This is an enormous
week in the United States: opening today with the Martin Luther King holiday, and culminating
with the Presidential Inauguration. So
it’s perhaps interesting to see a small spar between two individuals who are
close to each event. Congressman John Lewis,
a prominent figure from the Civil Rights Movement era, told NBC that he somehow
didn’t see President-elect Trump as legitimate and -concurrently- was "for the first time" boycotting the inauguration, which will be attended by current and former presidents Obama,
Bush, Clinton, and Carter. He did boycott George W. Bush's inauguaration however. And Donald
Trump rejoins on twitter that Congressman Lewis “should spend more time fixing and
helping his district, which is in horrible shape … not to mention crime
infested”. Now the assertion
that Russian President Putin hacked the election is one that has been discussed
already on this site (here,
here). So instead this article will (perhaps too narrowly
for some) focus on the crime statistics in Congressman Lewis’ district, which
has been the new curiosity though for many following this recent diminutive squabble.
Congressman
Lewis represents Georgia’s 5th congressional district, permeating
from the heart of Metro Atlanta (also the capital of Georgia). See this map below to get an idea of the
district. One familiar with Atlanta will
notice that the region covers downtown Atlanta, as well as the more affluent
and among the nation's most habitable spots of Buckhead and Decatur.
These are definitely fun areas, where I personally have spent plenty of
weekends enjoying the shopping, dining, and nightlife. But the district also includes sketchier
cities southwest of Atlanta. These
include areas surrounding East Point, College Park, and Morrow.
What’s more
important than the heterogeneity of crime in Congressman Lewis’ district, is a basic
sense of how the overall crime rate looks relative to other places. So let’s start with the murder rate (and we’ll
use Atlanta itself as a proxy). The
murder rate in Atlanta is over 20 (all homicide statistics are quoted in murders per 100,000
people), which anyway is 3 times the
murder rate in Georgia and 5 times
the murder rate nationally (and the murder rate nationally is already higher than the global average rate!) We should
note, perhaps related, that there is sometimes enhanced fear among Black Georgians
(a state that had record hundreds of lynchings of Blacks up through the Civil
Rights Movement).
So now let’s
look at Atlanta’s murder rate relative to other large cities nationally. Atlanta has a population of just over
450,000, implying it is the 39th largest city in the U.S. and the
median city of the 80 U.S. cities with a population of over a ¼ million.
This chart
shows a lot of contextual mortality statistics information. The average homicide rate among a
composite of the 80 cities is 9 (per 100,000 people). And the standard error about this is 8. Hence we see a red horizontal line on the
chart indicating the just over 17 violence rate (for 1 standard error above
average).
We show all
of the 80 cities plotted in the chart, with Atlanta and most of the cities
labelled for convenience. What we see is
that there a dozen cities that have murder rates higher than 1 standard
deviation, which is also about what is expected from a Gaussian distribution
(though the distribution we can see is empirically closer to a lognormal).
In this sense
President-elect is only somewhat correct in pointing out that Congressman Lewis’
district is crime ridden, though there are 8 cities with even higher murder rates on
the chart above. And it is also worth
noting the oft-cited Atlanta Journal-Constitution came out with an article
titled “Atlanta named one of America’s top ‘murder capital’”. While we can argue about how bad the violent
crime is in Atlanta, no one argues it is
instead one of the America’s safest cities.
Factoring demographics however (age, race, longevity, etc.), the conclusions do become slightly foggier, as noted below.
Now look at
another city, with 6 times the
population, and yet a little lower murder rate: Chicago. That’s right, Chicago which we are all
frightened about after last week's radical criminal-justice report from DOJ
Attorney General Lynch. The
demographics are different (with Chicago far closer to the country's diverse profile on age and racial mix), but on face value Chicago has a lower murder rate than Atlanta! Now some pundits of this analysis might state that geographic relative comparisons are less useful than noting the murder rate has come down over the years to a safe absolute level. Though Congressman Lewis has been in office for decades, through waves of homicides during that time, and murder rates have come down everywhere (all cities, and globally). There is still no denying -particularly among police makers- that for the "bad section" of his district, as well as for the same in Chicago and other "above 1-sigma cities", there is obviously no comfort in time series analysis because they live in justifiable fear.
And with these statistics ideas under our belt, it is with sincere hope that this week all Americans get a chance to remember the life of Martin Luther King. And also celebrate a key event in our free democracy: the inauguration of our 45th U.S. President.
And with these statistics ideas under our belt, it is with sincere hope that this week all Americans get a chance to remember the life of Martin Luther King. And also celebrate a key event in our free democracy: the inauguration of our 45th U.S. President.
So now correlate murder rate in cities to percentage of black population. Go ahead.
ReplyDeleteHa Jerry, that is so risqué! Also if you haven't seen: http://www.zerohedge.com/news/2017-01-16/losers-who-wont-lose-john-lewis-versus-donald-trump
Delete