Quantifying Hate Crime Hoaxes

The unraveling Jussie Smollet saga brings to mind this question, but how pervasive is the problem? Both 0% and 100% fake rate is a clearly absurd position to hold, but in between is much harder to ascertain. You first have to define what is a hate crime and that's a very slippery definition. The Southern Poverty Law Center tried to gauge the "Trump effect" on school bullying and (predictably) concluded bullying has increased. Can you guess what the two most common incidents were? Yep, "Build the wall" and talks of deportation. By far. The Anti-Defamation League tried to measure anti Semitic incidents but it ended up counting bomb threats hoaxes made by Jewish people, as well as cemetery vandalism made by clearly intoxicated individuals. The FBI is hopefully a more scrutinizing source, but even their report suffers from having an increased number of hate crimes reported as more law enforcement agencies participate.

Professor Brian Levin seems to be a frequently cited academic compiling hate crime statistics, including the false ones (Disclaimer: he still uses a aol.com email address and spends a breathakingly bizarre amount of space complaining about Russian election interference in his hate crime report).

I work as a public defender and deal with misdemeanor assaults quite often. By the time an arrest has been made, an exceptionally lucky chain of events had to have happened for the person to get caught. Either they were on video, a police officer was extremely nearby, or they were known to the victim. I don't know what the exact data is on clearance rates, but I would hazard a guess that random assaults happening on the street are comparable to property crime clearances which hover around 15%.

Crucially, a hate crime gets labeled as false only if it gets confirmed as false which is a somewhat remarkable hurdle to reach. If someone gets injured (say, walking home drunk) and then claims they were assaulted by bigots, there's literally no way to prove it as false unless there happened to be video nearby. The 0.3% false rate is therefore misleading and most likely severely undercounts the reality. A better statistic would be to count the number false crimes with the number of cleared crimes as a denominator. If 15% clearance rate is a reasonable starting point, false crimes could plausibly constitute 2% of all reported crimes.

So in essence, the issue suffers from a severe definition problem, a severe reporting inconsistency, and a severe lack of follow up. It's impossible to get any good statistical conclusion.