Thursday, June 22, 2017

Just Saying What We All Know...

h/t PawPaw

How far down on the last panel do you have to go to find a city with a Republican mayor/administration?

Fight crime: ban Democrats.


John ShootBetter said...

The Democrats keep talking about common-sense gun laws we can all agree on. I have one: No guns for Democrats!

loren said...

Interesting graph. I'm thinking if you made another that showed murders per capita of diversity the graph would be a straight line. Include cities in other continents just for giggles.

Aesop said...

I didn't make the first graph, but back in the mid-1990s, when they cared about honesty, National Review noted that if we looked at crime in America committed by caucasians, our crime rate is lower per capita than Switzerland; and that if you screened out Hispanic crime from caucasian crime, our national crime rate per capita is lower than that of Lichtenstein (Google it if you need to).

I doubt that was true only then.

Mike_C said...

Well now, that thought of yours would be badthink. That said, I may know of someone who actually made the plot and ran basic statistics.

Now, this person may have used demographic information from the 2010 US Census, and s/he may have used simple Pearson correlation to examine the association between the aggregate 2010-2015 homicide rate (from the figure) and the following:
1. percent population black/African-American (%B)
2. percent white (%W; ignoring the "non-Hispanic" thing, i.e. regardless of n-H or H you were just plain white -- except for Miami, where the n-H white figure was used)
3. percent non-white (%NW=100% - percent white)
4. out of curiosity, homicide rate vs total city population (because of the whole "mouse utopia" thing -- those of you who don't know what that is, LOOK IT UP)
Anyway, this badthinking person may also have tested the significance of the correlation.

It's rumored that the results are as follows (HR=homicide rate, %B, %W, %NW as above; TP=total population)
HR vs %B: r=0.68, p<0.001
HR vs %W: r=-0.46, p<0.05 [note the negative sign]
HR vs %NW: r=0.46, p<0.05 [this is a well, duh finding since %NW is 100%-%W, but presented for completeness]
HR vs TP: r=-0.16, p=nonsignificant

Mike_C said...

So the association between percentage of blacks in the population is significantly correlated (the p<0.001 indicates that there is less than a 1 in 1000 chance this is a spurious finding due to random chance rather than a true association) with homicide rate. Note that the percentage of whites and HR is inversely correlated (hence the negative sign on the "r" value); higher proportion of whites is associated with lower homicide rates. More non-whites is associated with more homicide, but the association is weaker than the association between blacks and homicide.

Note that the r=0.68 for %B vs HR gives an r2 (r-squared) of 0.47. Despite lazy stats jargon commonly in use, this does NOT mean that blackness "explains" 47% of the homicide rate. The r2 is a way of assessing how well the underlying statistical model (i.e. the badthinker assumed a straight-line relationship between %B and HR when s/he used linear correlation) fits the actual data. An r2=1.0 (maximum possible value) would indicate that a linear model perfectly fit the data. Which it doesn't. What this says is that there is a linear relationship between %B and HR, but the fit is only "okay" and not "good" much less perfect.

To expand on the above, even though these results show a significant linear association between %B and HR, it doesn't tell us about causation. For example, it could be that poverty is the cause of high HR, and blacks tend to be economically disadvantaged.

There are many limitations to this simple analysis, including the fact that the HR is averaged over 2010-2015 and does not fully capture recent spikes in homicide rates (q.v. Baltimore [AKA (c)Harm City, hat tip to James LaFond] since the Freddie Gray unfortunateness). Also, these findings are not adjusted for other factors such as income, education, differences in police staffing and ROES, etc. Finally, these results are unlikely to generalize to other nations.

So says this badthinking person I'm sure I don't know well at all. Certainly not a friend of mine.
A last note from me. Wikipedia has an entry on 50 cities (not a war zone, population>300k) worldwide with the highest homicide rates. Four US cities are on this list (St Louis at #14, Baltimore #26, New Orleans #34, Detroit #36). The list is dominated by Central and South American cities (especially Brazil); Africa is represented only by Cape Town (#13), Nelson Mandela Bay #43 and Durban #50. Two thoughts on this:
1. I am astonished to see not a single European* or Asian city on the list;
2. Only the RSA represents Africa on the list. This may be because the rest of Africa is safer, homicide-wise, than the RSA and the USA, or it may be because in much of Africa they are so screwed up that they aren't able to keep records of homicides. The reader is invited to decide him/herself.

*that may mean this is a shitty and incomplete wikipedia article; also, things are just beginning to hot up as more and more culture enrichment takes place

Aesop said...

Regarding African "statistics", as such, refer to my Ebola posts from 2014 onwards.
Getting stats from people who lie about them because they can (let alone must), and due to literacy and numeracy rates need their shoes off to get to twenty (assuming they possess both legs) bodes a poor source sample to work with.

Google Kim DuToit's old essay "Let Africa Sink", and work forward from that point.

Regarding Asia: Singapore, Seoul, Tokyo, and even Peking are one thing.
Bangkok, Djakarta, Manila, and Mumbai probably something else entirely.

In my experience, most large cities resemble most closely Pratchett's fictional Ankh-Morpork in any terms, but particularly crime, more than the city fathers of such districts would like to admit (and will deny, when the point is pressed).

So at that point, is a city better served by someone who thinks people are basically bad, and government is a guard rail on unchecked behavior, and neutral towards those staying in their lanes; or by one which thinks people are basically good, apologizes for the worst by blaming nurture over nature, gives them 27 chances to prove it, and seeks to force them all to be better, whether they like it or not, and nannies them 24/7/365?

The posted graph speaks volumes to that question.

Anonymous said...

Here's an article from one of the W4 posts. Goes right along with this.