Accelerating Future

Better All the Time? Wednesday, Feb 10 2010 

random Michael Anissimov 10:50 pm

Trackback URI

16 Responses to “Better All the Time?”

1. Michael Vassar says:

   February 10, 2010 at 11:21 pm

   Someone tell Mencius; the following is much more relevant and much more objective data, as it doesn’t just reflect a less free society or a higher rate of prosecution. Murder rate estimates have surely been much more reliable across time.

   It would be nice to have an ethnic breakdown of perpetrators and of victims too. If high murder rate demographics have immigrated, have lowered their murder rates, but have continued to murder one another but not British natives more than the British baseline murder one-another this increase in the British murder rate is win-win.

   Homicides per million population in England and Wales
   Year Homicides
   1900 9.6
   1910 8.1
   1920 8.3
   1930 7.5
   1940 ..
   1950 7.9
   1955 6.3
   1960 6.2
   1965 6.8
   1970 8.1
   1975 10.3
   1980 12.5
   1985 12.5
   1990 13.1
   1995 14.5
   1997 14.1

   Reply
2. TheMediumDog says:

   February 11, 2010 at 12:02 am

   A graph like this divorced from a narrative of political history doesn’t mean anything.

   Reply
3. Russell Coker says:

   February 11, 2010 at 12:20 am

   I agree with TheMediumDog. There have been changes to the extent of the law (EG more laws against drug use) and significant changes to the ability to detect crime.

   Michael Vassar says that homicide rates are more reliable, however recent scientific advances have made it possible to identify more murder victims. Many cases that in the past would have been regarded as suicide, accidental death, or disease are now identified as homicides.

   Reply
4. Joshua Fox says:

   February 11, 2010 at 2:07 am

   Yes, the increase shown in the graph is so extreme that it calls for some explanation beyond simply “crime increased.”

   Reply
5. Russell Coker says:

   February 11, 2010 at 3:00 am

   http://www.ted.com/talks/steven_pinker_on_the_myth_of_violence.html

   Steven Pinker gave an interesting TED talk about the reduction in violence as a general trend through history. I don’t think that an increase in the number of convictions is an indication against that trend.

   Reply
6. MZ says:

   February 11, 2010 at 5:09 am

   When there are more laws to break, people will commit more indictable offenses. In 1900, there were no drug crimes, no driving-related crimes (DUIs, etc), no internet and communications crimes, hardly any banking crime other than robberies, essentially zero crimes related to consumer fraud (other than theft). You can look at these increases in indictable offenses and think that we’re less free, or you can look at them and think that there’s less chaos.

   Reply
7. Michael Anissimov says:

   February 11, 2010 at 12:07 pm

   A graph like this divorced from a narrative of political history doesn’t mean anything.

   Really, nothing at all?

   It’s good to look at the data prior to looking for theories to explain the data, for two reasons. The first is that the current theories might not actually explain the data. The second is to debias against framing effects.

   Reply
8. Bob Mottram says:

   February 11, 2010 at 2:53 pm

   Interesting. The trend looks exponential until the mid 1990s. I wonder what happened then. One hypothesis is that the Blair regime was successful at bringing down crime. Another, probably more likely, is that the way crime was recorded was changed which made the figures look lower. This sort of creative accounting is very much in tune with the way that the post 1997 labour administration operates (they eliminated child poverty in the same way, just by reclassifying things).

   Reply
9. Smifly says:

   February 11, 2010 at 7:12 pm

   I think that the downturn is at least partly because they started to reclassify some killing as self defense/justified, for example if a police officer killed someone.

   Reply
10. TheMediumDog says:

    February 11, 2010 at 10:34 pm

    “Really, nothing at all?”.

    Well obviously not NOTHING in an information-theoretic sense. Principle of charity in interpretation, please.

    Thing is, have a read of this article (http://www.monthlyreview.org/090601holleman-mcchesney-foster-jonna.php), which displays a graph (chart 3) with a very similar shape, over the same years, for the U.S (and with regards prison population as opposed to offenses). And consider the explanation given therein.

    The point is that your graph occurs under the heading “Better all the time?”, as though it presents some kind of potential falsification of a thesis about ‘progress’. It’s too simplistic.

    I’m mixing up points here, about information-content, about statistics, and about explanation… But there we are…

    Reply
11. mitchell porter says:

    February 12, 2010 at 2:14 am

    If you follow the link below the graph, and then look at sources (page 31), you will find that 1900-1980 comes from a 1988 Cambridge University publication called “British historical statistics”. I’ll look it up tomorrow; perhaps it will tell us more about what’s going on.

    Reply
12. Shane Legg says:

    February 12, 2010 at 3:04 am

    Continuing from Vassar’s numbers, the latest results for 2008-2009 put the rate at 12, the lowest in 20 years.

    In comparison the current US rate is about 56.

    Reply
13. Ryan says:

    February 12, 2010 at 7:54 am

    Micheal said,

    “It’s good to look at the data prior to looking for theories to explain the data, for two reasons. The first is that the current theories might not actually explain the data. The second is to debias against framing effects.”

    I agree, framing is a big issue to consider when looking at and generating data. A almost comical example of this is here:

    http://politics.theatlantic.com/2010/02/want_to_serve_be_gay_or_lesbian_dont_be_a_homosexual.php

    Reply
14. Steve Burrows says:

    February 12, 2010 at 9:22 am

    Be interesting to compare this graph with one showing the number of law enforcement personnel per thousand population.

    I would hazard a guess that the graphs would look remarkably similar.

    Reply
15. mitchell porter says:

    February 12, 2010 at 8:05 pm

    I now have before me a copy of British Historical Statistics by Brian R. Mitchell. The data to 1980 comes from the chapter “Miscellaneous Statistics” (omitted from the Google Books preview), first table. Its figures in turn come from annual government publications.

    There are three categories of crime tabulated. In the three categories, the figures for 1900 and 1979 are: crimes against the person, 3500, 117000; crimes against property with violence (burglary and robbery), 7800, 562000; crimes against property without violence (other offences against property), 64000, 1695000. (A footnote says reporting methodologies changed in 1980 so I have used 1979 figures instead, but the figures for 1980 are very similar.) The increase in frequency for each category is therefore 33x, 72x, 26x.

    Reply
16. mitchell porter says:

    February 12, 2010 at 8:35 pm

    From Michael’s link (“A Century of Change”) I can see that during this period, population in England and Wales increased by half (32m to 49m, see page 6) and police numbers tripled (40k to 120k, see page 14).

    Reply