Is Crime High on the Peninsula?

Surveys and public comment make it clear: one of the very top concerns in Los Altos is crime. Well, crime and parking.

I'm savvy enough not to wander into the parking argument. But let's talk about crime. How high is it, really? We're never going to eliminate crime entirely. And the lower it gets, the more it's going to cost us in dollars and civil liberties to get it even lower. 

If you just want the answer, skip down to "What does it all mean?" But this is a data blog! So we're going to talk about data!

What is this data, anyway?

Let's see if we can use some data to put our crime rates in context, so that we are all working from the same facts. If we don't have the same facts, we'll never come to a good policy.

Most of the crime that Los Altans are worried about is property crime. Break-ins, theft, catalytic converters being stolen out of our driveways, and for some reason, gangs of South American organized criminals. Violent crime is less of a concern, mainly because it is blissfully low. White-collar crime is... well, that's a subject for a Ramble sometime.

Using CityProtect data, we can look at all the incidents reported by the Los Altos Police Department (LAPD, but not that one). CityProtect also has data for a few cities nearby: Saratoga, Cupertino, and Los Altos Hills through the Santa Clara County Sheriff's department*, and Union City through their police department.

In that data, there are a zillion different IncidentTypes, which refer to a detailed kind of incident -- for example, "Abandoned Vehicle", "Call for Help", or "Bomb Threat." (Yes, it's an interesting data set.) But there are high-level categories as well that will help us look at property crime. The high-level categories that seem like property crime are:

• Breaking and Entering. My understanding is that this is when someone forces a lock, breaks a window, or by some other way gets into a dwelling or business and then steals something. If you left your door unlocked and they walk in, that doesn't count -- see below.
• Theft. This is when someone steals something without forcing their way in. You are distracted by your phone and they take your briefcase. You leave your car unlocked and they steal your laptop.
• Vehicle Theft. This is a special category, just for stealing vehicles. I don't know if catalytic converter theft fits here, or somewhere else.
• Property Crime. There a three kinds of incidents that show up under this heading: grand theft, which is theft of something worth a lot of money; vandalism; and "STOLEN VEHICLE; GRAND THEFT". What the heck is that? Well, for one it's a good example of how hard it is to clean this data. But maybe someone had their car stolen with diamonds inside, and the officer didn't know how to categorize it.

For context, I also found data on per capita property crime rates for California and the Bay Area, and the US as a whole.**

As with all data analysis, we have to worry about how clean our data is. The state and national data claims to include all reported and prosecuted instances. The CityProtect data appears to include all reports of property crime when the police are called. However, that data is only as good as the officers reporting it. If no officer ever comes to the scene, it's not clear if that incident is in the dataset or not.

So, what does it all mean?

First of all, it's clear that property crime in Los Altos is lower than the Bay Area average, the state average, and the national average. By any measure, our community is safer than average. The communities in our data set that have even lower property crime rates are Los Altos Hills and Saratoga. Saratoga has half the population density by area; Los Altos Hills has a quarter. Perhaps that lower density leads to fewer property crimes.

But the bottom line is, when you put Los Altos' property crime in context with the rest of our area, state, and country, it's well safer than average. We are already doing well on this front. 

So why are people so concerned about property crime in Los Altos? I think there are two reasons.

• There have been a few high-profile commercial burglaries. These burglaries get a lot of press -- everyone knows that business, it's shocking, and so it gets a lot of press. People extrapolate from what they see in the press, and so if there are a lot of stories about crime, crime must be really high.
• Human nature prioritizes fear over possibility. It's pretty easy to stoke people's fears, and fear of being a crime victim is no exception. Media reports, NextDoor, even the police talking about crime makes people afraid -- even if the numbers show our town is safe.

Often, people we respond to these numbers by saying, "Well, sure, but have you ever been robbed? How would you feel if it were your house?" Of course it feels awful to be a victim of crime, and of course it would be great if we could snap our fingers and make it go away. But we have to also recognize that we cannot make our world risk-free. We can't live in bubbles, or in armored castles, and we wouldn't even want to. 

So we're going to have to make public policy choices about how to prioritize our response to lots of different risks. We'll have different legitimate opinions about that, which is great. But we have to start from the same set of facts. The fact: Los Altos is a very safe community.

Next time: "Sure, maybe crime is low now, but it's rising!" Let's see if that's true.

*CityProtect data for the Santa Clara County Sheriff's Department requires some extra analysis. Those records usually include a street and block number, but not the city in which the incident took place. To figure out the city, we had to use a reverse geolocation service. That worked pretty well, but not perfectly -- another source of noise in our data.

**Whenever you combine data from different sources, you have to worry that they aren't actually measuring the same thing. It seems like the definitions of property crimes used in the US, California, and Bay Area bars are pretty close, if not exactly, what we chose from CityProtect, but a good semester-long project for an undergraduate student would be to dig out the actual definitions from each data source, and then characterize the impact of differences as compared to the expected errors elsewhere (for example, data entry issues). At this point, I do not have any undergraduate students.