A Quick Check on Long-Term Violence Trends in the US

As Xander pointed out a few weeks back, the news reports outliers. Unfortunately we have cognitive biases--useful during our evolutionary days--that cause us to believe that because we happen to see something frequently, it happens all the time.

This is almost certainly exacerbated by the 24-hour news cycle and, particularly, social media. 

So let's look at how Americans feel, and then how the data looks

Americans think that violence in the country is getting worse. The national spotlight has also turned to two specific forms of violence: rape and police killings (both of police and by police).

How does the increased attention--and increased belief in frequency--stack up against the facts?

Violent Crime

Down 68% over 20 years. Just think about that. 

This comes with a significant increase in the total number of police in the country; violent crime has dropped so much that police now arrest more people for marijuana use than all violent crimes combined. (Imagine how bored they'd get if marijuana were legalized nationally.)


"In the space of one generation... the population-adjusted rate of rape has dropped by 55%." --Mother Jones

(The study MJ is citing attempts to take into account the likely increase in reporting during that time period.)

Shootings of Police

These have steadily dropped since 1970, almost without exception. Police are actually killed at a slightly lower rate than the average citizen (as we discuss in Wedged).

Shootings by Police

These peaked in about 1993--the height of violence crime in the US--bottomed out at about 2000, and increased slightly before leveling off. Again, this is as a rate by population, rather than an absolute number (which are never, ever helpful). 

So what was surprising here?

Are any of these very different from what you've believed to be true so far? Why do you think that might be? Does getting the data change how you feel about any of it? 

ERRATUM: This post previously stated that the above graph shows that police are killed at a rate of 100x the average citizen. While mathematically true, it is misleading: they're not actually killed by citizens at a rate of 100x compared to the rate that they kill citizens. See this correction post for details. 


Erik Fogg

We do politics, but we don't do the thinking for you.