There is no statistical correlation between crime reported and actual crime. It’s one of the first thing you learn in statistics classes. It’s too easy to get people to not report, or change the category of a report, ie petty crime, etc. Or people not bothering because they don’t see the point of reporting, like graffiti, smashed car windows, pick-pocketing, etc.
What’s the implication? The city’s festering with crime even if the numbers say it isn’t?
Surely the Tories would be right in there emphasizing the prevalence of crime if the numbers supported it in any way. It’s been one of their most reliable scare tactics.
@Kate – What it basically means is that it’s a nice number to show the public, but essentially it doesn’t mean anything. Another way of showing it is to say… if no one officially complains about the Quebec government discriminating in it’s hiring practices, does it mean that it doesn’t? No, it just means that no one is complaining about it. There is no relationship between official complaints and actual discrimination.
I once has a policeman tell me to stop calling when I saw drug sales. Sure… because if you don’t report crime, crime statistics go down. Doesn’t mean that the crime has stopped, just that it isn’t being reported.
Don’t other cities experience the same problem of reported crime versus actual crime? When we compare statistics with other countries/cities, aren’t we assuming the existence of this flaw in data collection all across the board?
Highly unlikely there is NO correlation btw the reported and the actual crime rate. They might diverge for a number of reasons, even by a lot. But I seriously doubt there is no correlation. If the factors anchoring that correlation (btw actual and reported crime rate) do not change over time, the variations in reported crime rate will parallel the actual crime rate. So trends in the reported rate will reflect the actual rate.
Other countries may define crimes in different ways, and with different police priorities and legal systems there’s no reason to assume that the percentage of unreported crimes are the same for all crimes in all areas.
(Also, Aubin’s article seems to be “well, StatsCan says we have fewer crimes, but more severe ones, but really we’re pretty safe overall”. And I don’t see the crime severity index, so I don’t know how various crimes are stacking up there.)
Yes, the correlation exists, but not statistically significant enough. The problem is simply a matter of not being able to measure how much crime is not reported or is reported in different ways. A robbery could be listed in many ways depending on what is taken, how much is taken, is a personal crime involved. So, if you are there it’s not burgulary or break and enter it may be a home invasion, instead. What is someone on the police decides to classifiy it as burgulary even though there was a rape. Or someone breaks my window at home and I don’t see the point of reporting it because the insurance won’t pay, but if I do report it, insurance rates may go up. There are just too many ways to manipulate the data. And that’s the point. It’s not like the correlation between pregnancy and giving birth. It’s more like the correlation between getting stopped for speeding in Montreal and accidents in Laval. Yes, they are related, but is that really significant enough to say that the more speeding tickets that I give in Montreal the less accidents they will have on the road in Laval? Not really.
Ephraim 10:16 on 2012/07/31 Permalink
There is no statistical correlation between crime reported and actual crime. It’s one of the first thing you learn in statistics classes. It’s too easy to get people to not report, or change the category of a report, ie petty crime, etc. Or people not bothering because they don’t see the point of reporting, like graffiti, smashed car windows, pick-pocketing, etc.
Kate 10:25 on 2012/07/31 Permalink
What’s the implication? The city’s festering with crime even if the numbers say it isn’t?
Surely the Tories would be right in there emphasizing the prevalence of crime if the numbers supported it in any way. It’s been one of their most reliable scare tactics.
Ian 10:50 on 2012/07/31 Permalink
Ah yes, the famous “unreported crime” trope. I believe Rick Mercer covered that one best: http://youtu.be/bdtuBFtU1t8
Ephraim 11:48 on 2012/07/31 Permalink
@Kate – What it basically means is that it’s a nice number to show the public, but essentially it doesn’t mean anything. Another way of showing it is to say… if no one officially complains about the Quebec government discriminating in it’s hiring practices, does it mean that it doesn’t? No, it just means that no one is complaining about it. There is no relationship between official complaints and actual discrimination.
I once has a policeman tell me to stop calling when I saw drug sales. Sure… because if you don’t report crime, crime statistics go down. Doesn’t mean that the crime has stopped, just that it isn’t being reported.
Matt 13:43 on 2012/07/31 Permalink
Don’t other cities experience the same problem of reported crime versus actual crime? When we compare statistics with other countries/cities, aren’t we assuming the existence of this flaw in data collection all across the board?
Alex 14:19 on 2012/07/31 Permalink
Highly unlikely there is NO correlation btw the reported and the actual crime rate. They might diverge for a number of reasons, even by a lot. But I seriously doubt there is no correlation. If the factors anchoring that correlation (btw actual and reported crime rate) do not change over time, the variations in reported crime rate will parallel the actual crime rate. So trends in the reported rate will reflect the actual rate.
jeather 14:30 on 2012/07/31 Permalink
Other countries may define crimes in different ways, and with different police priorities and legal systems there’s no reason to assume that the percentage of unreported crimes are the same for all crimes in all areas.
(Also, Aubin’s article seems to be “well, StatsCan says we have fewer crimes, but more severe ones, but really we’re pretty safe overall”. And I don’t see the crime severity index, so I don’t know how various crimes are stacking up there.)
Chris 22:24 on 2012/07/31 Permalink
I agree with Alex. Ephraim, are you sure you took that statistics class? Correlation is not black vs white, it’s a matter of degree.
Ephraim 15:05 on 2012/08/01 Permalink
Yes, the correlation exists, but not statistically significant enough. The problem is simply a matter of not being able to measure how much crime is not reported or is reported in different ways. A robbery could be listed in many ways depending on what is taken, how much is taken, is a personal crime involved. So, if you are there it’s not burgulary or break and enter it may be a home invasion, instead. What is someone on the police decides to classifiy it as burgulary even though there was a rape. Or someone breaks my window at home and I don’t see the point of reporting it because the insurance won’t pay, but if I do report it, insurance rates may go up. There are just too many ways to manipulate the data. And that’s the point. It’s not like the correlation between pregnancy and giving birth. It’s more like the correlation between getting stopped for speeding in Montreal and accidents in Laval. Yes, they are related, but is that really significant enough to say that the more speeding tickets that I give in Montreal the less accidents they will have on the road in Laval? Not really.