Monday, June 23, 2008

Videogaming and Crime Victimisation

This graph seems like it's rooted more in propaganda and wishful thinking than hard data. It also makes a mockery of any number of theories of social deviance. But I've spent the best part of two hours trying to pull it apart and I've yet to find a significant chink in its armour.

The numbers come from a US Bureau of Justice survey of violent crime in the United States between 1973 and 2005. The survey targets a "nationally representative sample" of around 77,000 households and asks questions about the experiences of individuals with crime (so this is based on victim perception of crime, rather than police reports, prosecutions or convictions).

It's more or less corroborated by the FBI's Uniform Crime Reporting Program, which gives numbers for reported crime.

The turning point in both sets of data appears to be 1993. It's possibly significant that the US Bureau of Justice made some major changes to their data collection in 1993 but if anything those changes should have raised the numbers rather than lowering them. Correctional data for the period is suggestive but doesn't seem to provide the answer.

Does anyone out there have any insight into what's really responsible for these dramatic statistics?

Thanks to Hibikiryouga for drawing my attention to the graph.


Juffles said...

(so this is based on victim perception of crime, rather than police reports, prosecutions or convictions)

There's your significant chink. Why would you survey 77,000 random idiots about their "experiences of crime" (whatever the hell that means) when you could collate statistics from every cop shop in the country? Sure, you would miss the unreported crimes, but if someone's not going to report a crime are they going to tell a stranger with a clipboard? (improved methodology or not)

Mainly I just have a problem with graphs that have a misleading baseline. :)

Greg Tannahill said...

Victimisation surveys are particularly valuable in getting accurate data surrounding sexual assaults; victims are more likely to report incidents of sexual assault in an anonymous survey than they are to a formal investigative process.

They're also quite good at picking up victims of crime who are themselves offenders. Those involved in property theft or drug offences are much less likely to report crimes to police but may be fine with responding to a survey.

Finally these kind of surveys can get data about victimisation among minority groups, particularly non-assimilated immigrants and indigenous populations, who may have difficulty interfacing with the criminal justice system or who may be suspicious of it.

Anyway, like I said, the reported crime numbers, which are the collated cop shop stats, more or less mirror the survey, although the survey numbers are a bit more dramatic.

So I think we have to keep looking for an answer.

Juffles said...

The reported crime numbers actually say otherwise - where the victims/1000 in the survey fell from a high of ~51 to ~21 (59% drop) the FBI data says there was a fall of 38%. (758.2 to 469.2 per 100,000)

So there's a couple of problems with the survey data -
1) There's 7x as many victims in the survey as there is reported crimes. Is the crime reporting rate really only 14%?
2) The fall is far greater in the survey data, which suggests...what? People hardened the f*** up? Got blase about minor crime? Stopped fiddling their daughters?

That's all kinda by the wayside anyway, as the FBI data still shows a nice drop in crime per capita. The thing is, the number of violent crimes that the cops knew about barely drops at all from 1986 to 2005 - 90,000 less, or 6.5%, and actually trends up slightly from 2005 to '06. Maybe the real question is why does it spike so high in the early 90s? I'd be interested in the official data pre-1986 to see what the longer-term trend is.

Greg Tannahill said...

Spike in the early 90s may correlate to baby boomers reaching the high-risk age groups for both offending and victimisation.

Fluctuation up to about 1993 is normal; that's the sort of numbers that you expect to see across all cultures across all time periods regardless of policing effort. It's the drop after that which is staggering.

Your 14% is probably derived from the "total violent crime" figures, but note that the survey includes simple assault numbers whereas the FBI reporting data doesn't. Simple assault clocks in at 13.3 per 1000 on the survey data out of the 21.0 total for violent crime so that's obviously a hefty chunk of the aggregate figure.

If you look at a category by category break-up it's less distressing. For example:
Aggravated assault 4.3 surveyed vs 2.9 reported;
Rape 0.5 surveyed vs 0.3 reported;
Robbery 2.6 surveyed vs 1.4 reported

The variance is within what I'd expect.

If you go to the Bureau of Justice site you can find a copy of the survey questionnaire which is fairly heavily skewed towards getting participants to over-report.

I suspect that the size of the discrepancy as we near the current day is not a fall in the surveyed victimisation rates but rather a comparitive rise in reporting rates; certainly crimes like sexual assault are far more likely to be reported today than ten years ago.

flashman said...

Think how much lower the crime rate would be if violent video games didn't exist. They must be the one thing stopping violent crime from disappearing altogether.

(Tongue firmly in cheek.)

flashman said...

Oh, Greg: I'm looking for recommendations on DS games for my girlfriend's birthday. She's not a full-on "girly gamer" but does enjoy Viva Pinata, word puzzles, the Sims, and casual games eg Bejeweled. I'm looking for something that won't insult her intelligence - what would you recommend?

Greg Tannahill said...

Flashman - The Phoenix Wright games are thoroughly recommendable for all audiences, although start with the first as they have continuing storylines. They're plot-driven legal shenanigans with very simple mechanics.

I've got a friend whose interests sound similar who is a fan of the Lost in Blue games, although I haven't had experience with them myself. (They're essentially deserted island sims.) Harvest Moon might also be appropriate although you're better off getting one of the Game Boy Advance versions to use with the DS GBA slot as the current-generation ones are only so-so.

Brain Training (aka Brain Age) is selling to casual gamers like there's no tomorrow even after several years, and it includes a couple of hundred Sudoku puzzles, which is awesome.

Puzzle Quest might also be a winner; it's an RPG where the fighting occurs as a Bejewelled clone.

If you need more options beyond that, let me know.

flashman said...

Thanks Greg, that's a great start. I've been a little overwhelmed by the sheer tonnage of DS games on the market :)

Greg Tannahill said...

It's like the gaming candy factory exploded and I'm up to my eyeballs in fun.