Comment by pjc50 | Hacker News Reader

Comment by pjc50 | Hacker News Reader

Parent Full thread

As soon as I saw John Lott, I knew I was going to be dealing with one-sided data. But I'm having trouble replicating that graph from https://www.gov.uk/government/uploads/system/uploads/attachm...

It did point out that I was wrong about no mass shootings (p57):

"In 2010/11, the police provisionally recorded 642 offences of homicide, which includes the 12 people killed by Derrick Bird in June 2010. Homicide offences increased by four per cent (24 offences) compared with 2009/10 (Table 2.04).3 Caution should be taken in looking at short term changes in the number of homicides, as they can fluctuate from year-to-year. For example, the rise in homicides this year follows a fall the previous year"

But the dead giveaway is buried in a footnote on p58:

"As measured by the Homicide Index. The number of currently recorded homicides peaked in 2001/02 at 794. There are more currently recorded homicides for 2002/03 (943), but these include the 172 victims of Harold Shipman, which were all recorded in 2002/03 but took place over a number of years."

Harold Shipman was a doctor who spent a large part of his career murdering elderly patients with medication. The huge spike in John Lott's graph is almost entirely due to this one set of crimes being recorded in 2002/03.

But are homicides down overall? What's the purpose of banning firearms if they don't prevent murders? Because if we compared home invasion rates (and perhaps some other categories), I'm guessing they'd look drastically lopsided. In the US state where I live 80% of those convicted of home invasion said they carefully picked a house where they believed the owner was not armed. And the news often carries stories of people stopping assaults, rapes, and home invasions with their personal firearm, nearly always with no one being harmed. These stats are of course not recorded.

"Home invasion" is not a category known to UK crime reporting; the same PDF says

There has been a general downward trend in police recorded homicides over recent years. If the provisional figure of 642 homicides is confirmed when the final figures from the Homicide Index are published, then this would represent a fall of 19 per cent in homicides since 2001/02."

and

"The BCS shows that the number of violent incidents increased gradually through the 1980s and then increased sharply after 1991 to reach a peak in the mid 1990s. The number of incidents then showed steep decreases in the late 1990s. Since then, despite non-statistically significant year-on-year changes, there has been an overall decline."*

The decline in violence is actually somewhat global and not really anything to do with firearms either way. What the gun ban does unambiguously reduce is the opportunity for mass shootings.

(Did you mean compare rates within the same country, or rates between the UK and US? The UK's overall murder rate is comparable to that of two or three US cities. Sadly this PDF only goes back to 2000/2001 and doesn't cover the 90s.)

This may be of interest to you:

https://en.wikipedia.org/wiki/Tony_Martin_(farmer)

I was sure there was a nasty bit of deception going on, I was digging through gov.uk's England/Wales crime stats PDFs looking for the explanation and it seems you've figured it out :) However if my previous chats with pro-gun types are anything to go by, please don't be surprised if there's a reply along the lines of "so if that was the cause then why don't you go ban medication..."

Washington DC has the lowest gun ownership per capita and yet the highest gun murder rates in America.

https://en.wikipedia.org/wiki/Gun_violence_in_the_United_Sta...

Please explain the lack of correlation between gun murders and gun ownership. Even when controlling for density, there is still no correlation.

Bad stats, there is no actual nationwide tracking of gun ownership in the US making such numbers pure BS.

Further, crime rates need to be adjusted for poverty or they become meaningless.

I agree on the tracking on gun ownership in the US, which is why I treat all such research with the same distrust. Whether they are pro or anti gun in nature.

I think DC is an anomaly because of its small size and proximity to Virginia, which has one very permissive gun sales laws. State-by-State reporting only really works when states are large enough that the populations you are studying live almost entirely within the border of such state and the actions that take place in the state are done by the same population. DC as a very small city (500K or so actual residents) is not going to be accurately captured by such a study because many of the activities that take place in DC are undertaken by residents or Maryland or Virginia. For a more mundane example than gun deaths, one could look at traffic accidents. If one took the number of traffic accidents that occur in DC and dvided by the number of DC residents, I would expect DC residents would look like bad drivers (high rate of accidents per resident), but that is a false conclusion because many of the cars on the road on any given day come in from neighboring states (I have been told that daytime population of the city is three times the number of residents, but I don't have a citation for that, so it may be wrong but is certainly directionally correct). So your denominator is wrong, you should look at traffic accidents in DC/cars on the road in DC or some similar metric.

I've seen numerous times people complaining that crime statistics are recorded differently from country to country and the differences being used strategically to prove an argument. I've seen such spikes used to prove and disprove the same arguments before.

Ha, I knew there was something off about that spike. Well spotted.

My point still stands below, despite that spike, the overall trend is still very much downwards. People just don't read charts very well.

Reply on news.ycombinator.com