Driving to the grocery store yesterday my boyfriend points to the car in front of me (a Honda Civic) and says “that’s one of the two most stolen cars in the country”. My immediate response was to think that this must mean these cars are the more easily broken into; there must be a reason why they are stolen more than any other car. But that’s not necessarily true. With the one piece of information I was given and nothing else, I am not actually able to know much more than I knew before I was given the statistic.
Most of us know the saying: “90% (or whatever number you choose!) of statistics are made up.” My concern is not with the statistics that are made up, but with the ones that hold some truth. Take the car thefts for example. Assuming it’s true, I have no idea how that statistic came about. I can guess that it is from a list compiled of vehicles reported stolen or missing in the USA over the period of a year. But then we aren’t left with much information. Yes, that particular car brand was one of the two most stolen cars in the country BUT maybe it was also the most popular car. Maybe if the total of each type of car owned was taken into account then this would not be the most likely car to be stolen.
What I mean is: if there are 100,000 Honda Civics and 1000 are stolen that’s 1% of all Honda Civics. And if there are only 1000 Mercedes and 500 are stolen then that’s 50% of all Mercedes. But, our original statistic would not represent this fact. It only states that more Honda Civics are stolen than Mercedes, which is true numerically but taking into account proportions it is way off.
This concept of statistical deception is one that is popular in the media, and is one that bothers me often, particularly with regards to public health studies as they are so frequently taken out of context and end up causing hysteria or creating new fads based on statistics that weren’t fully explained.
The most recent article I found was about beer in Baltimore: http://well.blogs.nytimes.com/2013/08/19/beers-implicated-in-emergency-room-visits/?smid=fb-share&_r=1
The argument is that out of all emergency room visits involving beer there were 5 brands that were most commonly implicated. So, what is your immediate response? Those 5 beers must be worse for you in some way. That’s the obvious explanation, the article cites some of those beers being malt beers as the reason. But is it?
Firstly, certain beers are associated with different people. Some of those beers are associated with fraternity parties, in which case I’m not sure the type of beer is the only thing to blame for people ending up in the emergency room.
Secondly, some of those beers are cheap beers. People looking to get drunk are often going to get the cheapest beer with a high alcohol content so as to be able to get drunk fastest and cheapest. People who are looking to get drunk are also more likely to end up in the ER.
And thirdly, this particular article on the study makes no note as to whether the people who drank these beers and ended up in the ER had been drinking other things too (like liquor), or if they had used any recreational drugs. All things that in combination could lead to a person being much more likely to visit an emergency room.
The most interesting thing about this is that when I clicked on the link to the study I discovered it was a pilot study: only 105 people interviewed. This is not enough people to get an accurate representation of anything like this in any town or city.
The point is that anyone can take real, accurate statistics and portray them in a way that makes you believe a certain thing if you take the statistics at face value. Whether or not I believe global warming is a real phenomenon, I could easily take real statistics from real studies and make an argument on whichever side of the debate I choose to. It’s important to be aware of how easy it is to do this.
So, my point? The statistics we read that “prove” certain theories and argument: take them with more than just a grain of salt. They might be true, but make sure you understand where they come from before you accept them as such.
EDIT: For anyone interested in the article about stolen cars, I have just found it here: http://www.autoblog.com/2013/08/20/honda-accord-civic-named-most-stolen-cars/