The National Academies presented a 95% lockdown-drill figure as active shooter drills

Counting drills works only if the thing being counted has fixed contents. The National Academies report that carried the 95% figure also found a lack of standardization across schools for drills and preparedness in general ^[15]. A survey can therefore collect a near-universal yes to "do you hold drills" while the drills differ in whether a child hears a gunshot. The same report found the lack of standardization has made it nearly impossible to say what is harmful and what is helpful ^[23].

Lockdowns are not specific to shooters either. They get called for parent disputes that turn physical, for nearby police activity and for wildlife ^[16]. The article's author reports that a student described her only elementary school lockdown as a response to a squirrel that had got into the building ^[17].

Among surveyed K-12 teachers, 55% said the drills practiced in their schools included no sensory elements such as fake blood ^[7], and 1% reported fake blood or explosions ^[8]. Those shares and the simulated-gunfire share sum to 62, and the remaining 38 percentage points of respondents sit outside the reported categories ^[21]. The categories can also overlap, since a drill with fake blood plausibly includes mock gunfire. And the denominator is teachers, so the numbers say how many adults have sat through a format, not how many buildings run one.

The barrier to measuring effectiveness is the base rate ^[12]. With attacks this rare, no feasible sample of schools accumulates enough real incidents to compare a simulation against locks, lights and out of sight, so money and time would not close that gap. What the literature does report is significant concern about psychological harm to students and teachers from high-intensity simulations ^[11], alongside mixed results on the effectiveness of both formats ^[19].

State law is where the two formats separate. Subtract the 37 states with an annual drill requirement from 50 and 13 are left with none ^[20]. New York schedules lockdown drills at several points in the year ^[3], and both New York and California restrict high-intensity simulations ^[9]. Texas requires lockdown and shelter-in-place drills while permitting active shooter threat exercises, simulations included ^[4]. Many states have no simulation guidelines at all ^[10].

So a Texas district can satisfy the requirement with a lockdown drill that involves locking doors, cutting lights and moving out of sight lines until an all-clear ^[5], and separately choose to stage actors with paintball guns ^[6]. In the record here, the two choices sit in different provisions, and the harm evidence is attached to only one of them ^[11].