Why data in context matters – On reporting about aggressive dogs

This morning the radio and television stations in Vancouver were broadcasting the results of an investigation by Estefania Duran from CKNW. Her story had the click-baiting title “What dog breed is behind the most attacks in Metro Vancouver? It’s not what you think” and presented the results of a investigation that included Freedom of Information requests from the various municipalities in the Lower Mainland of British Columbia.The outcome of her investigation was a graphic that showed that German Shepherds were the dogs most frequently declared by municipal governments as “dangerous” followed by Pit Bulls, Labrador Retrievers and Rottweilers. Wait you must be thinking, one of these things is not like the others? Yes, you read that right; according to this investigation Labs are the breed with the third most incidents standing behind only the Pit Bull and the German Shepherd. Now if this was the entire story then it wouldn’t be a big thing but this one data point is being expanded and stretched dangerously out of context to imply that Labs are an “aggressive” breed. You might ask: how could that be? and why would anyone do that? Well that is what this blog post is going to be about. How, when placed out of context, good data can result in the spread of bad conclusions and can ultimately lead to bad decision-making.

Let’s start with the potential conflicts of interest. I am the owner of a Labrador Retriever. We got our English Black Lab puppy six months ago and she has grown into a great big friendly lug. She joins us daily as I walk the kids to school. Only today, I was informed by a grandmother dropping off her grand-kids that our puppy is the third most dangerous dog out there. Since I hadn’t turned on my radio yet I had no clue what she was talking about and simply nodded but once I got plugged into the media I realized what she was talking about and that this was becoming a “big” news story, at least in our little part of the world.

Now our having a Lab is not a big surprise because Labs have been the most popular family dog in the US for over the last 22 years and in Canada for at least the last 20+ years  with the German  Shepherd coming in second on that list. On our street alone there are five Labs, one golden retriever and three other breeds that I now of. Labs make up 55% of the dogs on our street. So why is this important? Because were population stats included in the story it would change the entire dynamic of the discussions about this story.

Sadly, as Ms. Duran noted on the radio, none of the useful dog stat information is readily available so this next section will depend on info I have picked up where I could. According to a Georgia Strait story, in Burnaby pit bulls that made up only 2% of the licensed dogs in the city but were responsible for 24.7 % of the bites in which the breed could be identified. German Shepherds made up 5.4 % of all licensed dogs and 7.3 % of all dog bites. Anyone notice a slight difference there? Now to be clear I am not trying to be mean to Pit Bulls but by providing the numbers in context a different pattern appears. Per capita, German Shepherds look a bit less menacing but because of their higher numbers they end up with more total attacks.

As for Labs, well getting numbers of dogs by breed is a surprisingly hard thing to do online but a study showed that in Toronto there were approximately twice as many Labs as German Shepherds. If those numbers hold for Greater Vancouver, that would suggest that they might make up as much as 10% of the dog population and yet according to the report there are half the number of aggressive Labs as German Shepherds. When compared to Pit Bulls the numbers are even more telling: Labs are at least five times as popular as Pit Bulls but make up a smaller number of “aggressive” dogs than Pit Bulls.  See how putting the numbers into context makes all the difference?

So why you ask is Blair getting all annoyed by this story? Well my interest is in evidence-based decision-making and this story is already being used to advance/promote bad decision-making. If you go back to the online version of the story, who do they quote? A pit bull advocate; and what does the advocate say? That she “is hoping the new numbers will shift the conversation from breed to owner.” Other folks were also on the band-wagon, with the Georgia Straight running a similar story that once again highlighted the Labs while suggesting that we shouldn’t be banning Pit Bulls.

Now I am going to keep this post short as I just blew my lunch hour writing it and have to get back to my paying job. But I just want to make it clear that putting numbers into context makes a big difference for policy purposes. Since this report has come out I have seen it already used by supporters of Pit Bulls to discourage breed-specific bans. This story has already appeared on Global News and both online and on the radio at CKNW. As I noted, I have already had it mentioned to me on the street. I’m betting that by tonight I will be seeing it on the t.v. news and will hear about it more on the radio on my way home and I’m betting that few will point out the population demographics that underpin the story. To summarize, no Labs are not more aggressive than Rotweilers no matter what that out-of-context numbers appear to imply. There are just a whole lot more Labs out there than Rotweilers and a small percentage of a big population can often outnumber a large percentage of a small population. To conclude, any municipal politician who uses this data to pursue a decision about breed-specific dog bans is barking up the wrong tree.

Advertisements
This entry was posted in Uncategorized. Bookmark the permalink.

7 Responses to Why data in context matters – On reporting about aggressive dogs

  1. Legume says:

    Very good points here.

    We should also consider the integrity of the reported data. that not all reports of a bite necessarily equate to an aggressive dog / breed. Was the dog reacting to something? Was it scared or startled? What classifies the event as the result of aggressive behaviour?

    Also, the mis-classification of “pit bull” affects a lot here. A true pitbull is so rare nowadays that they would barely register on a national scale, let alone the Lower Mainland.

    Most reports of “pit bull” and “amstaffs” are mutts with primarily lab and boxer mixed in to give them “that pitbull look”.

    Like

    • rogercaiazza says:

      I agree with the issue of the integrity of the data points raised here. I would also add the possibility of the bias of the reporter. If you don’t like dogs or are afraid of them a big black lab bounding out with tail wagging madly in hopes of getting someone to pet him would probably count as an aggressive dog attack.

      Like

  2. Canman says:

    In my experience, black labs tend to be considerably louder than yellow labs. I also suspect black labs might be more popular in northern latitudes, since there coats should absorb more heat.

    Like

  3. chrism56 says:

    I suspect that there is flaws in the medical data collection as well, with all registered as dog bites, be they nips without puncture wounds or flesh torn off, needing major reconstructive surgery. Here in NZ, the breed that causes the most dog bites is fox terriers! All those pensioners must be in mortal danger from their hose dogs;-) It was only when pushed that the authorities admitted that in their eyes, a dog bite is a dog bite.

    Like

  4. Jeff Norman says:

    Sort of like saying 2016 was the warmest year on record without pointing out that record only goes back to 1979 (or whichever date you’d like to quote).

    Like

  5. Junior says:

    100% of Viszlas, in a sample of one, demand lay on their owners and demand scrunchies above their eyes. 0% are dangerous.

    Like

  6. Kevin Edge says:

    I heard the maddening reports on NW. It was obvious that neither the author or Simi had the most basic knowledge of statistics and that this report should never have been made in a public forum. Also reported was the aggressive nature of some small dogs, failing to mention the differences in pinprick injuries vs having your arm crushed or torn off…Simi was right onboard with erroneous conclusion that owners were the problem based on this “data”. Several yrs ago I read a stat that in the USA , ~75% of human deaths by canine were attributed to pit bulls or variations of this breed.
    If only every media outlet had a vetting process performed by an employee with a basic statistical math course…

    Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s