology' frame and the proxy-vs-real-question distinction are a genuinely useful contribution, though Scott had coined the term before and the underlying methodological points (demand effects, durability, multiple comparisons) are standard.
Real-world impact 2 · Minor. An excellent, fair methodological takedown that both demolishes a widely-misreported 'trigger warnings are harmful' study and names a transferable failure mode ('streetlight psychology' — measuring the wrong proxy). Conceptual/methodological influence within rationalist discourse, a focused single-study critique, no material change — low RWI.