What you probably don't know about homicide statistics

This article was recently featured in LinkedIn's 'Big Data'

This is part two of my homicide statistics series.

I have decided to merge this with part three so this is the final post of the series.

One day, I realized that everything I was reading about homicide rates was wrong. It was a YouTube comment that got the ball rolling, and I started following the activity of the user who wrote it. 

Back then, you could view a user’s comment activity.

What he was saying was very important and something that took me a while to get used to because I was so used to seeing things a certain way.

His comments were to do with population figures and the widespread naivety surrounding corrupted city limit populations that are touted– especially US cities.

If you read my first post in relation to this, you’ll have seen homicide figures I attributed to some cities in the US for 2015. In reality, those figures are somewhat misleading.

The idea that the size of a city is a limited, ill- and politically-defined area is wrong. A city can't be smaller than its urban footprint. It can’t be, by the laws of common sense.

Administrative borders are applied, but a city doesn’t obey administrative boundaries. Cities don’t just stop as if they had a mind of their own.

It’s similar to the idea that every London borough could be its own town with its independent authority. But they’re not, they’re merged to make up Greater London.

Mexico City is commonly said to have 9 million people, but its population is 20 million. To repeat, a city can't be smaller than it's urban footprint. It's the same city. It doesn't matter who governs or how many administrative boundaries there are, that would have nothing to do with not being the same city. It's still an obsolete boundary.

So, what about the misleading headlines you read?

Let’s take the 'city' of St. Louis and the 'city' of Ferguson (the place that made headlines after its riots). 

I say, Ferguson is a suburb of St Louis. And I say St Louis city is a municipality in greater St Louis - the real city. Like Manchester is to greater Manchester - same urban area, same labour movement, same city.

St Louis’ supposed city population is roughly 360,000. The real size, I say - that of Greater St Louis - is over 2 million (which gulps up the likes of Ferguson).

When calculating the number of murders committed in the arbitrary city limit population, the murder rate in St Louis for 2015 was about 52 per 100k.

Consequently, the papers will report that St Louis is ten times deadlier than NYC (in relation to population). But it’s not. Greater St Louis – far more fairly matched with NYC – has an annual murder rate of about 5 per 100k – much closer to NYC's rate than you have been told.

This is repeated over and over again. Ever wondered why America's deadliest cities are always tiny 'cities'. For example, Camden, NJ.
To compare Camden to, say, NYC is woeful.  You might as well compare it to Brownsville and East New York (two historically violent neighbourhoods in Brooklyn).

Otherwise, you are just comparing a grape with a watermelon. It's not a fair comparison, even when using ratios. That's why metropolitan regions are there - to capture cities' sizes as they grows.

There is – or, at least, has been – an annual publication that is often cited as a legitimate source of information – Morgan Quinto.

Every year it publishes America’s most dangerous city list (and safest). But in reality, it is a bad source of info.

You will find that most media outlets don't like questioning elevated murder rates or their origin if there are problems with the numbers. Unless numerous experts point it out, they'll tend to just go with it.

The FBI makes an effort to warn against city-by-city comparisons and encourages metro region comparisons. When you read the table for this, you will see a much different list of deadly cities compared with the ones we are so used to hearing.

Pine Bluff was the 2014 murder capital, followed by near-perrenial No. 1 New Orleans. Detroit probably isn't the top 10 even though it is always touted as America’s deadliest city (or in the top three). See here https://www.fbi.gov/about-us/cjis/ucr/crime-in-the-u.s/2014/crime-in-the-u.s.-2014/tables/table-6

Look at Chicago. Type 'murder' into Google and you'd think Chicago led the world in murders. But it has a population of 10 million and sees around 600 murders a year. That's 6 per 100k. A bad figure, but it pales in comparison with scores of cities outside the US. The same applies to the rest of the so-called US mega deadly cities.

Maybe I'm not being clear enough? See the link immediately below to learn more - it's one of the few articles I can find online to substantiate my perspective. Thank you, USA Today.

https://usatoday30.usatoday.com/printedition/news/20071129/opcomwednesday.art.htm

For a bit more understanding of populations, see Thomas Brinkhoff's notes about his agglomeration list on the City Population website. A lot of people don't use "city" to describe these things anymore because it's been so corrupted (particularly in the US where it's utterly useless). They think it's worthless. They use "metro" or "agglo" instead.

For the US and the West, it is relatively safe to compare homicide rates. But things change for the rest.

The US and the West are transparent but minor league

As bad as it gets in the US, it gets a whole lot worse in its southern neighbours. But again, the media get a bit stuck with its data.

Latin America has been getting more and more limelight in recent years, mainly thanks to Honduras and El Salvador. In the 1990s, Brazil and Colombia were in the news a lot, and rightly so, but access to info was much trickier then - especially for non-scholars like me just interested in crime and criminology. No internet.

But as dangerous as Latin America can get, it is not more dangeous than Africa or the Middle East. Its statistical infrastructure allows for greater access, thus, lopsided data analysis and reports.

Except for the Caribbean (not including Haiti) and South Africa, it has the most accurate homicide statistics in the developing world.

A city such as Kinshasa in the Congo is considered to be comparable with the most dangerous Latin American cities as far as insecurity is concerned, yet its official murder rate is 1.6 per 100,000. Haiti has an official rate of about 10 murders per 100,000, a country widely considered to be the most dangerous place in the Americas alongside Honduras and El Salvador.

Credit must go to the Guardian for its interactive most dangerous city map (I'm still searching for the link but I remember it well). Mogadishu was on there, as was Aleppo and a few other conflict zones.

We can sit here and pretend that Latin America is more dangerous than Africa or the Middle East because it has a recognisable statistical infrastructure, but it's not. 

I'm nearly sure that Syria's rate is a conservative 200-300 per 100k.

What a country includes in its death tolls affects things, too. Venezuela includes all deaths from external causes that take in suicides and accidents. This naturally inflates the number, but it is something that won’t be questioned by the media. Don’t get me, there is still a staggering level of violence in the country, sadly.

Here are three examples of what you will see in the news:

https://www.worldatlas.com/articles/most-dangerous-cities-in-the-world.html

https://uk.businessinsider.com/the-50-most-violent-cities-in-the-world-2015-1?r=US&IR=T

Even my beloved Economist

https://www.economist.com/blogs/graphicdetail/2016/02/daily-chart-3?fsrc=scn/fb/wl/bl/dc/st/theworldsmostviolentcities

They're not the worst lists I have seen. They rightly include notable cities that would have barely got a mention in the past and are probably unknown to most people. Data availability and a bit more effort are improving journalism from this point of view, but they are still off. I understand that limited data for many African and the Middle East nations makes things tricky, but they shouldn't be ignored. Detroit and New Orleans should be nowhere near those lists.

To wrap up: always scratch the surface. I apply that to many things. Especially when statistics are involved. Statistics are a great measuring tool, but can be manipulated and badly judged