Experienced Segregation

Here is a new and important paper from Susan Athey, Billy A. Ferguson, Matthew Gentzkow, and Tobias Schmidt:

We introduce a novel measure of segregation, experienced isolation, that captures individuals’ exposure to diverse others in the places they visit over the course of their days. Using Global Positioning System (GPS) data collected from smartphones, we measure experienced isolation by race. We find that the isolation individuals experience is substantially lower than standard residential isolation measures would suggest, but that experienced and residential isolation are highly correlated across cities. Experienced isolation is lower relative to residential isolation in denser, wealthier, more educated cities with high levels of public transit use, and is also negatively correlated with income mobility.

Here is the NBER link.  Here is an earlier and ungated version.

Comments

The small southern town where I grew up turns out, by this measure anyway, was not segregated. Every day the blacks, who resided on the west side of Pine Street, crossed over to the east side to work for the white folks who resided there. See, no segregation. It's a novel measure all right. "When I use a word," Humpty Dumpty said, in rather a scornful tone, "it means just what I choose it to mean—neither more nor less." "The question is," said Alice, "whether you can make words mean so many different things." "The question is," said Humpty Dumpty, "which is to be master—that's all."

What? "small southern town", "to be master"...dats racist.

I think the study is saying that in rich cities, like raywards says, the poor folk cross the wrong side of the railroad tracks to work for the rich more often than you would expect if they were in a poor town, and rich people are more mobile that poor people, since they don't have to depend on public transportation. There's enough profundity in these trite observations for a PhD thesis or two in some social studies program.

This is the Stanford Graduate School of Business, not 'some social studies program'. The authors are expert economists including a John Bates Clark medalist.

Yes, but what are they really measuring and why. It appears to have no practical value, merely a talking point to be used to support your bias.

Economics has provided many useful insights: tradeoffs, opportunity cost, marginal cost, invisible hand, seen versus unseen. The new stuff is just a lot of useless busy work.

I simply couldn't depart your web site beffore suggesting that
I really enjoyed the standard info a person supply on your visitors?
Is gonna be back frequently to check up on new posts

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

No, I get the point of the study. In my small southern town blacks and whites were in close contact every day - though not at school. Is that why blacks back then aspired to the same things as whites, for their children to have a better life than them? It's true, I lived in a totally segregated place, but blacks and whites had more contact and were actually more alike than they are today. "Acting white" wasn't considered an insult, and naming a black child after Washington or Roosevelt was common. Again, I get it. But isn't this like that idiot from Arkansas, Tom Cotton ("cotton pickin" Cotton), who yesterday explained that slavery was a necessary evil so that America could prosper and become the world's beacon for freedom.

Come off it, he said nothing of the sort.

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

As usual, the perception people have of their level of isolation is wrong. In general there is far greater mobility and interaction between people of different races and ethnicities and classes but this fact is not narratively useful to those who have a need to perceive it being otherwise. Density does not change the ratio, it is just proportional.

I am not sure whether races, ethnicities, or classes are interacting when paying at the cash register.

When in modern American life "paying at the cash register" is for many people the only interaction they have with anybody, I most definitely categorize that as an interaction. There are high-density residential housing and apartment blocks where people don't even know who their neighbors are, so it makes no sense not to classify it this way. Something, regardless of intensity, is not nothing.

'for many people the only interaction they have with anybody'

What a sad perspective, without commenting on how pervasive those circumstances may be..

"What a sad perspective,..."

Subjective. Although I think many people (there was a SSC post long ago about this very trend) would assume and anticipate that isolation will only get more and more common. It's entirely possible by 2070 people will interact with the physical world through avatars, but ymmv.

Technology truly has revealed for many people just how little they need other people, even in a transactional way. Assigning a moral value to that in my opinion isn't helpful. For many people this future might represent a golden age.

Of course it is subjective - how could it not be?

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Rather than trying to trump this data with anecdote, I'll try popping up a level.

1) I take it the big data backlash is over?

2) I'll guess that this was privately collected and quite personal data, hopefully well anonymized, before being shared with researchers?

3) Where else does this data go? Note that private data may be purchased by essentially any branch or level of government, bypassing any limits they face on "collecting" the same.

Interesting times.

2a) I hope there is nothing embarrassing in our national private databases that allows "race" to be recovered for a random cell phone number.

That was the oddest part of the study.

Respond

Add Comment

Don't worry, I'm sure they just used some proxy, like, obviously for instance, certain strings of numbers sound more black or white.

Respond

Add Comment

Respond

Add Comment

File under: no one bothers to read the papers.

We use the inferred home location to impute race and other demographics. We match each home geohash7 to both Census tracts and blocks as follows. We match the geohash7 to the tract that contains its centroid. This yields a matching tract for 99.53 percent percent of devices in our sample. We then match the geohash7 to all Census blocks that overlap its area, and assign demographics to the geohash7 by taking a simple average of these Census blocks.4 This yields a match to at least one Census block with non-zero population 98.12 percent of the time.5 We drop the remaining 2.27 percent of devices.

It uses census data, that thing you want to get rid of.

The ungated version was giving a 500 error when I looked. But that's actually one answer I considered.

"impute race and other demographics"

Of course that's bad, and yes I still want to get rid of the "race" question. It has no biological foundation, and is fundamentally anti-American.

"It has no biological foundation,"

And yet it is routine on medical questionnaires. Have you told the AMA about their ignorance of science yet? /sarcasm

Obviously race and ethnicity have biological components. This is well known and indisputable.

"Q: WHAT DO MY RACE AND ETHNICITY HAVE TO DO WITH MY HEALTH?
A: Although we are all individual people, our racial and ethnic backgrounds may place us at differing risks for some diseases. We can work to reduce these risks by making sure that everyone gets high quality health care. "

https://www.hcup-us.ahrq.gov/datainnovations/raceethnicitytoolkit/ca6.pdf

Wow, way to out yourself as a fan of "race." No wonder you were one of the last in the Trump pool.

And fwiw, US medical practice lags on this, probably for reasons of historic racism. Scientific American: What Role Should Race Play in Medicine?

A "fan" of race? Maybe just reality. In medicine, a person's race is valuable information. White and Black people react differently to some medications. Better to treat people well than to feed ignorance.

For the lazy:

Genetic variation across geographic loci is continuous—like a color spectrum or gradient—though medical literature often communicates race as immutable—like clearly separated colors. Research demonstrates that genetic differences are higher within racial groups than between racial groups—that two black patients sitting in the waiting room will have less genetic overlap with each other than with their white, Asian, or Hispanic neighbors. And while ancestral alleles can impact rates of disease and pharmaceutical metabolism, these alleles do not align neatly with reductive racial categories often employed to represent geographic origin.

So, assuming that Susan Athey, Billy A. Ferguson, Matthew Gentzkow, and Tobias Schmidt know this, what are they up to?

Are they imputing a *social* grouping and then looking at the mobility for the imputed group? Maybe. And maybe that might be useful. But I'd say in a very limited sense.

You've gone from linking to things that don't support your position to copy and pasting them. For race to be a spectrum, which I agree with, there must be ..... race.

Come on, man. Where was it used as a spectrum? In this study? Were white and black normalized to 0..1 real numbers? How, given the great genetic diversity in Africa would there even be a 1 to endpoint that scale?

"Come on, man" Joe, is that you?

You don't think people know everyone, or anyone even, isn't 100% of any race? Everyone is a mix. To think there's no biological basis though is a bit insane. Maybe there's a doc out there who could explain why Asian couples out there tend to think they're going to have an Asian kid.

Here's a pro tip: Everyone has an ancestry, and you can drop the word "race" when you discuss it. If you are talking about a Tibetan or a Tamil it's both more accurate and less loaded to be specific.

No, you can't because we have a lot more knowledge about the crude racial categories than the fine ones, because the results of such studies are far more useful and because it's much easier to get study participants.

So the reality is we can expect that, for example, if you are African American you probably have a reduced pain tolerance compared to a European American. That can be important clinical information. But we have no idea about the relative pain tolerance of people of West African or East African descent, relative to each other. So knowing someone is African American may be useful information, knowing the region of Africa is not.

By the way this "Research demonstrates that genetic differences are higher within racial groups than between racial groups—that two black patients sitting in the waiting room will have less genetic overlap with each other than with their white, Asian, or Hispanic neighbors" is complete nonsense. The first half is true but the second half is bizarre.

Now when we're talking about a social study your objections are far more bizarre, since race is experienced primarily as a social phenomenon and here the racial categories are what is overwhelmingly relevant.

Comment sections rejecting science to insist race is biological.

Sad, but very 2020.

https://www.nationalgeographic.com/magazine/2018/04/race-genetics-science-africa/

Rejecting "Woke-Science" Actual science is still intact.

You poor guys don't stop to think about what you are doing to the "conservative" brand when you make it "anti-anti-racist."

Yes, that's a key difference. I care more about the truth than I do about a brand. You care more about the brand.

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Han parents have Han children, Vietnamese parents have Vietnamese children, Korean parents have Korean children, and Japanese parents have Japanese children. Not a single one of those parents would ever describe their children as "Asian" in their own language.

Odd how they don't understand they are actually having Asian children, as those four groups of people consider themselves completely distinct (one even claiming that distinction is based on divine circumstances).

This is also why American ideas of whiteness - oops, definitely Whiteness in this context - fail so completely in Europe. Europeans of such inclination already have the idea that their 'race' is superior, without grasping that Italians, Greeks, Germans, Poles, Finns, and Danes are actually all white people first, and the actual ethnic group they identify belonging to with second.

So ethnic groups aren't races. We knew that. Thanks for the attempt though. And acknowledging the biological nature of race does not imply one is superior to the other.

Can't wait until you find out some people have pee pees and some don't. THAT'll be fun.

Ethnic groups are most definitely races. Just ask someone from Japan if they are the same race as a Vietnamese. Cannot wait until you find out some people have a different idea of race than you. THAT'll be fun.

But their kids won't be Japanese because none of this is biological, right?

' .. and Japanese parents have Japanese children. '

I am not sure what world you are living in, but having children is 100% biological. You were aware enough to point out that some people have pee pees and some don't.

Exactly. Races and ethnicity are biologic.

The Japanese consider themselves a race, not an ethnicity.

Still cannot wait until you find out some people have a different idea of race than you. THAT'll be fun.

So what? What is your point? "Asian" is a real, biological category of significance. So is "Japanese".

'"Asian" is a real, biological category of significance.'

To you. The Japanese see themselves as distinct from other 'Asians.' The same change of perspective is true of the French and Russians, who consider themselves basically as being two distinct races, not one.
As pointed out below, even American defined 'white' is not a very useful distinction depending on circumstances.

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

The Japanese word is 民族 (which means people family). Koreans use the same word with a slightly different pronunciation (minjok). They don't think "Asian" is a "race." Whorf aside (no word = no concept, no way to think about), Japanese noticed that some of the crew members on Perry's ships were "of color", and understood that foreigners classified groups by color.
There are three different ways to be a member of a group. I leave it to woke and unwoke readers as an exercise to guess what they are.

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Lmao prior_approval. This is White Boomer cringe on god tier.

Yes, Asians refer to their children as Asian. We have a concept of what Asian is. Is your source a wikipedia page? jfc

'Not a single one of those parents would ever describe their children as "Asian" in their own language.'

My sources are mainly Japanese and Vietnamese people. What are yours?

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

"Genetic variation across geographic loci is continuous—like a color spectrum or gradient—though medical literature often communicates race as immutable"

I don't think that's charitable. At some point, doctors will be able to sequence a patient's DNA and let that guide the diagnosis and treatment options. Until then, "race" is a rough proxy measure that a Bayesian would be unwise to discard when diagnosing an illness, deciding whether or not to order additional tests, or prescribing medication or treatment.

+1, that's a logical and rational point of view.

It may be, but if you really believe it, you cannot also believe that race is a real biological category at the same time.

As that long piece at Sci Am documents.

You know, another item in the news today was that darker skinned black women get longer prison sentences than lighter skinned ones.

We should not pretend that sort of "proxy" is used in sentencing and not in medicine. And given the way genetic actually work, skin tone is going to be a very poor proxy for "how African" an individual is.

"if you really believe it, you cannot also believe that race is a real biological category at the same time."

False

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Asking someone if they are white is not going to help when prescribing metamizole.

"The Spanish Health Ministry has issued a warning regarding the misuse of metamizole, a painkiller whose most popular brand name in Spain is Nolotil.

Following reports of British citizens who died after being prescribed the medication in Spain, the country’s health authorities are reminding users about the risk of a dangerous and unpredictable side effect known as agranulocytosis, which drastically lowers the patient’s white cell count and can even lead to death.

On October 30, the Spanish Medication and Health Products Agency (AEMPS) issued a reminder that metamizole can only be purchased with a doctor’s prescription, even though many Spanish pharmacies continue to sell it over the counter.

Last weekend, British news outlets reported that a dozen UK citizens had died after taking metamizole in Spain. It is one of the most commonly prescribed anti-inflammatory medications in Spain and Latin America, although it is banned in several European countries, including Sweden, the UK and Ireland." english.elpais.com/elpais/2018/11/06/inenglish/1541519618_515280.html

Respond

Add Comment

We can already do whole exome sequencing, but that is expensive and few payers will touch it for anything remotely run-of-the-mill.

Even then, we are going to find things that are going be things not seen in the genetics that will correlate with race (be it asthmogen exposure, or specific dishes like spinach leading to inappropriate warfarin dosing). Ideally your physician would ask you everything up front, but there are many times when the patient is obtunded or otherwise unable to answer so I play odds.

And I do this for all manner of things. If a patient comes in wearing a hijab with AMS, my first thought is not going to be AUD or DT and even if it looks a lot like DT my threshold to go for a benzo will be higher. Eventually I should get some real data and can dispense with empiric treatment for any patient, but at first pass I am quite willing to use any stereotype or group marker that increases the odds of the patient surviving.

Individual data trumps group correlates every time, but if I need to start before I can get the data and use empirics with some potential trial and error I am going with the correlates every time.

This is the definition of white supremacy and racism in medicine. It's exactly what Scientific American was warning against. A white doctor treats a Muslim woman differently in the hospital. Treats a black patient differently in the hospital.

Morally disgusting. This is what Trumpism has wrought.

Oops, this is the point where someone feels bad enough to impersonate me again. *My* real answer at 8:24pm, below

Name - anonymous

Claim - people are impersonating

Verdict - counting to potato

You get very frustrated, and then the impersonations appear minutes later.

Coincidence?

anonymous, you have terrible reading comprehension.

"You get very frustrated, and then the impersonations appear minutes later."

The impersonation? you mention responded to Sure, you are responding to Skeptical. IE two different posters.

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

I am much more comfortable with physicians using it as a hint than I am with the government using it as a citizen category.

Spot the troll.

Race is not a biological category and it has no place in science or medicine. That a white physician would treat patients differently based on the color of their skin is exactly what Martin Luther King Jr warned against.

Sure is a black doctor, for what it's worth

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Yes it’s racist to use risk factors.

Okay Boomer

What, did you get a 500 error on that article?

You should probably read your own linked article.

He clearly didn't read the Scientific America paper he linked to. But then he doesn't appear to read or at least to comprehend single line comments he responds to.

From the article he linked to:

This verifies that race is a risk marker:
"Rather than a risk factor that predicts disease or disability because of genetic susceptibility, race is better conceptualized as a risk marker—of vulnerability, bias or systemic disadvantage."

This condemns the US system for treating all Asians as one group instead of being more specific when it comes to race:
"And if there are 11 standards for countries in Asia, why and how are these measures collapsed into a single instrument for “Asians” who live in the United States? "

Once again this confirms that race is a risk marker:
"The American College of Cardiology’s cardiovascular disease Risk Estimator—a clinical tool that helps doctors decide if patients should start anti-cholesterol medication—adjusts its recommendation depending on whether the patient sitting on the examination table is “African-American,” “White,” or “Other.” "

Slow down and read all that again.

Seriously.

A "marker" is not a genetic category, and it confirms my statement, the one you opposed: "I still want to get rid of the "race" question. It has no biological foundation, and is fundamentally anti-American"

Sounds like you want to erase Black identities and Black experiences, probably because of your white fragility.

At one point MLK did have a dream.

White supremacy and racism are serious problems in our society that affect us all. We're not going to overcome them by pretending race doesn't exist. White supremacists sure don't. You need to learn about white allyship and what your role as a person of privilege should be in combating racism. It sounds like you have a long way to go, if you're quoting MLK, as a white person.

New favorite troll account.

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

It is White fragility, except among those ashamed of their race.

No, it's entirely Boomer fragility, except among those who properly know when to put the keyboard and mouse down.

The boomers always deserve lower case spelling.

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

By the way, you should have finished that section on the cardiovascular risk estimator:

These examples, along with even a short perusal of biomedical literature, demonstrate that researchers are inconsistent with their paradigms of race. While some delineate racial difference through “black/non-black” or “white/non-white,” others substitute permutations of other labels such as “African-American,” “African,” “Caucasian” or “European.” In other instances, ethnicities are co-opted to represent race (Ashkenazi Jewish, Pima Indian) as are nationalities and regional identitis (Hong Kong, German). Consider, for example, how “Native American”—a single U.S. racial category—collapses over 500 distinct tribes that lived across a geographic area the size of Europe. If “European” isn’t a racial category, should “Native American” be?

In this thread anonymous becomes a parody version of a Shark Lasers-esque /pol commenter arguing for fine tuned micro level racial categories. Wouldn't be out of place in a 1930s German salon. Ironic.

Let's just be non-insane Bayesians and make sure Black people get screened for diabetes and sickle cell anemia.

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

+1, for digging that out. I suspected the answer was straightforward and not some psuedo-conspiracy involving private cell phones.

fan of "imputing race" are you?

If you've got a complaint about the papers methodology, perhaps you should take it up with the authors of the paper.

I gave Skeptical a +1, because he went to the trouble of looking at the paper to find an answer to where the data came from instead of jumping to conclusions.

I looked, and got the 500 error. Still, I don't think you quite grasp how this study looks bad either way. Either it had a race look-up (bad) or it imputed race (I say bad as well).

The census is a bizarre windmill to tilt at.

But tilt away

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Note that one could do a similar, invasive, study of mobility without race. Take your cellphone number and impute, or literally look up via credit information, income, and then look at daily travel.

One should not be surprised that people stuck in low income surroundings have less mobility in both senses of the word.

Respond

Add Comment

Respond

Add Comment

"Our GPS data are provided by a company that collects anonymous location data from mobile
applications on users’ smartphones. "

how is this data actually collected and how can I ensure my data is not collected?

additionally the "geohash" resolution goes as fine as 4.8mx4.8m which can be used to positively identify "anonymous data" especially at their "home location"

It's creepy that this study can be conducted like this.

Data privacy laws are underrated.

Respond

Add Comment

Respond

Add Comment

"Experienced isolation" seems like a relatively useless concept. The area I live in is ~ 80% white but a lot of the service workers are not. So my "experienced isolation" is rather low. What residential isolation (vs. this "new and improved" measure picks up) is the segregation of substantive relationships. To extend the measure of segregation to include many more points of contact and find that that measure finds lower levels of segregation than a measure that includes fewer points of contact is as good an example of "No s%!t, Sherlock" I can find.

Yeah, this seems like a fairly moronic study. By this categorisation, a slave plantation would have very low levels of “experienced segregation” because everyone is in the same place!

Respond

Add Comment

Respond

Add Comment

I am exhausted by identity politics, and the topics of race, ethnicity or gender. American social scientists have studied these topics from every possible angle---well, so I thought. Now we can track cell phones.

BTW, there are YouTubes put up by American Black comic-entrepreneurs. Some are pretty funny. (Trigger warning: most are profuse with profanity).

The racial segregation in the YouTubes made by American Blacks is almost complete---very scarce white people.

Another study needed!

Respond

Add Comment

Whatever the scientific basis for or against the concept of "race" , the American constitution NEEDS to be and MUST be (the People demand it) be amended to guarantee equal rights for People of all "races" and "colors" (except maybe not White who need to have less rights in order to compensate for the history of brutality and oppression). That's what BLM and Antifa are all about, amending the Constitution and BORs even they were the product of White racist slave-owning male chauvanist pig supremicists.

Respond

Add Comment

That's an excellent comment Dr. Peretti-Nueves. However one thing should be noted: The impact of White and Black police brutality falls disproportionately on the backs of Caucasians, particularly White criminals who are standing up and speaking out for and exercising their constitutional rights. More people of color should learn from their example. No one should have to suffer the oppressive brutality of being arrested or detained merely because they tried to pass counterfeit money. It is grossly unjust to expect Black people to follow White supremicist laws after all the brutal oppression, rape and disrespect, we and other People of Color have suffered and are struggling with everyday.

Respond

Add Comment

My first reaction to this paper and all the research mentioned as related in the paper is that the researchers are willing to call "social interaction" to the simple fact that we coexist, and therefore we make at least visual contact.

For the past 10 years, I have been living in a two-tower condominium of apartments. I have seen many people around the two buildings --their population has been changing in many forms and I expect it will continue changing. I know, however, that the social interactions among residents as well as residents and other people providing services to them have been quite limited. Perhaps, it has been limited to less than 5% of residents among themselves (my personal experience is close to 0 in 10 years) and the only social interactions are (a) between residents and the regular personnel controlling access to the buildings, and (b) between residents and their nannies. Today, thanks to the quarantine, interactions (a) have increased sharply because residents rely on regular personnel to receive supplies.

Indeed, in large cities, coexistence implies many "contacts" (when using the elevator I often say hello to someone going up or down; when I lived in Hong Kong I had to take the metro to go downtown daily I had contact with hundreds of people moving in the same direction). "Contacts" may be a first step for "social interactions" beyond our family and our intimate, little world. I may make contact daily with hundreds of people, and only have a market exchange (a basic interaction for all of us) with just one or two of them. I don't interact with fellow travelers, I interact with coworkers, colleagues, clients, suppliers.

Social scientists continue to have a hard time defining and classifying relevant social interactions, but to rely on "segregation" is nonsense. Segregation is an outcome of how we make contact and how we interact with other people, and it assumes we are able to distinguish between contacts and interactions. It may be an appropriate word to sell research today but not to understand how "social divisions" in big cities and elsewhere originate and evolve.

Respond

Add Comment

Respond

Add Comment