Avontus' crime theory still shaky

| | Comments (10) | TrackBacks (0)

A commenter on my “Bothered by crime? Protest the war” entry suggested I was manipulating the facts because:

  • my analysis looked at all crime, rather than only at property crime
  • my numbers focused only on crime within only 1 block of Avontus Software’ office, ignoring the possibility that employees are walking to work, going to lunch, or parking several blocks away
  • as of 2006, property crime rates in Berkeley exceeded that of Oakland or Richmond

The first two statements are true. The last one’s irrelevant, because I’m only addressing Avontus Software’s claim that high crime rates were linked to the Berkeley Police Department being tied up with protesters.

I re-ran my numbers based on these suggestions, looking only at property crimes within “three blocks” of Avontus Software’s office, as calculated by the indispensable Berkeley Crimelog — an area which in reality encompasses 43 actual human square blocks and about 10% of UC Berkeley, including four parking garages, almost a hundred restaurants, and almost the entirety of Berkeley’s downtown, arts, and transit centers.

The results? Avontus’ suggestion that property crime rates are going up because the police are distracted by anti-war protesters is still a misleading statement, given that January-March property crimes near Avontus Software’s office are actually at a four-year low (down 26% over the past two years), at precisely the time that Berkeley Police have supposedly been the most distracted by having to deal with the anti-war protests two blocks away.

0 TrackBacks

Listed below are links to blogs that reference this entry: Avontus' crime theory still shaky.

TrackBack URL for this entry: http://www.chatterjee.net/mt/mt-tb.cgi/21


You're still misrepresenting the data by focusing on a small area. There are just too many reasons why this is innacurate.

You can't make the assumption that people live in a bubble of a 3 block radius, it's not reality. It's feasible that you could run numbers in a 3 block radius elsewhere and show a tremendous increase.

Geographically, the area you're looking at comprises less than 1% of the city. It's statistically irrelevant. Also, downtown areas are variable throughout a 24 hour period since they're comprised of office spaces. This compounds your innacuracy.

Crime stats are always reported in comparison with other communities to eliminate exactly what you're trying to point out here, and several others. Stats in adjacent communities typically rise and fall at the same rate. A mere 'drop' in crime is not the point here. You must compare the entire city with others.

It would be interesting for you to get into the reason for why this community has stopped reporting the comparison with others in the area in thier public reports.

I know you want this to work, but it simply doesn't. When cities lose focus on crime (budget constraints, hiring problems, whatever) they have the highest probability of increased crime rate compared with other communities.

The fact that property crime in this community has a history of being higher than adjacent communities is a fact that simply cannot be ignored.

Hi Statgal. I'm not trying to suggest that crime rates (or specifically, property crime rates) are headed down across Berkeley, today, or in the future. Or that the city or BPD's potentially distraction may not have effects down the road.

Rather, I'm responding to the very specific assertion that the particular crime experienced by Avontus Software and the subsequent police response was substantially hampered by the BPD's needing to deal with anti-war protests two blocks away, and that this single data point somehow constitutes evidence of a broader trend.

The data's not there. Unless reporting rates are down, or the police are lying, property crime rates in the neighborhoods around Avontus Software office's have clearly decreased since the protests began. This is the case across the entire zone where typical downtown Berkeley office workers and their clients might reasonably be expected to spend their work-related time -- parking garages, restaurants, cafes, the post office, office supply stores, etc.

This decrease suggests that Avontus and its neighbors have reason to feel more, not less, safe. Local property crime rates were unambiguously higher when Avontus chose to move to downtown Berkeley in 2007, before the protests began.

Could that trend change? Is it possible that budget constraints, hiring problems, etc. will have negative impacts in the future? Absolutely. But possibly to BPD's credit, the significant predicted negative impact has simply not yet taken place for downtown businesses like Avontus.

This is waht I'm saying, why your limted set does not work. You cannot know what crime any individuals have experienced from looking at a generalized set of data. You're trying to correlate specific life experiences of an individual (or group of people), with another specific set of data. For your correlation to work you have to ensure that these people were subjected to the crimes that are in your set of data, which you cannot do.

But there are facts in the article that you can use for correlating your set of data. The police admit to not being able respond to issues while the protests are happening, which means these crimes aren't in your set of data. If you had to assume that a decrease was caused by anything, I'd say this is it.

Moreover, this correlation happens repeatedly in the article. The police sargent says twice that they have been unable to respond.

1. "police Sgt. Mary Kusmiss acknowledged that the increased manpower at the protests has sometimes taken away from other police roles"

2. "Property crimes have been historically our greatest challenge," Kusmiss said. "There have been days when the protests and demonstrations have had an impact on our ability to provide services to the community."

Further, their leaders even admit to their inability to service the members of their own community: "I don't see how the businesses in that square can sustain the kind of controversy"

The fact is, Berkeley has a history (as shown in their own data) of having higher crime than the surrounding communities. There's no way you can change that.

Yep, I think the drop you're showing indicates that the cops aren't responding to crime, just like they say they're not doing in the article. Can't record anything if they're not listening. Makes sense to me...

That crimelog thingy is pretty cool. But I wonder, dude, are you insane? Whether the crime goes down, up, mellow, whatever, there was like 100 incidents within this 3 block area in the last 30 DAYS!! Whoe in the hell would live in an area like that??? Holy christ, they need gun laws in Berkeley just to protect yourself.

When I first moved to California in the late 70’s, Berkely was still a lovely place to shop and walk through. Now it’s just a cesspool. I haven’t patrinized a business their in over ten years. Besides, Emeryville is the place to be now.

Why would you use 5 years of data when they've only been there a year? Bad use of data, sounds like you're masaging things to reach your own conclusion, just like the conservatives do.

Statgal, you're still fundamentally not getting me. You're trying to rebut me by saying:

(a) Looking at Avontus' neighborhood, the place where they work, is unfair; instead, we should look at unrelated places (where employees might live, bowl, pray, etc.?)

Avontus moved to downtown Berkeley from Oakland, the very next city over. It's incredibly unlikely that their employees moved or otherwise meaningfully changed their lives when their office made such an incredibly small move. We need to hold distant-from-office crime constant because there's no reason to believe that employee weren't exposed to the identical distant-from-office crime while Avontus was in Oakland.

(b) the BPD apologized for not doing as good a job as possible under certain circumstances

...and yet, though the BPD says it sometimes didn't do as good a job as possible, property crime rates near Avontus' office are still down. Maybe they could be lower. But they're certainly not higher, which is what was being suggested.

(c) the local city councilperson worried about the impact of the controversy of local businesses

You're throwing out totally irrelevant statements; how does this have to do with crime experienced by Avontus Software, and that likely to be experienced by local small area small businesses in the future?

(d) Berkeley has a lot of crime. I can't disprove that.

Duh. Again, totally irrelevant to the issue. Berkeley crime rates were also high before Avontus moved to the area, and the protests started.

"None" suggests that the BPD isn't recording crime stats. That's a fairly significant charge, and I'd like to see some evidence for that.

"Meaningless" says looking at five years of data is excessive and possibly misleading. That's fine. Just look at one year's data -- it proves the very same point. There were 247 property crimes near Avontus Software's office in January through March 2007, and 206 in the same period in 2008, when the protests were at their highest point. If property crime affecting small downtown businesses like Avontus Software is supposedly going up because of the protests, then why is it actually going down during the period when the protests have been at their peak? The numbers don't work.

Exactly. That's my point. The bare bones point of the article, this whole discussion from my point of view, is that the business is moving because of crime, you're trying to twist some sort of anti-war thing into it. I'm confused why you're questioning them, the guy even says he's anti-war (obviously, he's in Berkeley). You're trying to use the numbers to say that they not moving because of crime, yet you admit that the area has a lot of crime? The logical conclusion is that they're moving because they're sick of it and they're sick of their city not responding to it. I'd be pissed if my neighborhood had that much crime and they didn't do anything.

Oh my God! That's about 2.7 per day, and within 3 blocks of one location!? Who cares if it drops to 2.2 per day? Even at 1 per day it's horrible. I would move.

Indeed, it looks like we agree.

The original article's headline read "Local Company to Leave Berkeley Due to Protests," and that's precisely how the article was received by blogs commenting on it. I believe Avontus' CEO was right when he talked about the impact of bad press on his business, and about the less-than-optimal police response Avontus and/or its employees may have received after a crime. However, the suggested linkage between the protests and crime affecting businesses like Avontus is weak, based on a plain reading of local crime stats. (As you said, property crime was high before; as the numbers indicate, it's been marginally lower during the period that the police have been monitoring protesters.)

I don't like seeing plainly wrong things in the news. "Local Company to Leave Berkeley Due to High Crime" would have had a very different meaning (and reception) than "Local Company to Leave Berkeley Due to Protests."

Thank you, Statgal. It's nice to be able to refine positions and come to a consensus with strangers online.

Incidentally, property crime in Downtown Berkeley's high, but per Berkeley Crimelog, 60% of that is theft, which I'd guess disprortionately affects ground-level high-volume retail stores (e.g. Ross, Walgreens), rather than offices like Avontus', which is (or was) based in a high-rise office building with security guards and check-in procedures. I have a friend who used to work in the same building as Avontus, and another who's worked for years at the Morgan Stanley branch in the next building over, and neither's ever mentioned anything about crime.

Leave a comment


Anirvan Chatterjee is a San Francisco Bay Area tech geek and bibliophile.


Enter your email address:

About this Entry

This page contains a single entry by Anirvan Chatterjee published on April 9, 2008 11:44 PM.

Class in India was the previous entry in this blog.

The joy of CPAN is the next entry in this blog.

Find recent content on the main index or look in the archives to find all content.

Recently read