You need a substantial period of time to prevent localized events from affecting the numbers. COVID measures started 2 years ago, which could be involved. The

You need a substantial period of time to prevent localized events from affecting the numbers. COVID measures started 2 years ago, which could be involved. The "defund the police" movement caught wind within that period, it could be that crimes are being over-reported. Analyzing crime rates over a couple of years is fraught with sampling bias. E.g. there was an 83% decrease in mass shootings between 2018 and 2020, does that mean those aren't a problem anymore?