The "Social Connectedness Index" data is now available globally and more disaggregated ([email protected]).

It captures the likelihood that individuals across region-pairs are friends on Facebook

This allows to better understand the determinants and effects of social networks

As an example, let's compare the social connectedness to two Chicago-area zip codes:

Bronzeville (60653): Med HH Inc: $28k, 4.5 non-hisp white

Wilmette (60091): Med HH Inc: $148k, 82% non-hisp white

--> Very different structure of networks

Similar patterns emerge when we plot the connectedness of the same two Chicago-area zip codes to zip codes in New York City. Wilmette is connected with wealthy parts of Manhattan, Bronzeville with zip codes with large African-American populations.

Things get even more striking when we move to the national level.

Wilmette has many links to the coasts, to fancy holiday resorts across the country, as well as to college towns:

Bronzeville has few links to the coasts, but instead strong links to the South, likely a persistent feature of the Great Migration.

The SCI data now covers most of the rest of the world. Let's compare the social connections of San Francisco County, CA, and Kern County, CA, first to North America.

SF more links to North East

Kern more links to Mexico, Oklahoma (Dust Bowl Migrants) and North Dakota (oil)

And now globally:

San Francisco is much more connected to the rest of the world than Kern County is.

Many more interesting patterns globally. For Europe, the data exist at the NUTS2 and NUTS3 levels.

Here are social connections to Limburg, a Dutch-speaking province of Belgium

Compared here to the friendship links to Namur, a French-speaking province.

And here, one more: The links to Johannesburg, suggesting substantial migration of South Africans to work in the U.S. oil industry

As we have shown in earlier work, this rich data is helpful for understanding a large number of economic and social outcomes, such as patent citations, migration, and trade flows.

http://pages.stern.nyu.edu/~jstroebe/PDF/BFKS_SocialConnectednessUrbanAreas.pdf

http://pages.stern.nyu.edu/~jstroebe/PDF/BGHKRS_InternationalTradeSocialConnectedness.pdf

http://pages.stern.nyu.edu/~jstroebe/PDF/JEP_SCI.pdf

But we think that it likely has uses across many other fields: Sociology, Political Science, History, etc.

--> Please make your networks in those fields aware of the availability of the data

A natural area of application these days: Epidemiology

While we are not epidemiologists, we have worked on a "proof of concept" to highlight the potential usefulness of these data for COVID-19 modeling (with all the appropriate caveats when you venture outside your core area of expertise):

http://pages.stern.nyu.edu/~jstroebe/PDF/SCI_and_COVID.pdf

We find that those counties with stronger friendship links to Westchester County, NY (an early US COVID-19 hotspot) had more subsequent COVID-19 cases (with or without controlling for variables such as population density or distance to Westchester).

We find similar patterns for social connectedness to Lodi province in Italy (an early hotspot in that country) and subsequent COVID-19 cases).

What this is trying to achieve:

The underlying data for all of this is accessible to researchers and non-profits by emailing [email protected]

More details are here: https://dataforgood.fb.com/tools/social-connectedness-index/

And the original paper introducing the index:
https://pubs.aeaweb.org/doi/pdf/10.1257/jep.32.3.259

Happy to help anyone hoping to understand this data

Some more fun maps: Comparing the local and national social networks of an Upper East Side zip code (10021) to an East Harlem zip code (10035).

Social network of Upper East Side

Social network of East Harlem

And a different part of the world. Social connectedness to South Asia from West Delhi and Mumbai City. @arpitrage had some interesting thoughts on this one.

And one more comparison between the global social connections of London boroughs Brent and Tower Hamlet -- showing importance of immigration links.

Brent has sizable immigrant communities from Brazil, Romania, Somalia and Afghanistan.

The social network of Tower Hamlets shows strong links to Bangladesh (Bangladeshi is largest single ethnicity, with 32% of borough).