Friday, September 18, 2009

Atlanta crime data: Larceny from Auto

This is the second in a series of posts looking at APD crime data. Previously I looked at residential burglaries. For an overview, including methodology, see Atlanta crime data: project overview.

Looking at the the Larceny from Auto (car break-ins) data is rather different than the home burglary story. With home burglaries, we saw rather localized results, where it was possible to identify areas of the city that saw disproportionate increases. With car break-ins, we basically have several major problem areas in the city for all years of our data set.

For the city as a whole, car break-ins are up. Things started off well, as car break-ins actually dropped in 2005, but picked up again in 2006. 2008 was a particularly bad year.

Overall, I can't see significant trends in the data other than the fact that 2008 was a very bad year for car break-ins. Without the 2008 series, many of the NPUs have average annual increases in the single digits, and a few NPUs actually have fewer car break-ins. So for, the first half of 2009 is showing a decrease in major problems areas, as well.

Don't park your car in Buckhead, Midtown, or Downtown

Breaking down car break-ins by NPU very quickly shows that the it is dangerous to park your car in densest areas of the city with the greatest amount of on-street parking and open-air parking lots. Combined, NPUs B, E, and M account for 54.83% of all break-ins over the period studied.

What you can also see here is that the trends aren't as strong for car break-ins as they were for home burglaries. For a lot of NPUs, the numbers of incidents are relatively level. The biggest increases are seen in 2006 and 2008, but in different areas. All of the increase in car break-ins in Buckhead (NPU B) occurred in 2008, while 2006 was a bad year for Midtown (NPU E). Midtown, incidentally, actually had a net decrease in activity because the increases in 2006 were offset by large decreases in 2005 and 2007.

Other areas with significant car break-in issues include NPU F (including Virginia-Highland), NPU N (including Inman Park, LIttle Five, and Cabbagetown), NPU W (including East Atlanta), as well as NPUs T and V (Summerhill, Mechanicsville, Peoplestown, Capitol View). For the most part, these problem areas correlate to areas with lots of nighttime activity. For crying out loud, don't leave stuff in your car when you go out at night!

East side again a problem area

The concentration of car break-in activity also means is that relatively modest increases in activity in these NPUs gets magnified. When car break-ins increase by 30% in Downtown over five years, it accounts for 24% of the increase in all break-ins. By contrast, care break-ins in NPU P increased by 156% yet that only contributed to 4.6% of the overall increase in activity.

Compare this graph, showing just the additional incidents by each NPU, with the second graph, showing the overall percentage change in car break-ins for each NPU.

Several things jump out to me. First, while downtown accounts for a lot of the increase in burglaries, Midtown and Buckhead are less culpable. NPUs N and O show both significant percentage and absolute increase in activity. Almost all of the increases for N and O occur in 2008.


So far, most of the major problem areas are all showing decreases in car break-ins for the first six months of the year compared with the first six months of 2008. Buckhead, Midtown, Downtown, East Atlanta, and Little Five NPUs are all down, while Virginia Highland is up. NPU O, which includes Edgewood and Kirkwood, has really taken off, though. Car break-ins have almost doubled there - no surprise given the recent reports there.


  1. I frequent the heart of Va Hi a lot. Los Angeles next to Highland Hardware is constantly covered in shattered glass. I'm told they have a special police officer just for car break-ins who will meet you at the fire station to file a report.

  2. Half the images aren't coming in for me (could be my net connection, Google, or both), so I can't read all the graphs. So apologies if this is a dumb question.

    With higher density, isn't it natural for there to be a higher number of break-ins going by sheer numbers alone? Is there any sort of per-capita data here?

  3. Rusty - You are absolutely right that density is a big factor here, but it depends on what you mean by "density". Downtown is relatively sparsely populated, but has huge numbers of visitors from other parts of the city, metro region, and state on a daily basis. Regular "per capita" numbers based on residents wouldn't be accurate. Likewise, I suspect areas like Virginia Highlands and East Atlanta see large numbers of nighttime visitors.

    I don't think Per Capita is the right driver for trying to compare relative problem areas. Yes, obviously Midtown and Downtown are going to have higher absolute numbers. But what should an "average" level be for those areas, if we have established that per capita is a bad way to standardize the numbers? I am not really sure, so please feel free to brainstorm with me on this.

    Also, I think that is why the percentage increases is helpful, since it takes out the magnitude issue. This, for me, is what makes Inman Park/Little Five and Kirkwood (NPUs N and O) really stand out - combined with the fact that there are already significant levels of incidents there, the percentage increase is important.


Note: Only a member of this blog may post a comment.