Every objection below is one we think is serious. They are stated here in their strongest form, before our response, because an objections page that argues with a weakened version of the objection isn't worth reading. If you only read one section, read the fourth.
The objection. Arrests and referrals to law enforcement are the incident categories with the best-documented racial disparities in American schools. Two schools with identical student behavior can produce very different numbers depending on whether administrators call police. A score built on those counts partly measures how a school polices its students, not how safe it is — and it will systematically disadvantage schools serving Black and disabled students, who are referred and arrested at higher rates.
What we do about it. Suspensions, expulsions and corporal punishment — the most discretionary responses — are excluded from the safety score entirely. Every school page shows the severity mix behind its score, so a score driven by arrests is visibly different from one driven by weapons or assaults. What we do not claim is that this solves the problem. It doesn't. A reader comparing two schools should look at what kinds of incidents each reported, not only at the number.
The objection. Schools self-report to CRDC once every two years, with no audit. A large share report implausibly little. Any ranking built on that data mostly ranks reporting diligence, and quietly rewards the schools that disclose least.
What we do about it. This is the objection the site is built around. We compare each school's reporting to ~1,500 schools of its size and grade range nationally, and refuse to publish a safety score for any school whose record doesn't support one — about 42% of them. The share of schools reporting nothing ranges from 13% in Georgia to 96% in Puerto Rico, and we deliberately do not adjust for state, because doing so would define each state's reporting culture as its own normal. See the methodology for the exact test.
The objection. The bar you compare schools against is itself built from self-reported data. If most schools under-report, the bar is too low, and a school that under-reports moderately sails past it with an A.
This is correct and we don't have a fix. The comparison is circular by construction, so it is biased toward leniency: we only flag records that are implausible even by the permissive standard of what schools actually file. We under-flag on purpose. An A means "this record is consistent with its peers," not "this record is complete and true." Read it that way.
The objection. Publishing school-level safety comparisons can accelerate the decline of schools already struggling — families with the means to leave do, enrollment and funding follow, and the ranking becomes self-fulfilling. The families least able to act on the information bear the cost of it being public. Several major real-estate platforms removed crime and safety layers for versions of this reason.
Our answer, and its limits. This data is already public and already federally mandated; what was missing was a comparison. A raw incident count with nothing to compare it to is what actually misleads a parent — it looks like a fact and behaves like an insinuation. We think the honest response to "this could be misused" is to build against misuse: no score for thin records, denominators shown everywhere, no "worst schools" page, no predictive claims, and a published method anyone can check.
That is a position, not a proof. If you think a school's page here is wrong or unfair, we would rather hear it than not.