Measures of cultural distance

A new paper with many authors — most prominently Joseph Henrich — tries to measure the cultural gaps between different countries.  I am reproducing a few of their results (see pp.36-37 for more), noting that higher numbers represent higher gaps:

Distance from the U.S.

Algeria: 0.15

Australia: 0.03

Brazil: 0.07

Canada: 0.02

China: 0.17

Ecuador: 0.12

Egypt: 0.24

Ethiopia: 0.14

Georgia [the country]: 0.15

Hong Kong: 0.09

Indonesia: 0.19

Japan: 0.11

Malaysia: 0.12

Nigeria: 0.15

Switzerland: 0.06

Egypt is the most distant, then Yemen, with Canada as the closest.

As for cultural distance from China, we have:

Great Britain: 0.20

Hong Kong: 0.09

Japan: 0.14

Russia: 0.09 (not *so* far away)

Taiwan: 0.10

Vietnam: 0.06

Overall the numbers show much greater cultural distance of other nations from China than from the United States, a significant and under-discussed problem for China.  For instance, the United States is about as culturally close to Hong Kong as China is.


America is a viral culture. We make anything popular or sellable part of our culture. Pizza, sushi, hip hop, irreverence, democracy (significant since we are a constitutional republic), even ironic representations of our own culture (MURICA!). It doesn't surprise me that we are a more central culture than the Central Kingdom.
Consider K-Pop, which is incredibly popular beyond Korea. But what is by far the most popular Korean song? Gangam style! Now which is more "American", the average K-Pop mega hit or Gangam Style? There is no coincidence the more popular one seems more American.

?? Gangnam style is in Korean lyrics. And you misspelled it, not 'Gangam' (sic).

Bonus trivia: a temporarily sick mind often makes spelling mistakes, a social study paper found, and that's how a recent smartphone app tracks mental health.

I think that spelling Nazi's are the one's with sick minds.

It's "Nazis" and "ones" you fool

It's the people's 'postrophe and you'll learn to use it soon enough.

So, for Algeria - are we measuring the cultural distance of Arabs or Kabyles? Or do we just ignore the different cultures ones finds in a geographic area in favor of one culture? - after all, what is the cultural distance between the Amish and the Acadians (or Cajuns if American)?

Random samples that are as representative as they can get, I'd guess (and that's tough to do). Measurement is supposed to be affected by diversity such that more diverse groups are downweighted in their distance from others though (arguably this makes sense, arguably it does not).

Take India - what is the culture of India? And if the attempt is to mash all the distinctive cultures within India to a single 'Indian' one, what is the possible value of such an artificially contrived measure?

Even Switzerland is funny, because though one can certainly talk about a Swiss identity, the idea that the French speaking Swiss and German speaking Swiss share a culture is somewhat iffy.

Then take the Belgians.

Using borders drawn on maps as a way to define the seemingly single culture inside seems simplistic in many cases.

Not that something being overly simplified so as to fit into an arbitrary model has apparently ever bothered an economist.

I don't see a table of all-to-all distances, but surely they can generate one from their data. Surely the distance between France and Netherlands gives you an upper bound for the error bar on the Belgian location due to ignoring the great divide. I imagine it's not large.

It would indeed be interesting to see comparable data for sub-groups of India, geographic or, erm, class-based. My guess is that they would vary on these measures much less than the food & language divides would lead you to believe...

'Surely the distance between France and Netherlands gives you an upper bound '

Why? Belgium has its own history, and in terms of religion - well, Belgium is not really in the Dutch or French mold- Both of whom, through separate paths, are quite secular in many ways that the Belgians share no historical connection with - after all, religion is one reason that Belgium is not part of the Netherlands, and Belgium most definitely did not go through the radical secularization represented by the French Revolution.

Not to say that such a method might not be valid. However, apparently, the authors of the study do try to make some allowance for intra-regional variance. Which would almost seem to defeat the point of comparing Belgium to Algeria, admittedly.

As for cultural distance from China, we have:
Hong Kong: 0.09
Russia: 0.09 (not *so* far away)
Taiwan: 0.10

And that's how we know that this study is bunk.

Good catch. I don't believe these numbers either. Also Brazil, a rather strange place culturally for Americans I could imagine (TR?), gets the same mark as Switzerland.

I'm not so quick to dismiss these out of hand. There are reasons that could support the view that HK and Taiwan have drifted away from mainland China: Japan's colonial history in Taiwan, the UKs colonial history in HK, 80 years of divergent economic growth in both cases, the influence of communism over the longer term in China, and durable language differences in HK which few on the mainland share outside of Guandong province. And both territories share a sizeable antagonistic attitude toward China which is based on some fundamental differences of opinion.

Put this another way, what is the cultural difference between north and south korea? Or between east germans and west germans?

Right. If one was not overly influenced by Sino-centric propaganda, based on the last 100-120 years, the most natural way to think of Taiwan and HK would be as former Japanese and British colonies, respectively.

Well, then let us be a bit more precise, as Taiwan was a Japanese colony to the end of WWII, then became a Chinese Nationalist colony for approximately the next two generations.

Obviously the Communists and Nationalists were (and remain) different, but it is not that both of them are somehow not culturally Chinese. What makes Taiwan confusing is the large number of people who had two different colonials overlords in succession.

Have you met any Hong-Kong-ese? My quick impression based on manners & modes of interaction is that they are easily mistaken for Japanese with better English skills, not chinese. So I'm not surprised that this kind of survey would notice that. (Obviously there are also senses in which they are 100% chinese!)

I'm not sure what Hong Kong has to with Taiwan's recent history of being colonized by both the Japanese and the Chinese Nationalists.

Or possibly, the threading was a bit off.

They could also be underestimating the cultural similarities between China and Russia. People forget that Russia extends all the way to the Pacific and shares a border with China.

Given the large number of characteristics underlying these distances I believe that countries with similar distances could also be far apart from each other. How many cities are within 300 miles of St. Louis?

A third pole in this study, besides US and China, was Yemen, which seems rather plausible.

I thought the technique, borrowed from Cavalli-Sforza's old genetic studies, was clever.

I forget, though, how they weighted the components of their estimates. Did they just weight each input equally? That's not a bad way to do it, but of course the numbers could and likely would change if you did additional studies focused on cultural traits that make, say, Burma or Bolivia or Tonga or Congo distinct from other countries.

@SS - yes, good point. See, "The AI Delusion" by Gary Smith, (c) 2018.- Principal components and factor analysis are similar to ridge regression in that they are based on the statistical properties of the variables, with no concern for what the numbers represent. Because principal components and factor analysis are so similar, I will just discuss principal components. Suppose we are trying to predict the chances that a person will be involved in a car accident and we consider these five variables:

F = gender, 1 if female or 0 if male
Y = age if under 30, 0 otherwise
O = age if over 60, 0 otherwise
T = number of traffic tickets
D = average number of miles driven each year

If these variables are correlated, then it is somewhat redundant to look at all five. Instead, we might use principal components to determine two weighted averages:
C1 = 0.30F + 0.01Y + 0.26O + 0.05T + 0.31D
C2 = 0.26F + 0.19Y + 0.24O + 0.12T + 0.14D


Hard to know what to make of a study that shows Russia to be culturally closer to China than is Taiwan. I haven't read the study, so I can't say it's wrong. On the surface it seems almost as implausible as arguing that Afghanistan is culturally closer to Sweden than is Denmark.
But as Tyler recently noted in another post, "culture" is one of those slippery concepts that's hard to pin down, so who knows?

About the same as the polity-4 data which shows that Israel and Zimbabwe have the same quality of government. I am writing a book chapter that needs measures of political and economic freedom and I threw that one in the garbage.

Doesn't polity-4 simply show a rough approximation of how democratic/open the goverment is? It does not need to correlate to "quality". After all, ancient Athenian democracy (at the historical point of just one polis) would measure about the same as current USA system. Even though logistics and robustness is quite different. From what I understand, you are not viewing this data in the correct angle (albeit the data is quite shallow, that's true).

Maybe the weights for public acceptance of street defecation, shoving elderly women on the metro, spitting, and holding one’s child while they urinate on a wall downtown were set too high.

That would put the US and Taiwan pretty close.

You pick some outlier behavior that has nothing to do with culture, and in reality only a negligible number mainland Chinese would actually do (remember there are about 1.37 billion of them), and totally ignore the real culture factors such as language (both speaks Chinese), moral values (both under the influence of Confucianism, despite the "Culture Revolution" in mainland China), and religion (Buddhism vs Christianity) etc.

Takes about 1-2 seconds to spot the mainlanders on a street in Taipei. If you think it’s a negligible percentage, you need to visit China and Taiwan again.

If behavior is separate from culture then we’ve redefined culture to be meaningless.

Henrich isn't even listed first, yet you decided to give the bulk of visibility to him. Bad form, Tyler.

The Fourth Law from 'The Formula: The Universal Laws of Success':

"While team success requires diversity and balance, a single individual will receive credit for the group's achievements."

A book I read on the recommendation of this blog, by the way.

I find that study to be robust, but it has some very strange results.

As a Russian, I find it quite interesting, that Russia is culturally one of the closest to China, however that distance is the same as Russia-USA distance. An anecdotal evidence: my son is learning both Chinese and English (I am heavily westernized by Russian standards).

Would all those saying the paper is wrong please present their data, methods, calculations and results.

Why? Disproving the epicycles of the Ptolemaic Model involves simply accepting that the Earth is not the center of the solar system, but the Sun.

If one wishes to believe that a nation state like India has a single culture (something that the Hindus and Muslims within India's borders just might dispute, to provide one historical example that continues to the present), then it is much like using epicycles to explain observed reality. The model itself fundamentally flawed.

On that point, the study also has results for within-nation regional variation, and finds that "India, 'the country of a hundred nations and a hundred tongues, of a thousand religions and two million gods' (Mark Twain, “Following the Equator”) has the largest mean regional diversity" of all countries, followed by the EU (not a country). The vector of diversity isn't just religion, though, like you suggest.

Thanks for the info - even with javascript on, nothing was offered for download/viewing.

Religion was just one example. One assumes that the differences between Keralans and Punjabis are as roughly 'religious' as the differences between Hindus and Muslims. The conflict between Hindus and Muslims is why the India that gained independence from the UK was split into two countries. Of course, the difference between Urdu and Hindi should not need to be mentioned in this comment section, Urdu being the language used by the invaders. Then used by the next wave of people to rule India as the preferred way communicate from and to English.

Urdu being the language used by the invaders

This is wrong. Urdu is an indigenous (north) Indian language. In spoken form, it's virtually identical to what we call "Hindi" now. Both Hindi and Urdu sprung out of middle Gangetic-plain dialects; it's just that the former uses more Sanskrit vocabulary while the latter uses more Farsi vocabulary. The literary forms of the language can sound quite different, yet are very mutually intelligible.

Indian history is fascinating, and here what is written about the Hindi-Urdu controversy in wikipedia - 'Urdu, along with English, became the first official language of British India in 1850. Urdu was being written, spoken and translated to and from English in all courts, schools, official documents, and government institutes. Although the need to have a language for Hindus developed in the 1850s, the irrevocable momentum of the Hindi language movement occurred around 1880. Urdu is a name derived from the Turkic word ordu (army) or orda. When the Turco-Persian Mughals who invaded the Indian subcontinent made camps, the colloquial Persian spoken by the soldiers, known as the "language of the camp", or "Zaban-i-Ordu", became gradually more and more Indianised, as they themselves settled and married into the local communities, until the dialect they spoke was readily intelligible to the native Hindu population.'

Oddly, though, Indian and Pakistanis do not seem to believe they share a language (the simple difference of the names might be a hint). Possibly in the fashion that the Dutch do not actually share a language with the Belgians, if the somewhat snarky opinion of several Dutch (to German) translators concerning Flemish is any guide.

Clearly, language might not be the right word, yet if among two groups of people speaking a language, a number consider 'their' language different, it is hard to judge from the outside. As noted, the Dutch speakers I have talked about on the subject of Flemish do not consider it Dutch (apart from undeniable historical connections), even if clearly there are not two separate languages involved.

And one Dutch person I know recently recounted a story of being pulled over by Belgian police - though they were clearly Flemish, they only used French to officially deal with the Dutch traffic violators. That is clearly a case of two different languages - and of the police being as intentionally rude as possible to someone that they reasonably would believe considers Flemish inferior, if at at least intelligble.

This can go the other way too, of course - German speakers feels they share a language, even when they cannot understand each other due to accent and dialect (this region in Germany can more or less deal with Schwitzerdeutsch or Austrian German, but that definitely does not apply to people from Hamburg or Rostock).

You don't understand the burden of proof in academic research. It is not necessary for a critic to present data, methods and calculations. It is only necessary to find a single counterexample, absurdity, false premise, or logical flaw.

Some Canadian friends and myself often discuss that we are surprised how American Australia and Australians can feel sometimes, given the geographic distance from the U.S. At least among a small Canadian network of mine, we tend to think Australia is more culturally close to the U.S. than Canada in some, but not all, regards.

Well, as an Australian I have to say Canada is really weird. Canadians touch each other all the time. Just have a conversation with a Canadian involves more physical contact than sex does in Australia, as we try to minimize it on account of the heat.

As an Australian / Canadian dual citizen who currently lives in America, the idea seems ridiculous. Someone from Manitoba is essentially indistinguishable from someone from Minnesota. Of course a Montrealer is different from a Texan though. And a Montrealer is also different from an Albertan. Being American is not just one culture. Canadian is actually a part of North-American culture. I reject the idea that border is a big deal culturally. Sure it is politically.

When I have been in Australia for a few years, and arrive in the USA I feel as though I am basically home in Canada. When I have been living in North-America, arriving in London makes me feel like I am basically home in Australia. I'm going to get into a car on the correct side, people are going to speak to me in essentially my own language usage, the taxi driver is talking to me about the cricket and the rugby, I am going to get offered a cup of tea somewhere, the bakeries are selling sourdough bread. For take away food, do I feel like grabbing a pie and sauce or stopping by a fish and chips shop?

It seems strange to me that this study makes Australia much closer to America culturally than Britain.

K-pop is a J-pop with Korean lyrics, and J-pop is Anglo-American pop with Japanese lyrics. Nothing remotely indigenous about either J or K pop. You'd have to search long and hard to find a Japanese or Korean person who actually prefers their own indigenous musics. Japanese enka (演歌) is also Anglo-American pop with Japanese lyrics.

Have they got the questions right, are the weighting them right, etc? People will say they have or have not based on whether it conforms to their intuitions, which is the problem with these things.

My sense is China today is closer to the US in the 60s than Trump's America is to the US in the 60s, but Trump's America is closer to China in the 60s.

Right. One man single-handedly changed the culture of 350 million people. And he did it overnight, in your mind, I'm sure.

You're not deranged at all.

"Cultural gaps"? Wikipedia: A culture gap is any systematic difference between two cultures which hinders mutual understanding or relations. Such differences include the values, behavior, education, and customs of the respective cultures. It's the gap not the culture that is being considered; thus, the culture in China and Taiwan could be similar but the culture gap between the two (i.e., the differences that hinder relations) could be large. That's the way I interpret what's being measured here. Here at home, the culture of Republicans and Democrats is very similar, but the culture gap is large.

Interesting that China and the US are both equally distant from Hong Kong (9 bps) but almost double that distance from each other (17 bps). Geometrically, this would mean that HK, China and the US would lie on a straight line with HK in the center.

It would be cool to see these plotted in two or three dimensions. It would help us understand the methodology better.

I'm only vaguely familiar with Brazil and surprised it is as close to the US as it is. I wonder where Great Britain is compared to the US.

Quantifying cultural distance, and thus culture itself, with a single value is a fool's errand.

This is ostensibly an economics web site - reducing things to a single value is the sort of errand that economists love to perform.

For rigor, they should have researchers from each pole conduct a similar study, using the same statistical methods, but allowing each culture to select its own traits to use in the study. My concern is, how much of American culture is coded into the traits they used? What constitutes the culture (or significant enough aspects to include) is part of the culture.

At the very least, having the other three poles run the study will give us different perspectives, which is valuable in a study like this.

What are the numbers for the cultural distance between the US and Mexico and Russia?
I'm curious to know if Russian or Mexican immigrants would be more easily culturally assimilated into the US.

Ahhh.... it has the full PDF available.

Mexico: 0.075
Russian: 0.092

Wow, aside from the fact that the abstract claims psychological not cultural value in such a reductive approach, how do you weight the various elements of the vector? How sparse is the vector? Decades ago there was an interesting (to me) article in Science mag. about scales and scaling. Turns out it matters how you scale your observations. The difference between listening to sounds at db 10 and db 20 is quite a bit different than listening to sounds at log(db) 10 and log(db) 20. Try it, if you don't believe me. Given the (probably) non-linear, discontinuous, and non-monotonic nature of the underlying observations (not to mention the probable non-gaussian nature of their distributions), I find the idea that useful information on population-level differences can be determined this way. (Although, there may be useful information on the observer and his/her choices)

It takes a certain kind of academic to come up with a measurement that ranges from 0.02 to 0.24.

I'm surprise that Japan is not the furthest from us. Japan seems so odd.

Japan is so Americanized. I've been here since 1983, minus ten years in Korea, Thailand and Brazil. It seems more like America than America, where I went to grad school and occasionally visit. Much more like America than Brazil, England, Germany, Korea, or France for example, where I also worked/lived. Thailand however is less Americanized.
Korea was half America, half Japan, last time I was there. Suspect it still is.

A model that places Sweden closer to the US than Norway suggests a need to return to the drawingboard.

Comments for this post are closed