Thursday, February 21, 2013

Incorrect statistics portray Americans and Canadians as more prone to criminality

From incorrectly calculated foreign crime rates to tabloid TV

Part 1: Incorrect statistics portray Americans and Canadians as more prone to criminality
Part 2: Yonhap reports on the KIC foreign crime study
Part 3: Joongang Ilbo: "Get a Korean woman pregnant": Shock over manual for foreign men
Part 4: JTBC's "We are Detectives" looks at foreign crime using the KIC report

Part 1: Incorrect statistics portray Americans and Canadians as more prone to criminality

So yesterday it was noted at the Marmot's Hole that the Herald Gyeongje had reported on a study by the Korean Institute of Criminology:
According to the study, entitled "Research into Crime and Public Safety in Areas with a High Concentration of Foreigners," the Mongolian community had 7,064 criminals per 100,000 Mongolians in 2011, followed by the American community (6,756 per 100,000), Canadians (4,124 per 100,000) and Russians (3,785 per 100,000). These rates were higher than both the general foreigner rate of criminality of 2,763 per 100,000 and the Korean rate of criminality of 3,692 per 100,000.
Now, having stats from 2010 on foreign criminality in Korea from the Supreme Prosecutors Office, I knew that these stats weren't right. Here are statistics from the 2010 report:

Nationality / Total population (A) / Number of charges (B) / Crime rate (B/A)
Overall Crimes in South Korea / 47,990,761 / 1,816,223 / 3.8%
China / 608,881 / 16,540 / 2.7%
Vietnam / 103,306 / 2,733 / 2.6%
US /  127,140 / 2,046 / 1.6%
Mongolia /  29,920 / 1,803 / 6.0%
Thailand /  44,250 / 1,636 / 3.7%
Uzbekistan /  25,895 / 666 / 2.6%
Philippines /  47,241 / 613 / 1.3%
Sri Lanka /  18,377 / 573 / 3.1%
Taiwan /  24,760 / 480 / 1.9%
Canada /  20,435 / 401 / 2.0%
Others /  211,210 / 10,464 / 5.0%
All non-Korean nationals / 1,261,415 / 33,586 / 2.7%

As can be seen by looking at these statistics, these 2010 crime rates are much lower (except for Mongolia) not only than the 2011 crime rates described in the article, but also the 2010 crime rates visible in this graph accompanying the article. As we'll see, the crime rates on that graph are based on there being 22,543 people out of 1,261,415 foreigners in total arrested in 2010, which is a lower figure than the 33,586 charges listed in the SPO stats above, and yet the crime rate in the article is still higher.

So a look at the study itself is necessary. Searching for the Korean Institute of Criminology report (in Korean:  "외국인 밀집지역의 범죄와 치안실태 연구") at the KIC website leads to this page, where it can be downloaded (note - after downloading, you must add '.pdf' to the file name in order for it to work).

The report is 301 pages, so, to be clear, I haven't read it all! It starts, however, by saying that in 2001 there were 4,328 foreigners arrested, which made for 0.22%  of the total figure of 2,005,476 people arrested that year, but in 2011 27,144 foreigners were arrested, or 1.43% out of the total figure of 1,900,489 people arrested. This makes for a 527% increase in the number of people arrested between 2001 and 2011. However, between 2001 and 2011 the number of registered sojourners increased by 146% (or if you include these people and those who registered addresses together, you get an increase of 357%, a higher figure. [I have no idea what's being talked about at the end; 146% is the correct population increase figure, as the total number of foreigners rose from 566,835 in 2001 to 1,395,077 in 2011.]

The report says that between 2007 and 2011 the crime rate for Koreans for the 5 serious crimes (murder, burglary, rape, theft, assault) decreased from 1,113 to 1,033 per 100,000 people, while among foreigners, it increased from 636 to 918 per 100,000 people, a 44% increase. In the four areas with high concentrations of foreigners (Ansan, Siheung, and Yeongdeungpo and Guro in Seoul), the crime rate for the 5 serious crimes was higher than the foreign average for such crimes (twice as high in Guro, and 'almost approaching the [related] Korean crime rate' in the other three areas).

It also notes that illegal aliens committed 5.7% of crimes by foreigners, though they make up 12.0% of the foreign population, so they have a low crime rate.

Moving along to the crime statistics, we're first shown this chart (as with all charts, click - or open in a new tab - to enlarge). Columns are by year/ total number of criminals / foreign criminals / foreign criminal increase over previous year / foreign criminals as percentage of total

A graph illustrating the rising number of foreign criminals, 2001 - 2011.

Total number of criminals by year and nationality (with percentage out of all foreigners in brackets).

 The problematic chart showing foreign nationalities by crime rates (per 100,000 people).

Since the previous chart gave us the actual numbers for people arrested, I crunched the numbers using 2011 immigration statistics and came up much lower crime rates. I suspected that for some reason KIC decided to use 'registered' foreigners (which may mean those registered with immigration offices, but I'm not entirely sure). Page 14 of the PDF here (the zip file has .xls files with the correct total figures) lists the number of 'registered' foreigners (followed by legal / illegal foreigners):

The problem is that, while most 'registered' foreigners make up around 70% to 80% of a nationality's population in Korea, only 30% of the Canadian total, 57% of the Russian total, and 20% of the US total are represented in the above figures, which explains why those three countries are listed in the KIC report as having such high crime rates.

In the case of the US, the 26,466 'registered' Americans doesn't even cover the 31,000 US soldiers (or 10,000 others connected to the military), even though the report notes that USFK criminals are included in the crime statistics! To very briefly sum up the American population in Korea in 2011, 41,000 were connected to the military, 19,000 were on tourist visas, 11,000 were on E2 visas, 5,000 on F1 visas, 2,000 on F2 spousal visas, and 40,000 were on F4 dongpo (gyopo) visas.

In the case of Canada, only 6,572 of 21,812 are listed as 'registered,' but, once again, it's not clear what that means. For example, a breakdown by visa of Canadians in Korea at the end of 2011 gives us 2,737 visitors as visa-free tourists, 66 on short time work, more than 110 as students, 74 investors, 93 in trade management, 165 professors, 3,987 on E2 visas, 18 on E3 research visas, 30 on E4 tech instruction, 59 on E5 special work visas, 16 on E6 entertainer or athlete visas, 254 on E7 specific activity visas, 281 F1 accompanying another visa holder, 836 on F-2 spousal visas, 238 on F3 accompanying visas, 11,290 on F4 dongpo (gyopo) visas, 342 on F5 series visas, and 31 on the F6 spousal visa. That the 2,700 tourist visa visitors wouldn't be registered is understandable, but that 70% of Canadians aren't makes little sense, unless 'registered' means something different.

Also, in order to state at the beginning of the report that between 2001 and 2011 the number of registered sojourners increased by 146%, they used the correct statistics, so it's not like they didn't have access to them.

I made this chart to show the correct population totals, the incorrect totals used in the report, and both the incorrect and correct crime rates:

So the correct order from highest to lowest crime rate per 100,000 people would be Mongolia, 5,249, Uzbekistan, 2,447, Pakistan, 2,378, China, 2,312, Russia, 2,152, Vietnam, 2,097, Thailand, 2,068, Taiwan, 1,915, USA, 1,353, Canada, 1,242, Philippines, 1,125, Bangladesh, 920, Indonesia, 462, and Japan, 223.

In listing its own, incorrect version of the crime rates, the report does warn that one should be careful not to make interpretations which generalize based on the statistics, which is better advice than one might have first imagined. On the one hand, when national assembly representatives use incorrect statistics and portray 22,000 E2 visa holders running amok outside of Immigration's control, it's not so difficult to understand, what with all the other things like keeping track of donations, buying cake boxes and booking room salons which need to be taken care of. But when a government-sponsored research institute makes a similar mistake with immigration statistics, it doesn't exactly inspire confidence.

On the bright side, there is now a set of crime statistics for both Koreans and foreigners available and easily accessible, so there's no complaints about that here.


K said...

I think the point about people registering addresses together is saying that there was a 150% increase in the total number of foreigners, but a 350% increase in foreign households.

idlewordship said...

Was waiting for your post on this. Something seemed wrong in the media portrayal. As always. Great work!

kushibo said...

A proper comparison of anything would have to have similar definitions of crime, both using arrests or both using prosecutions. And it should include age-adjusted rates.