Tapsearch Com EditorU.S. leads world in incarcerations
December 10th 2006 22:08
This story piqued my interest....so I spent several hours doing statistical research on it. According to Reuters, the U.S. has the highest number of prisoners in the world due to "tough laws."
There's a lot of truth to that, as U.S. incarceration rates have risen and fallen with the drug war and mandatory sentencing laws. Now the U.S. towers over the other 27 "full democracies" (as decided by The Economist's Democracy Index). The U.S. rate is 738 for every 100,000 people; Uruguay comes in second at 193.
But the thing is, there is a very strong statistic in America toward minorities being more crime-inclined. Oddly enough, America has a higher incarceration rate than the countries these immigrants came from (first in the world, after all). Besides America being just plain tough -- which is true -- what could be going on here?
For readers who don't like statistics, the long and short of it is that diversity is linked to incarceration rates in the world's leading democracies. I believe that is due to the facts that (A) immigrants and other minorities tend to be poorer and (B) it's easier to be tough on crime when a significant minority will bear the burden; the smaller race is an "other."
This is not necessarily to say that diversity is a bad thing, just that it has this very negative effect.
I decided to do a correlation between diversity and incarceration rate in the world's full democracies. I kept it to full democracies because it doesn't tell us much that Rwanda, a failed state, has an incarceration rate of 152. It also doesn't help that China clocks in at 118. As a U.S. writer I'm interested in states most like the U.S.
I measured diversity in terms of ELF, or the Ethno-Lingustic Fragmentation index -- the odds that two random people from a country will be of different ethnicities or speak different languages. I used James D. Fearon's numbers when available (they are here, and you can open them with Stata or convert them to Excel with StatTransfer), but sometimes I had to go with Philip Roeder's (already in Excel; they're from 1985 but match up with Fearon's fairly well in the countries for which both have data).
I realize mixing and matching data isn't a good thing, but as Rumsfeld might say, "you do statistics with the numbers you have, not the numbers you want." If anyone has a full and recent set of ELF numbers, please send them to me (or better yet, calculate yourself!).
The incarceration rates came from the International Centre for Prison Studies, the group cited in the Reuters article.
Because the U.S. is such an outlier in terms of incarcerations (and because it's one of the few countries Fearon and Roeder have very different numbers for), I excluded it.
What I came up with (using this online calculator) was really quite striking. I had an R value of .4603, the square of which is .212 -- diversity explains 21 percent of variation in incarceration rates (that does not necessarily mean it causes 21 percent of incarcerations). It is quite statistically significant, with a P of less than .01596 (less than .05 is the benchmark). In a given country, the more likely two random people are from two different ethnicities, the more likely one of them is incarcerated.
Please bear in mind I'm not a statician. If you'd like to re-run the equations, shoot me an e-mail and I'll send my data (robertv4311[at]gmail.com). And always bear in mind that correlation is not causation; I above suggested some of the reasons diversity and incarceration might be linked, and I'm sure there are many, many more. It's not as simple as diversity causing prison terms.
Robert VerBruggen blogs at http://www.therationale.com and http://robertsrationale.blogspot.com.
UPDATE: I just found something very odd.
I e-mailed The Audacious Epigone about this post -- he actually has a clue when it comes to statistics. He suggested I look at criminality in addition to incarceration rates. So I did, using Wikipedia's list of most-recent-available rates.
I excluded the U.S. as an outlier (also because of the varying ELF numbers) again.
My theory was that if diversity and incarceration were linked more strongly than diversity and criminality were, there was something to my idea that diversity added to incarceration in ways other than bringing in crime-prone minorities. In diverse societies, tough-on-crime measures are easier to pass because the criminals are less like society at large.
I used the murder rate to measure criminality because it's best -- as Charles Murray pointed out, murders are almost always reported, so the data isn't affected by police priorities, failure to call law enforcement, etc. Presumably, sometimes murders fluctuate independently of other crimes, but it's the best measure available.
I was far more right than I'd imagined, assuming I'm measuring what I'm trying to. Diversity and murder are not even correlated -- my P value was <.6737, which translates to "holy crap is that statistically insignificant." Even if we ignored that, the R squared was .0072, meaning diversity explains .72 percent of the murder data. Negligible.
I was a little skeptical -- maybe the Wikipedia numbers were just bad. So I ran a correlation between incarceration and murder. Crime priorities vary by country, but the two should at least be linked in the world's best democracies. Here I got P<.01008 (very statistically significant), R squared .238. The murder rate explains about 24 percent of the incarceration data, which sounds about right. Other crimes can fluctuate differently than murder, and different countries take very different crime control steps.
So what does this say about the diversity debate? Well, for one, liberals can certainly argue there's something intolerant about these countries or even human nature.
One thing I've noticed from experience is that diversity often actually leads to intolerance. I grew up in a lily-white Wisconsin suburb (I could count my high school's minorities on two hands, maybe one), but I had some friends who'd moved up from more diverse cities. One once made a comment to the effect that, "I like it here. You can like minorities without having to be around them."
It was an unguarded moment, and he probably didn't completely mean it, but these numbers seem to show that's not uncommon.
UPDATE II: All the data is available on Google Spreadsheets here.
There's a lot of truth to that, as U.S. incarceration rates have risen and fallen with the drug war and mandatory sentencing laws. Now the U.S. towers over the other 27 "full democracies" (as decided by The Economist's Democracy Index). The U.S. rate is 738 for every 100,000 people; Uruguay comes in second at 193.
But the thing is, there is a very strong statistic in America toward minorities being more crime-inclined. Oddly enough, America has a higher incarceration rate than the countries these immigrants came from (first in the world, after all). Besides America being just plain tough -- which is true -- what could be going on here?
For readers who don't like statistics, the long and short of it is that diversity is linked to incarceration rates in the world's leading democracies. I believe that is due to the facts that (A) immigrants and other minorities tend to be poorer and (B) it's easier to be tough on crime when a significant minority will bear the burden; the smaller race is an "other."
This is not necessarily to say that diversity is a bad thing, just that it has this very negative effect.
I decided to do a correlation between diversity and incarceration rate in the world's full democracies. I kept it to full democracies because it doesn't tell us much that Rwanda, a failed state, has an incarceration rate of 152. It also doesn't help that China clocks in at 118. As a U.S. writer I'm interested in states most like the U.S.
I measured diversity in terms of ELF, or the Ethno-Lingustic Fragmentation index -- the odds that two random people from a country will be of different ethnicities or speak different languages. I used James D. Fearon's numbers when available (they are here, and you can open them with Stata or convert them to Excel with StatTransfer), but sometimes I had to go with Philip Roeder's (already in Excel; they're from 1985 but match up with Fearon's fairly well in the countries for which both have data).
I realize mixing and matching data isn't a good thing, but as Rumsfeld might say, "you do statistics with the numbers you have, not the numbers you want." If anyone has a full and recent set of ELF numbers, please send them to me (or better yet, calculate yourself!).
The incarceration rates came from the International Centre for Prison Studies, the group cited in the Reuters article.
Because the U.S. is such an outlier in terms of incarcerations (and because it's one of the few countries Fearon and Roeder have very different numbers for), I excluded it.
What I came up with (using this online calculator) was really quite striking. I had an R value of .4603, the square of which is .212 -- diversity explains 21 percent of variation in incarceration rates (that does not necessarily mean it causes 21 percent of incarcerations). It is quite statistically significant, with a P of less than .01596 (less than .05 is the benchmark). In a given country, the more likely two random people are from two different ethnicities, the more likely one of them is incarcerated.
Please bear in mind I'm not a statician. If you'd like to re-run the equations, shoot me an e-mail and I'll send my data (robertv4311[at]gmail.com). And always bear in mind that correlation is not causation; I above suggested some of the reasons diversity and incarceration might be linked, and I'm sure there are many, many more. It's not as simple as diversity causing prison terms.
Robert VerBruggen blogs at http://www.therationale.com and http://robertsrationale.blogspot.com.
UPDATE: I just found something very odd.
I e-mailed The Audacious Epigone about this post -- he actually has a clue when it comes to statistics. He suggested I look at criminality in addition to incarceration rates. So I did, using Wikipedia's list of most-recent-available rates.
I excluded the U.S. as an outlier (also because of the varying ELF numbers) again.
My theory was that if diversity and incarceration were linked more strongly than diversity and criminality were, there was something to my idea that diversity added to incarceration in ways other than bringing in crime-prone minorities. In diverse societies, tough-on-crime measures are easier to pass because the criminals are less like society at large.
I used the murder rate to measure criminality because it's best -- as Charles Murray pointed out, murders are almost always reported, so the data isn't affected by police priorities, failure to call law enforcement, etc. Presumably, sometimes murders fluctuate independently of other crimes, but it's the best measure available.
I was far more right than I'd imagined, assuming I'm measuring what I'm trying to. Diversity and murder are not even correlated -- my P value was <.6737, which translates to "holy crap is that statistically insignificant." Even if we ignored that, the R squared was .0072, meaning diversity explains .72 percent of the murder data. Negligible.
I was a little skeptical -- maybe the Wikipedia numbers were just bad. So I ran a correlation between incarceration and murder. Crime priorities vary by country, but the two should at least be linked in the world's best democracies. Here I got P<.01008 (very statistically significant), R squared .238. The murder rate explains about 24 percent of the incarceration data, which sounds about right. Other crimes can fluctuate differently than murder, and different countries take very different crime control steps.
So what does this say about the diversity debate? Well, for one, liberals can certainly argue there's something intolerant about these countries or even human nature.
One thing I've noticed from experience is that diversity often actually leads to intolerance. I grew up in a lily-white Wisconsin suburb (I could count my high school's minorities on two hands, maybe one), but I had some friends who'd moved up from more diverse cities. One once made a comment to the effect that, "I like it here. You can like minorities without having to be around them."
It was an unguarded moment, and he probably didn't completely mean it, but these numbers seem to show that's not uncommon.
UPDATE II: All the data is available on Google Spreadsheets here.
| 68 |
| Vote |
Add To: 
subscribe to this blog
