What FBI crime statistics tell us — and what they don't

What FBI crime statistics tell us — and what they don't
© Getty Images

On July 4, 2002, a lone gunman opened fire at the airline ticket counter of El Al, Israel’s national airline, at Los Angeles International Airport (LAX). Two people were killed and four others injured before the gunman was fatally shot by an El Al security guard.

As a special agent assigned to the FBI’s violent crime squad in Los Angeles, I was called into action. Weeks later, after an intense pursuit of the facts, especially regarding the motivation of the gunman, I took part in a roundtable discussion, the purpose of which was to recommend how the FBI should classify the incident.

Was it a hate crime in which a lone, anti-Semitic gunman opened fire on a group of Jews heading to Israel? Or, was it an act of political terror committed by a radical Muslim against one of Israel’s most widely known symbols of strength, its normally impenetrable national airline?

ADVERTISEMENT

Israel lobbied for the latter.

So goes one of the many quandaries when engaging in the often complicated task of classifying crime.

With the FBI announcement this week that the number of reported hate crime incidents in the U.S. decreased slightly from 2017 to 2018, and media pundits everywhere touting the decrease through their political lens, it is important for the public to understand what crime statistics tell us — and what they may not tell us.

While I believe the LAX shooting was properly classified as an act of terror, it is easy to see how classification bias can play a role. There are many reasons, political and otherwise, why certain crimes may not be reported as hate crimes when they empirically or arguably are. With over 16,000 entities reporting hate crimes in the FBI’s Uniform Crime Reporting System, the possibility of classification bias potentially can be significant.

Moreover, as any statistician will tell you, there are two critically important factors when evaluating the meaning and relevance of statistics if they are to be actionable: trends and causation. While the decrease in reported hate crime incidents is potentially encouraging, the reduction appears negligible and, in any case, would be meaningful only if the trend continued over a longer period. 

ADVERTISEMENT

Two consecutive years is not enough.

Perhaps more importantly, without using math and data to identify and test why any trend in American society has occurred, attributing it to politics, policy or celebrity influence such as the Kardashians is just huff and puff.

If you doubt this, perhaps James Franklin, the Penn State head football coach, is more believable.

This week, after the Nittany Lions’ loss to Minnesota in a top 20 matchup with college football playoff implications, Franklin was asked to comment on the fact that his team is ranked 10th out of 14 in pass defense in the Big Ten Conference. Franklin’s response was telling: “I think you have to be careful with statistics. … A lot of times the guy that leads the conference in tackles … they are on a team where the defense is on the field the most. It’s not necessarily that … they’re the best [player].”

A reduction in hate crimes is encouraging if it is statistically significant. The reduction itself, however, as James Franklin knows, may not mean what you think it means.

James S. Davidson was an FBI special agent for 23 years. He investigated major crimes in Texas and California, and served in Ukraine, Israel and Washington, D.C. He is now president of Protect the FBI, a non-partisan 501(c)(4) whose mission is to safeguard the FBI from partisan politics. Follow him on Twitter @jamessdavidson2.