Quote: Originally Posted by Colpy
Property crime is falling, violent crime is rising slightly......but Canada is still an amazingly peaceful kingdom.
This is a fairly commonly held misconception. Yet, from our own government:
It is true that television and the Internet are giving us access to a lot of this information. We are seeing a lot more of it, but data on crime shows the opposite. It shows that crime is reducing. I do not have to repeat too much of that. The data is out there. Since 1991, for reasons that sociologists have not ever been able to fully explain, our violent crime rates and our overall crime rates are decreasing and continue to do so.
My emphasis. It is true that in some districts police reported crime rates are increasing, but this is due to the self-evident fact that if you increase the number of police officers then you also increase the number of crimes that will be reported. It is also true that neighbourhoods which see more police presence will see higher rates of crime reporting and subsequently, convictions. It is for these reasons that the immigrant population in Toronto is so demonized, and this selection bias (external - login to view)
is sometimes used to ignorantly portray the immigrant population as innately more violent.
Quote: Originally Posted by IdRatherBeSkiing
No, I still think state by state comparisons are not valid. I stand corrected about NY however.
Looking at the chart, we can also say that the notible drop around 1995 could be tied to the adoption of the death penalty by NY and KS. You can interpret statistics any way you choose.
It is not true that statistics are amenable to any interpretation, quite the opposite, and that is exactly the common failing of frequentist statistics which necessitates bayesian statistics. In fact, the only thing frequentist statistics can do is to discount the null hypothesis, they can not be used to favour one theory over another. Bayesian statistics can. From the charts I posted, the only thing you can say is that the death penalty is correlated with a higher murder rate, why that is remains a mystery until some criminologists develop some fairly complicated predictors for murder rates and Bayesian statistics can then be used to favour one theory over another.
State by state comparisons are quite valid. It is equivelent to person by person comparisons when searching for say a correlation between gender and average height: a binary relation compared to a multivariate, continuous one.