A Government that really wanted to reduce crime, rather than one that was simply pandering to populism to get elected, would concerntrate on the well-known and understood causes of crime – the biggest of which are unemployment-induced poverty and deprivation.
When I challenged this theory I was directed to the following graph;
This is a really interesting representation. First let me show you what the crime line looks like when fully depicted;
+1999+to+2009.jpg)
Now, if you can imagine the unemployment measure on a right axis also starting from zero, the line depicting the unemployment rate would deviate a long way from the crime line. It would drop well below it. Here is a crude sketch;
Or let's put it another way. A forty five percent drop in unemployment, which delivered maybe a 10 percent drop in crime, does not show a strong association.
Return to their graph and consider something else. Using their scale, when the right-hand axis reaches zero - or full employment - the left-hand axis will still be showing about 78,000 crimes.
This "conversation" came out of a post they did today which was actually about violent crime and how it has risen under National. But it had been rising under Labour too. The following is violent recorded offences;
Now, one last thing. Imagine if the same treatment was given to this graph as occurred to the first. There would be no correlation whatsoever.
8 comments:
You might send this little gem to your slobbering mates at The Standard.
http://pajamasmedia.com/blog/recession-shatters-myth-of-poverty-causing-crime/
Now, if you can imagine the unemployment measure on a right axis also starting from zero, the line depicting the unemployment rate would deviate a long way from the crime line.
Why do you think that means anything? The units in which two factors are expressed aren't relevant here and there is no need to have line graphs include 0 on their axes (especially when on is per 100 and the other is per 10 000 !)
David, It allows quantification and comparison of the magnitude of change. What a graph with axes starting from zero shows is that there has been a small change in the crime rate and a large change in the unemployment rate.
David
clearly you work for Labour - or wish to.
Leaving zeros of charts is one of the most basic techniques in How to Lie with Statistics
We just had ten years of pain and finally a free election (somehow) to get rid of this sort of thing.
Lindsay,
What a graph with axes starting from zero shows is that there has been a small change in the crime rate and a large change in the unemployment rate.
That has nothing to do with how strong the correlation is. Moreover, the difference in crime is not small, the correlation identified predicts something like 7 000 crimes for every 1% change in employment.
Sinner,
It is important that charts that compare some value for several categories (bar charts in particular) include 0 on the y-axis. The Standard's plot isn't doing that - it's looking at a correlation (in fact, a scatter plot would have been a more effective display, but I guess that wanted to show the time series as well) so having is not important, and in fact it it preferable to truncate the axis so the variance in each time series can be properly displayed.
David, This is the most recent comment I left at The Standard reiterating my point;
Snoozer said, "If unemployment goes from 3.5% to 6.5%, the correlation identified by Marty suggests there will be an increase in crime from under 0.1 offences per person to 0.115, a 15% increase. That’s about 50,000-60,000 more offences across the country."
Actually, roughly the inverse did happen. Unemployment went from 7.2% to 3.4% with a drop of around 7% or 31,000 offences (1999 – 2005).
So what is accounting for the other 400,000 offences? You can show a correlation at the margin but that is not the same as what is claimed in this post ;
“A Government that really wanted to reduce crime, rather than one that was simply pandering to populism to get elected, would concerntrate on the well-known and understood causes of crime – the biggest of which are unemployment-induced poverty and deprivation.”
If the biggest cause of crime was unemployment induced poverty then when the unemployment rate dropped by over 50 percent there should have been a much larger drop in crime. Didn’t happen.
I am not disputing that unemployment is a factor in crime. I am disputing that it is as important as you, or the writer of the post, believe it to be.
And I dispute it only because if we continue pursuing policies based on wrong assumptions matters are not going to improve.
Also,
"Moreover, the difference in crime is not small, the correlation identified predicts something like 7 000 crimes for every 1% change in employment."
When there were 442,540 crimes (recorded offences) in the year to June 2009 I am afraid 7,000 is a small difference.
Unemployment, both in the U.S. and the world as a whole, marches ever higher because the field of economics doesn't account for the relationship between population density and per capita consumption.
Following the beating the field of economics took over the seeming failure of Malthus' theory, economists adamantly refuse to ever again consider the effects of population growth. If they did, they might come to understand that once an optimum population density is breached, further over-crowding begins to erode per capita consumption and, consequently, per capita employment.
And these effects of an excessive population density are actually imported when a nation like the U.S. attempts to trade freely with other nations much more densely populated - nations like China, Japan, Germany, Korea and a host of others. The result is an automatic trade deficit and loss of jobs - tantamount to economic suicide.
Using 2006 data, an in-depth analysis reveals that, of our top twenty per capita trade deficits in manufactured goods (the trade deficit divided by the population of the country in question), eighteen are with nations much more densely populated than our own. Even more revealing, if the nations of the world are divided equally around the median population density, the U.S. had a trade surplus in manufactured goods of $17 billion with the half of nations below the median population density. With the half above the median, we had a $480 billion deficit!
If you‘re interested in learning more about this important new economic theory, then I invite you to visit my web site at http://PeteMurphy.wordpress.com.
Pete Murphy
Author, "Five Short Blasts"
Post a Comment