Facebook Regions of America
February 9, 2010 8:11 AM Subscribe
Researcher uses data regarding connections on facebook to map distinct regions of the United States.
Stayathomia vs Everythingelse could use a control for inter-city distance averages. People move far away from home in the West because unless you live in the same bedroom you were born in, you have to be far away. There's no middle distance.
posted by DU at 8:25 AM on February 9, 2010 [6 favorites]
posted by DU at 8:25 AM on February 9, 2010 [6 favorites]
This is an interesting start. I'd like to see the same data analyzed more computationally, with quantitative criteria for determining region boundaries. I also liked where he was going with popular fan pages and first names; it would be nice to see longer, un-cherry-picked lists for every area. Any maybe international maps?
posted by The Winsome Parker Lewis at 8:28 AM on February 9, 2010
posted by The Winsome Parker Lewis at 8:28 AM on February 9, 2010
That's a lot more interesting than I expected, but the names for the regions are going to incite a lot of hurf-durfery from the tl;dr set who look at the map and don't read the post.
posted by ardgedee at 8:34 AM on February 9, 2010
posted by ardgedee at 8:34 AM on February 9, 2010
People from here do this, but people from here do this....
posted by jonmc at 8:37 AM on February 9, 2010 [1 favorite]
posted by jonmc at 8:37 AM on February 9, 2010 [1 favorite]
I have to wonder about Facebook as a selection sample. I'm not there, and my connections span 3 continents, because I move around a lot.
posted by Goofyy at 8:38 AM on February 9, 2010
posted by Goofyy at 8:38 AM on February 9, 2010
Re "Mormonia": I didn't expect to see Twilight showing up quite so much though, I have no idea what to make of that!
I feel like there's a second cultural divide here, between people (like the author) who are perplexed by the obvious and others (ME! ME! ME!) who can do a basic scan of Wikipedia.
I have to wonder about Facebook as a selection sample.
It's a convenient database of self-reported information and it includes a larger proportion of the populace than you think.
posted by kittyprecious at 8:46 AM on February 9, 2010 [6 favorites]
I feel like there's a second cultural divide here, between people (like the author) who are perplexed by the obvious and others (ME! ME! ME!) who can do a basic scan of Wikipedia.
I have to wonder about Facebook as a selection sample.
It's a convenient database of self-reported information and it includes a larger proportion of the populace than you think.
posted by kittyprecious at 8:46 AM on February 9, 2010 [6 favorites]
This is pretty cool, but I bet separating out urban/rural populations and college educated/not would make this both more interesting and easier to fight about.
posted by oinopaponton at 8:47 AM on February 9, 2010
posted by oinopaponton at 8:47 AM on February 9, 2010
It's a convenient database of self-reported information and it includes a larger proportion of the populace than you think
Yes, however focusing on individuals who have decided to leave their profiles public (or don't realize they're public) may skew the results somewhat.
posted by Adam_S at 8:49 AM on February 9, 2010
Yes, however focusing on individuals who have decided to leave their profiles public (or don't realize they're public) may skew the results somewhat.
posted by Adam_S at 8:49 AM on February 9, 2010
I also liked where he was going with popular fan pages and first names; it would be nice to see longer, un-cherry-picked lists for every area. Any maybe international maps?
posted by The Winsome Parker Lewis
Yeah, this is just an analysis and overview. If you want more data go here.
posted by vacapinta at 8:49 AM on February 9, 2010 [2 favorites]
posted by The Winsome Parker Lewis
Yeah, this is just an analysis and overview. If you want more data go here.
posted by vacapinta at 8:49 AM on February 9, 2010 [2 favorites]
It's a convenient database of self-reported information...
You just restated the problem.
posted by DU at 8:55 AM on February 9, 2010
You just restated the problem.
posted by DU at 8:55 AM on February 9, 2010
This is surprisingly reasonable and intriguing. And nothing is named Michiana, or Kalamazoo.
posted by The Esteemed Doctor Bunsen Honeydew at 8:57 AM on February 9, 2010
posted by The Esteemed Doctor Bunsen Honeydew at 8:57 AM on February 9, 2010
> Yes, however focusing on individuals who have decided to leave their profiles public (or don't realize they're public) may skew the results somewhat.
The sample set is 210 million profiles. It is a limited, self-selecting population which very likely includes a number of sockpuppets, shills, and dead people, but as a subset of the roughly 300 million people in the country (many of whom can't or won't be on Facebook), it may be more comprehensive than any survey other than the national census.
posted by ardgedee at 8:57 AM on February 9, 2010
The sample set is 210 million profiles. It is a limited, self-selecting population which very likely includes a number of sockpuppets, shills, and dead people, but as a subset of the roughly 300 million people in the country (many of whom can't or won't be on Facebook), it may be more comprehensive than any survey other than the national census.
posted by ardgedee at 8:57 AM on February 9, 2010
You just restated the problem.
Do people lie about where they live and who they know? I think that Adam_S's concern is valid, but otherwise there's not much in this exercise in analysis to get up in arms about.
posted by kittyprecious at 9:03 AM on February 9, 2010
Do people lie about where they live and who they know? I think that Adam_S's concern is valid, but otherwise there's not much in this exercise in analysis to get up in arms about.
posted by kittyprecious at 9:03 AM on February 9, 2010
iEsto es mi cara sorprendida!
posted by Mister_A at 9:15 AM on February 9, 2010 [3 favorites]
posted by Mister_A at 9:15 AM on February 9, 2010 [3 favorites]
This is great, thanks. I love comparing these with the Nine Nations of North America map.
posted by A dead Quaker at 9:20 AM on February 9, 2010 [2 favorites]
posted by A dead Quaker at 9:20 AM on February 9, 2010 [2 favorites]
but otherwise there's not much in this exercise in analysis to get up in arms about.
That's never stopped us before!
Everybody ready? Arms UP!
posted by notyou at 9:28 AM on February 9, 2010 [1 favorite]
That's never stopped us before!
Everybody ready? Arms UP!
posted by notyou at 9:28 AM on February 9, 2010 [1 favorite]
What, people are predominantly friends with people who are near them geographically? Local geographic regions are more often linked to neighboring regions? How..... hamburger.....
posted by cavalier at 9:29 AM on February 9, 2010
posted by cavalier at 9:29 AM on February 9, 2010
Looks like we've borked it. Or, rather, the several dozen sites that discovered this in the last 24 hours have borked it.
posted by gurple at 9:31 AM on February 9, 2010
posted by gurple at 9:31 AM on February 9, 2010
What, people are predominantly friends with people who are near them geographically?
I suggest a second reading.
posted by gurple at 9:32 AM on February 9, 2010 [4 favorites]
I suggest a second reading.
posted by gurple at 9:32 AM on February 9, 2010 [4 favorites]
Not knowing that Stephanie Meyer is a Mormon is a pretty big oversight, but isn't as surprising as considering Phoenix a small town.
posted by Pater Aletheias at 9:40 AM on February 9, 2010
posted by Pater Aletheias at 9:40 AM on February 9, 2010
I'm surprised there are so many open profiles. I wonder what the percentage is? If I just randomly click around friend to friend from my profile, I only see maybe 25% profiles that include location.
posted by smackfu at 9:41 AM on February 9, 2010
posted by smackfu at 9:41 AM on February 9, 2010
Is it just me, or is Atlanta surprisingly connected on the US Cities page? I would have guessed that it was in the St. Louis or Minneapolis range, but it looks like it's just about as dark as Chicago.
posted by dd42 at 9:51 AM on February 9, 2010
posted by dd42 at 9:51 AM on February 9, 2010
I have to wonder about Facebook as a selection sample. I'm not there, and my connections span 3 continents, because I move around a lot.
Yeah, my friends on Facebook (not an unwieldy list -- maybe 200 people) are in twenty-some-odd countries on six continents*, and this does not appear to be unusual among my peers. Conversely, the people I work with every day I do not have as Facebook friends because, well, I see them every day.
*Not anti-Antarcticanist.
posted by ricochet biscuit at 9:51 AM on February 9, 2010
Yeah, my friends on Facebook (not an unwieldy list -- maybe 200 people) are in twenty-some-odd countries on six continents*, and this does not appear to be unusual among my peers. Conversely, the people I work with every day I do not have as Facebook friends because, well, I see them every day.
*Not anti-Antarcticanist.
posted by ricochet biscuit at 9:51 AM on February 9, 2010
Can anyone point me to a reference (dead tree or electronic) on the math behind this? I have some knowledge of computer-sciency graph stuff (minimum spanning trees, etc), but I'm really curious how you go from a collection of (loc1, loc2) and partition them into a set of "most connected" locations.
posted by bonecrusher at 9:57 AM on February 9, 2010
posted by bonecrusher at 9:57 AM on February 9, 2010
I don't think there are as many world travelers as you think, especially in the US.
posted by smackfu at 9:57 AM on February 9, 2010 [2 favorites]
posted by smackfu at 9:57 AM on February 9, 2010 [2 favorites]
...twenty-some-odd countries on six continents...
Well, yeah. Exactly. Those connections, and the similar connections of your peers, fade into a background smear.
But if you pick any three of your local peers, you will all have connections within your geographical area, and perhaps to certain other geographical areas that people in your area tend to connect with.
posted by gurple at 10:00 AM on February 9, 2010
Well, yeah. Exactly. Those connections, and the similar connections of your peers, fade into a background smear.
But if you pick any three of your local peers, you will all have connections within your geographical area, and perhaps to certain other geographical areas that people in your area tend to connect with.
posted by gurple at 10:00 AM on February 9, 2010
bonecrusher, the cluster analysis wiki page might be relevant, if I could understand more of it.
posted by smackfu at 10:01 AM on February 9, 2010
posted by smackfu at 10:01 AM on February 9, 2010
Looking at the Columbus / Charleston thing, I think a map of the interstate highway system would prove enlightening for a lot of these kinds of situations.
I suspect that I'd have issues with the way the data is analyzed if I had a better feel for what they are doing - particularly given this statement: "even New York only has one really long-range link in its top 10". If this describes their methodology then a small town in Indiana where 25% of the links were to other people in town, 25% of the links are to 10 nearby towns and the remaining 50% are scattered evenly around the world would be hugely "stay at homey".
posted by Kid Charlemagne at 10:23 AM on February 9, 2010
I suspect that I'd have issues with the way the data is analyzed if I had a better feel for what they are doing - particularly given this statement: "even New York only has one really long-range link in its top 10". If this describes their methodology then a small town in Indiana where 25% of the links were to other people in town, 25% of the links are to 10 nearby towns and the remaining 50% are scattered evenly around the world would be hugely "stay at homey".
posted by Kid Charlemagne at 10:23 AM on February 9, 2010
I suggest a second reading.
Ok, I realize my original comment was a bit snarky, but I truly did just read this again and I am wondering where you see something different? The datapoints of names and which is a favorite are neat, but I'm confused as to how it's anything other than obvious in nearly all of his geographic observations that the local city centers are highly linked?
posted by cavalier at 10:23 AM on February 9, 2010
Ok, I realize my original comment was a bit snarky, but I truly did just read this again and I am wondering where you see something different? The datapoints of names and which is a favorite are neat, but I'm confused as to how it's anything other than obvious in nearly all of his geographic observations that the local city centers are highly linked?
posted by cavalier at 10:23 AM on February 9, 2010
The datapoints of names and which is a favorite are neat, but I'm confused as to how it's anything other than obvious in nearly all of his geographic observations that the local city centers are highly linked?
You seem to be trying to derive some sort of broad meaningful message from this data presentation, and I agree, there isn't really one there.
But the details of the analysis are neat. I find it interesting that Chicago makes the top-10 list for friends of people in Seattle but Vancouver, Canada, doesn't. And that Spokane is higher up the list than Olympia. Neither of those things is intuitive to me.
His summary of the connected regions is interesting, too. I have a lot of Mormon friends on Facebook (I grew up in Utah), and I think Mormons are disproportionately represented on Facebook. So I suspected that you'd see a strong Mormon Belt connection, but I bet that's not obvious to everyone, and I was surprised at the strength of that belt's ties to Phoenix.
I think you're missing the trees for the forest.
posted by gurple at 10:31 AM on February 9, 2010 [1 favorite]
You seem to be trying to derive some sort of broad meaningful message from this data presentation, and I agree, there isn't really one there.
But the details of the analysis are neat. I find it interesting that Chicago makes the top-10 list for friends of people in Seattle but Vancouver, Canada, doesn't. And that Spokane is higher up the list than Olympia. Neither of those things is intuitive to me.
His summary of the connected regions is interesting, too. I have a lot of Mormon friends on Facebook (I grew up in Utah), and I think Mormons are disproportionately represented on Facebook. So I suspected that you'd see a strong Mormon Belt connection, but I bet that's not obvious to everyone, and I was surprised at the strength of that belt's ties to Phoenix.
I think you're missing the trees for the forest.
posted by gurple at 10:31 AM on February 9, 2010 [1 favorite]
What, people are predominantly friends with people who are near them geographically?
I think the interesting part is how "near them" changes meaning depending on where you are in the country.
posted by smackfu at 10:33 AM on February 9, 2010 [1 favorite]
I think the interesting part is how "near them" changes meaning depending on where you are in the country.
posted by smackfu at 10:33 AM on February 9, 2010 [1 favorite]
Well, cavalier, sometimes using available data to investigate 'obvious' statements will turn up interesting results. The result that Arkansas is closer to Texas than Georgia is kind of surprising. As is the fact that Seattle is pretty much cut off from the rest of the West.
posted by kaibutsu at 10:37 AM on February 9, 2010
posted by kaibutsu at 10:37 AM on February 9, 2010
Yeah, I agree with DU about intercity distance. I live in coastal NH/Southern ME, and the bulk my FB friends range from Philadelphia to NY and NJ, through to VT and Northern New England, west to Wisconsin. Even counting out some of my far-flung friends on the West Coast and in HI and Europe and all that, the distance between me and my nonlocal friends is anywhere from a 3-hour to a 12-hour drive. In East Coast terms, those are different worlds. NYC and Philly, Wisconsin, and Northern New England are all different worlds. I spend so much time on the road between visiting all these people that "Stayathomia" really doesn't feel like what's going on.
posted by Miko at 10:49 AM on February 9, 2010
posted by Miko at 10:49 AM on February 9, 2010
> The sample set is 210 million profiles.
From http://www.facebook.com/press/info.php?statistics
* More than 400 million active users
* About 70% of Facebook users are outside the United States
That's 120 million US profiles. Still large, but under 50% of the US population. Combined with private profiles, and profiles just not providing location information, makes the sample set much less representative of the whole US population.
posted by topogopo at 11:15 AM on February 9, 2010
From http://www.facebook.com/press/info.php?statistics
* More than 400 million active users
* About 70% of Facebook users are outside the United States
That's 120 million US profiles. Still large, but under 50% of the US population. Combined with private profiles, and profiles just not providing location information, makes the sample set much less representative of the whole US population.
posted by topogopo at 11:15 AM on February 9, 2010
Saw it before it got posted here; had half a notion to post it myself, then saw the Mormonia/"Twilight" reference and realized I was dealing with a researcher who was a character in "Big Bang Theory".
posted by oneswellfoop at 11:29 AM on February 9, 2010
posted by oneswellfoop at 11:29 AM on February 9, 2010
The reason it "doesn't look like Washingtonians are big travelers compared to the rest of the West", is because it is so far to anywhere. Outside of the Seattle, Tacoma, Bellevue, Everett area, it is a three hour drive to Portland or Vancouver BC. It is a five hour drive to Spokane. The next major city to the south after Portland is Sacramento, a 12 hour drive. You can drive from Paris to Berlin in a shorter time. To the east, there isn't very much until you reach Minneapolis.
We don't go anywhere, because there is no where to go.
posted by Xoc at 11:32 AM on February 9, 2010
We don't go anywhere, because there is no where to go.
posted by Xoc at 11:32 AM on February 9, 2010
Is it just me, or is Atlanta surprisingly connected on the US Cities page?
Many of us wind up here. Mostly because we worked for the phone company. I'd say easily half of the folks that I work with all started out elsewhere and we're as befuddled as the next man as to how we got here.
"I was born in Pittsburgh, my folks moved to California, then Arizona then back to California. I migrated to Florida and now...(empty, white space)...I live in a brick colonial house in a dense growth tree neighborhood. How the .....? "
At any rate, if anyone has any ideas for how to get out of Atlanta, I'm willing to hear them!
posted by Ruthless Bunny at 11:57 AM on February 9, 2010 [2 favorites]
Many of us wind up here. Mostly because we worked for the phone company. I'd say easily half of the folks that I work with all started out elsewhere and we're as befuddled as the next man as to how we got here.
"I was born in Pittsburgh, my folks moved to California, then Arizona then back to California. I migrated to Florida and now...(empty, white space)...I live in a brick colonial house in a dense growth tree neighborhood. How the .....? "
At any rate, if anyone has any ideas for how to get out of Atlanta, I'm willing to hear them!
posted by Ruthless Bunny at 11:57 AM on February 9, 2010 [2 favorites]
Xoc: But then why isn't Portland similarly isolated?
topogogo: I expect that the sampling is, if not a complete, census-style set of data on the entire US population, at least 'significant' in the statistical sense...
posted by kaibutsu at 12:07 PM on February 9, 2010
topogogo: I expect that the sampling is, if not a complete, census-style set of data on the entire US population, at least 'significant' in the statistical sense...
posted by kaibutsu at 12:07 PM on February 9, 2010
One fascinating thing about the link is that if you click on the map you can sort the connections by country and it will show you about 30 thumbnail profile pics of people in that country. I guess it shouldn't surprise me that people's profile pics in Greenland and Burkina Faso look basically the same as the pics of my (mostly) American friends.
posted by ekroh at 12:20 PM on February 9, 2010
posted by ekroh at 12:20 PM on February 9, 2010
kaibutsu: "Stasitical significance" means "unlikely to have occurred by chance". Whether the sample set is representative as a whole is unclear...
posted by topogopo at 12:24 PM on February 9, 2010
posted by topogopo at 12:24 PM on February 9, 2010
Huh — the interesting result for me (like for most of the other people here, it seems) is the way Seattle's isolated... I grew up there, and since the majority of my close friends ended up moving to either L.A. or the Bay Area I expected to see Seattle and what he's calling "socalistan" tightly interlinked. Maybe what I've observed is just a side effect of my friends being for the most part hacker types. I hypothesize that if one were to make a version of this map selecting only people who list themselves as working in tech-industry jobs, Seattle would show up as being practically a suburb of San Francisco.
posted by You Can't Tip a Buick at 12:29 PM on February 9, 2010
posted by You Can't Tip a Buick at 12:29 PM on February 9, 2010
topogogo: The sample size very large from a statistical point of view (what is the sample size of a Gallup Poll, for instance). Where the sample set is taken is more relevant. This is a representation of the Facebook population of the United States. What does that mean, exactly? I could say that in general this only represents the connections with young, internet-savvy people. Immediately I have removed from this sample set those of the United States who do not have a computer (or visit a library or cafe to use one) and those who do not have the time or inkling to be on Facebook (there goes the population in poverty and a large number of elderly Americans). What is the ethnic disparity for a Facebook population versus the US census population?
This is an interesting look at the Facebook Regions of America, as titled by the OP far more accurately than PeteSearch's title.
posted by linux at 12:31 PM on February 9, 2010
This is an interesting look at the Facebook Regions of America, as titled by the OP far more accurately than PeteSearch's title.
posted by linux at 12:31 PM on February 9, 2010
One fascinating thing about the link is that if you click on the map you can sort the connections by country...
Oh, I missed that on the first go-round. If you mouseover a country, it highlights the top six other countries that members have connections to. For the U.S., I definitely would've guessed the UK and Canada, and probably Australia and the Philippines (and less-possibly Israel), but Indonesia's kind of unexpected...especially being ahead of both India and Mexico.
posted by kittyprecious at 1:08 PM on February 9, 2010
Oh, I missed that on the first go-round. If you mouseover a country, it highlights the top six other countries that members have connections to. For the U.S., I definitely would've guessed the UK and Canada, and probably Australia and the Philippines (and less-possibly Israel), but Indonesia's kind of unexpected...especially being ahead of both India and Mexico.
posted by kittyprecious at 1:08 PM on February 9, 2010
Oh, I missed that on the first go-round.
Not just you. It hasn't loaded for me in the entire time this post has been up.
posted by smackfu at 1:18 PM on February 9, 2010
Not just you. It hasn't loaded for me in the entire time this post has been up.
posted by smackfu at 1:18 PM on February 9, 2010
Phoenix is a small town?
posted by reductiondesign at 1:21 PM on February 9, 2010
posted by reductiondesign at 1:21 PM on February 9, 2010
Not just you. It hasn't loaded for me in the entire time this post has been up.
It loaded and loaded and loaded for me on my Linux box, but on my Mac it popped right up.
posted by gurple at 1:32 PM on February 9, 2010
It loaded and loaded and loaded for me on my Linux box, but on my Mac it popped right up.
posted by gurple at 1:32 PM on February 9, 2010
I think you're missing the trees for the forest.
Right on -- I get what you're saying. Different strokes / folks here; not trying to say I'm a sociological wunderkid, but I found the details remarkably.. well.. obvious? But let's say I'm over simplifying the result set -- it's still scratching the surface of what could really be interesting if we branched all of these connections out and then tried sorting them by patterns, fandoms, behaviors, what have you.
I've been trying to reload it every hour or so hoping to get the full data set. It's definitely rich!
posted by cavalier at 1:47 PM on February 9, 2010
Right on -- I get what you're saying. Different strokes / folks here; not trying to say I'm a sociological wunderkid, but I found the details remarkably.. well.. obvious? But let's say I'm over simplifying the result set -- it's still scratching the surface of what could really be interesting if we branched all of these connections out and then tried sorting them by patterns, fandoms, behaviors, what have you.
I've been trying to reload it every hour or so hoping to get the full data set. It's definitely rich!
posted by cavalier at 1:47 PM on February 9, 2010
Its very cool and fascinating, but I notice that there are some egregious misclassifications due to US cities/states with the same name as cities and countries outside the US. For example if you click on Georgia (the US state), the number one favourite thing is something in cyrillic? and the list of top ten names look pretty unlikely for the state of Georgia. Anyway I theorise this is because he's mistakenly reading facebook users in the country of Georgia as being in the US state of Georgia.
posted by Joh at 2:26 PM on February 9, 2010 [1 favorite]
posted by Joh at 2:26 PM on February 9, 2010 [1 favorite]
That's 120 million US profiles. Still large, but under 50% of the US population. Combined with private profiles, and profiles just not providing location information, makes the sample set much less representative of the whole US population.
A well-selected sample size of .01% of the U.S. population would do nicely for most statistical purposes. At even 10% you've got an insanely large sample size that might even drown out some selection problems. While it may skew young and middle class, I doubt it skews regionally or by geographic mobility. Combine that with a fairly well-normalized notion of acquaintance and it seems pretty likely to me that for the purposes of this topic, the dataset is probably not only pretty good but beyond what a lot of social science researchers are normally able to gather.
posted by weston at 2:35 PM on February 9, 2010
A well-selected sample size of .01% of the U.S. population would do nicely for most statistical purposes. At even 10% you've got an insanely large sample size that might even drown out some selection problems. While it may skew young and middle class, I doubt it skews regionally or by geographic mobility. Combine that with a fairly well-normalized notion of acquaintance and it seems pretty likely to me that for the purposes of this topic, the dataset is probably not only pretty good but beyond what a lot of social science researchers are normally able to gather.
posted by weston at 2:35 PM on February 9, 2010
I doubt that Facebook users skew all that young - lots of older users find it easier to understand the walled garden than the wilds of the internet. I was sure someone had linked to the age demographics of Facebook before on MeFi, but I can't find it now.
posted by harriet vane at 4:17 AM on February 10, 2010
posted by harriet vane at 4:17 AM on February 10, 2010
Its very cool and fascinating, but I notice that there are some egregious misclassifications due to US cities/states with the same name as cities and countries outside the US. For example if you click on Georgia (the US state)...
Wait...he's just scraping names?
Does that mean that when I click Spain I get people who chose their location as "Spain" rather than people who chose "España"? So, we get expats instead of locals. That would explain why everyone has so many friends in the US...
posted by vacapinta at 4:56 AM on February 10, 2010
Wait...he's just scraping names?
Does that mean that when I click Spain I get people who chose their location as "Spain" rather than people who chose "España"? So, we get expats instead of locals. That would explain why everyone has so many friends in the US...
posted by vacapinta at 4:56 AM on February 10, 2010
« Older Born from jungle techno, the amen break, hip-hop... | Wanna fancify your living room? Newer »
This thread has been archived and is closed to new comments
(Ctrl +)
posted by brandman at 8:22 AM on February 9, 2010 [6 favorites]