Metafilter: 56% Conservative Readership

April 29, 2010 11:42 AM

Slate has introduced a tool to analyze the news sites you read online. The tool is based on a paper that studied ideological isolation in news consumption online and off. It analyzes your history to determine which sites you read and looks at readership data to determine how much of an echo chamber, if any, your choice of news sources creates.

You can see for yourself, but the conclusion is that "Many people go to sites whose readers don't share their politics. To use their terminology, they found a low degree of "media isolation" among Web surfers compared with the political isolation most Americans experience in their daily lives. Stacked against the networks in which we work, live, and socialize, the network we increasingly use to get our news—which is to say, the one you are using right now—is relatively integrated."

posted by furiousxgeorge (72 comments total)

This is awesome.
posted by anotherpanacea at 11:47 AM on April 29, 2010

I clicked through to see the list of sites they check for in your history, to see if MetaFilter is on it. It is.
posted by FishBike at 11:50 AM on April 29, 2010

I wish there was a way to get this to pull from my Google Reader. As it is, it has CNN on there, which I think I've visited once to see a picture of bears on a playground set or something.
posted by shakespeherian at 11:51 AM on April 29, 2010

MetaFilter (56% conservative)

Could have fooled me.
posted by I_pity_the_fool at 11:51 AM on April 29, 2010 [23 favorites]

Wow, metafilter is 56% conservative, so pretty close to the middle. Am I reading this correctly?
posted by Think_Long at 11:52 AM on April 29, 2010

PBS readers are 76% conservative, by way of comparison.
posted by swift at 11:52 AM on April 29, 2010

It's an interesting exercise, but seems to be missing some crucial data from the "Profile Me" function, or something. My results only returned 3 news sites visited (including MetaFilter), despite my reading articles via Google News (and MeFi) all the time, meaning that I have visited a lot more than just three news sites in the past while. Slate didn't even list itself in my results, when I know I've read at least one article off of Slate per day in the past week.

Fascinating idea, interesting macro results, but I'm not so sure I trust the individual analysis.
posted by hippybear at 11:52 AM on April 29, 2010

Yeah, are they still compiling stats on the sources? I didn't have the time/stamina to read through all the gritty details (I just came off of The Big Short and my ability to consume and comprehend new facts is australopithecine right now) and that accounting for MetaFilter seems way off. Drudge Report even seems a bit too far right at 93% conservative.
posted by CRM114 at 11:55 AM on April 29, 2010

Seriously, what? @ 56% conservative.

It's interesting, although I'm sure the fact that all this internet content is 'free' skews the whole greatly. I'd really like them to have unraveled the term "media isolation" a bit more. If you see the media itself as an echo chamber, or at least as something on a page / screen, what kind of scale is this really? What about when the internet is a big part of your daily experience? Don't most of us imagine /experience that as greater isolation?

Also, they posit that on the internet we're mixing it up more than in meatlife. But there's also a relatively recent but significant shift towards personalization, and if these two things are at odds, I have a feeling personalization is going to win out sooner or later -- if only because it's a lot more lucrative to advertising.
posted by mondaygreens at 11:56 AM on April 29, 2010

PBS readers are 76% conservative, by way of comparison.

Goddamn socialistswait what?
posted by Think_Long at 11:58 AM on April 29, 2010

Yeah, it certainly has some odd data. Two things to keep in mind: It only looks at Liberal or Conservative, and more people identify as Conservative in general.

It also only looks at home page visits, apparently. So if you hit a deep link to a news story it won't pick it up.
posted by furiousxgeorge at 12:00 PM on April 29, 2010

Wow, they missed a few of the websites I read and I think it skewed the results. It seems like it doesn't take into account how long I spend at a site. I read something at msnbc.com? When? I don't recall doing that.
posted by fuq at 12:03 PM on April 29, 2010

I suspect that at least some of the data that looks funny to me comes from the fact that they don't count anyone who self-identifies as a moderate or independent. It seems likely that many internet users who frequent places like Metafilter would tend to identify themselves as moderates or independents, even if they would be considered more to the "left" on the usual political spectrum.
posted by jefeweiss at 12:03 PM on April 29, 2010

Sites Visited:
MetaFilter (56% conservative)
Slate (49% conservative)
Mail Online (49% conservative)
BBC News (22% conservative)
At these sites, the readership is on average 44 percent conservative, 56 percent liberal

Your isolation index is -41, meaning that, on the bell curve of all readers, your news diet is 41 percentage points to the left.

So, Metafilter is more conservative than the Daily Mail? Little bit skeptical.
posted by TheophileEscargot at 12:03 PM on April 29, 2010 [5 favorites]

For example, I just visited the Fox News home page and swung from -18 liberal to +1 conservative sites. Fox News did not show up in my first profile even though I read stuff there all the time, because I never visit the home page.
posted by furiousxgeorge at 12:03 PM on April 29, 2010

They have the BBC but not the Guardian. Is that too left-leaning for their survey?
posted by hydrobatidae at 12:04 PM on April 29, 2010

Apparently I'm a fascist.
posted by Astro Zombie at 12:04 PM on April 29, 2010 [8 favorites]

Is this *interactive test* something you need to *not use a feed reader with* to understand?

It says I've only read Metafilter from all of those sites listed, but thanks to Google Reader, I've probably read stuff from much on that list, without actually having gone to the site.

In other words, I wonder that since this might tend towards the less technical (not that using RSS feeds makes me a super-techie or anything) may or may not make this skew a certain way.
posted by MCMikeNamara at 12:05 PM on April 29, 2010 [1 favorite]

How did they miss that I read Al Jazeera English? Wonder where they would have put that on the left/right scale.
posted by QIbHom at 12:06 PM on April 29, 2010

The New Yorker is 60% conservative? And here I thought I was being part of the liberal elite.
posted by ecab at 12:08 PM on April 29, 2010 [1 favorite]

I feel like this thing was kind of hastily coded in order to get in on this whole conservative "epistemic closure" thing that's been going around.
posted by Think_Long at 12:08 PM on April 29, 2010 [2 favorites]

MetaFilter (56% conservative)

Could have fooled me.

Not me. A vast majority here support the centrist Democrats.

In determining if I live in an echo chamber, does this site also analyze the utterances entering my ears by way of my nutjob coworkers and racist, homophobic, anti-social (i.e. anti-tax) bumper stickers?
posted by DU at 12:10 PM on April 29, 2010 [1 favorite]

Metafilter is the only site they found for me, which is a little strange since I do read a lot of news from a wide variety of news sites. I hit BBC and CNN every day, for example.
posted by swimming naked when the tide goes out at 12:11 PM on April 29, 2010

Not me. A vast majority here support the centrist Democrats.

I think "supporting centrist democrats" is probably different than your personal political ideology. I would be willing to bet that a lot of us swallow our more liberal urges in the interest of getting some semblance of a return from the far right.
posted by Think_Long at 12:13 PM on April 29, 2010 [2 favorites]

They don't track Talking Points Memo.
posted by gurple at 12:13 PM on April 29, 2010 [2 favorites]

I_pity_the_fool: "MetaFilter (56% conservative)

Could have fooled me."

OMG its the silent majority!
posted by charred husk at 12:13 PM on April 29, 2010

When you're thinking about sources like the BBC and the Daily Mail, I'd bet that the majority of Slate's data comes from Americans, and everyone knows that only crumpet-eating anglophile liberal surrender monkeys read British news. Probably for the soccer.
posted by CRM114 at 12:15 PM on April 29, 2010

FTA:
Conservatives and liberals were the only types counted in the study; self-described moderates and independents were not included.

A bit of dimension reduction going on here, no?
posted by polymodus at 12:17 PM on April 29, 2010

Interesting idea, but it sounds like a pretty poor tool, especially if it doesn't work with RSS readers
posted by delmoi at 12:20 PM on April 29, 2010

Very weird analysis, at least from the Canadian perspective. The Globe and Mail was considered Conservative? Its pretty much Canada's national newspaper - read by virtually everyone - as a guy who self identifies as a socialist, I have no problems calling them fairly middle of the road and moderate.

The CBC commentators are mostly squishy, left-liberal types... and their web site contains a ton of wire stories which probably throws off the accuracy of the entire analysis. Also the good Canadian news blogs aren't big corporate sites which probably came under the radar of this study.

Most of the "liberal" sites rated are really more like multi-edit blogs, which I'd say have a tendency to talk mostly about issues which are US-focused and of not much interest to people outside the United States. What Al Franken or Michelle Bachmann does is USA-insider stuff... most Canadians couldn't identify Anne Coulter, or Glenn Beck. Also "liberal" doesn't equal leftist in Canada...
posted by Deep Dish at 12:21 PM on April 29, 2010

Most news sites online aren't even really regional anymore unless their content is such (for eg Pittsburgh Post-Gazette versus BBC News) The trouble with most surveys, in fact, most of the english speaking internet is that it still assumes it hasn't spread out and reached the world outside the motherland. A conservative in Scandinavia for eg may not be the same as one in the US and certainly a socialist is far more rational than the media would have you believe. This is a facet of the webz that is going to simply have to come up for review soonish - 3 to 5 years if not less, at the rate print has been declining in OECD nations
posted by infini at 12:22 PM on April 29, 2010

The Globe and Mail was considered Conservative?

No no--the research states that it's the readership that is conservative. But even that is skewed: if just 51% of the readers are self-described conservative, then the "Globe and Mail" gets tagged "Conservative" for the purposes of this study. Which is different from the political stance the paper itself.

At least, that's my interpretation of the article. Correct me if I got this wrong.
posted by polymodus at 12:24 PM on April 29, 2010 [1 favorite]

*raises hand*

Right here dude...
posted by republican at 12:30 PM on April 29, 2010 [1 favorite]

Astro Zombie: "Apparently I'm a fascist."

No, no, that's the *underwater* zombies.
posted by mwhybark at 12:30 PM on April 29, 2010 [3 favorites]

This web toy is extremely limited. But that doesn't discredit the research it's based on, which is very interesting.
posted by grobstein at 12:31 PM on April 29, 2010

Metafilter 56% conservative? While I've said for a while that the site's not as diehard liberal as some here find it convenient to say it is, that's still very odd to me. It's possible that the various frequent links to whatever conservative outrage is going on are skewing their reckoning.

What sites are considered liberal by them?
posted by JHarris at 12:33 PM on April 29, 2010

Forget MetaFilter, whitehouse.gov is 58% Conservative, which sounds about right.

Also, The Huffington Post is 70% liberal and 30% tits, which is spot on.
posted by felix betachat at 12:37 PM on April 29, 2010 [1 favorite]

MetaFilter (56% conservative)

Yeah, that surprised the heck out of me too, but remember that they're profiling readers, not posters. I think if they had a way to profile active commenters, it would swing way left.
posted by Malor at 12:39 PM on April 29, 2010

It's possible that the various frequent links to whatever conservative outrage is going on are skewing their reckoning.

They're looking at how the readers of the site report their political slant, rather than the political slant of the site's content. From the paper:

For each outlet in our sample (a newspaper, a particular website), we measure the share conservative: the share of users who report their political outlook as "conservative,"among those who report being either "conservative" or "liberal."

But I guess readers is not the same thing as registered users. Are there a lot of people who visit the MetaFilter home page who self-report as "conservative" but don't actually participate here?
posted by FishBike at 12:41 PM on April 29, 2010

Also there are links to non-political stuff here

So...maybe some conservative people also like cool links

It is not really too surprising
posted by internet fraud detective squad, station number 9 at 12:41 PM on April 29, 2010 [1 favorite]

I mean -- I frequently feel like I'm wildly conservative compared to most posters here, but compared to real conservatives, I'm way, way liberal.
posted by Malor at 12:42 PM on April 29, 2010 [2 favorites]

56% conservative readership isn't that bad when you look at the data they are using. They eliminate independent and moderate. In the general US population...


Two new studies (here and here) conducted by the Progressive Studies Program at the Center for American Progress breaks down the electorate on a new 5-point scale of political ideology that reflects the primary approaches people ascribe to today. Under this schematic, 34 percent of the country self-identifies as ‘conservative’, 29 percent as ‘moderate’, 15 percent as ‘liberal’, 16 percent as ‘progressive’, and 2 percent as ‘libertarian’.

Metafilter is a general interest site, not a political one. AskMe is even more general interest, so it's not like there is much reason to expect an overwhelmingly liberal or conservative readership aside from the fact that the dominant community culture definitely makes it feel that way.
posted by furiousxgeorge at 12:45 PM on April 29, 2010

I'd like to see this concept revisited with an open database of web sites that includes everything, not just news and politics. Something like Last.fm that scrobbles web sites visited instead of songs played. Just like that site can calculate your musical "nearness" to other people, this one would generate a fairly accurate picture of your interests and political leanings, with more dimensions accounted for than the OP's "left/right" slider.

Then again, sharing and comparing your browser history with the world is an idea almost as bad as Blippy. So, uh, nevermind.
posted by The Winsome Parker Lewis at 12:56 PM on April 29, 2010

Also, The Huffington Post is 70% liberal and 30% tits, which is spot on.

Funny, I would have said HuffPo is 70% liberal, 30% tits, and 100% ass.

Also:

Astro Zombie: "Apparently I'm a fascist."

No, no, that's the *underwater* zombies.
posted by mwhybark at 12:30 PM on April 29 [1 favorite -] Favorite added! [!]

Really? Just my favorite? C'mon, people, that was a good line. Shockwaves? No? Fine. Hmph.
posted by Amanojaku at 12:57 PM on April 29, 2010

It's an interesting idea; I can't wait to see how long it'll be long before it's used as more than that. I'm adding this to my list of reasons not to let my browser keep cookies past a single session unless I have to manually save and restore them.
posted by Avelwood at 1:08 PM on April 29, 2010

I'm adding this to my list of reasons not to let my browser keep cookies past a single session unless I have to manually save and restore them.

It's almost certainly not using cookies but one of these CSS visited-link techniques.
posted by enn at 1:10 PM on April 29, 2010 [1 favorite]

I think the Slate demo is using the CSS history hack.

On the general subject of browser-history-based profiling, you might be interested in cssfingerprint. The root interest there is de-anonymizing visitors based on a browser history profile (kind of like panopticlick, but potentially more effective since it is based on your behavior rather than the idiosyncrasies of your browser or computer). Classifying visitors according to ideology is probably easier than trying to identify individual people.
posted by hattifattener at 1:29 PM on April 29, 2010 [1 favorite]

(Ah, on not-preview, what enn said.)
posted by hattifattener at 1:31 PM on April 29, 2010

Wow it says I only go to MetaFilter, and it also says that I should KILLKILLKILL.

I will obey you, computer thingy!
posted by Mister_A at 1:31 PM on April 29, 2010

Smells like bullshit to me. Poor methodology, even poorer implementation.
posted by aeschenkarnos at 1:34 PM on April 29, 2010

This ranked me as being conservative because my regular news sources are so far left that Slate hasn't heard of them.
posted by stammer at 1:39 PM on April 29, 2010

What, are you kidding? Left is the new Right!

/me checks that his shirt is inside-out and backwards
posted by not_on_display at 1:43 PM on April 29, 2010

I clicked the profile button, but all I got was a Java animation of a giant crying panda bear, and a message from the editors of Slate magazine to seriously consider how my browsing choices reflect upon my mother despite all she most likely sacrificed to raise me.

I don't get it.
posted by dgaicun at 1:44 PM on April 29, 2010

I was inclined to believe it until:

Mail Online (49% conservative)

Now come on...
posted by Webbster at 1:48 PM on April 29, 2010

Hmm. cssfingerprint says the best odds are that I'm young, black and rich, whereas I'm actually middle-aged, white and middle-class heading toward poor. Apparently I go to the wrong websites.
posted by Pater Aletheias at 1:48 PM on April 29, 2010

Or your wife...

Nah. Couldn't be.
posted by pracowity at 2:00 PM on April 29, 2010 [1 favorite]

The Atlantic (62% conservative)
Rolling Stone (63% conservative)
posted by kirkaracha at 2:08 PM on April 29, 2010

Thank God I'm only 40% liberal. If they knew I read Counterpunch, Alternet, Truthdig, the Daily Bleed, Al Jazeera and Dissident Voice (none of which Slate monitors), I'd probably be put on the no-fly list.
posted by kozad at 2:14 PM on April 29, 2010 [2 favorites]

The Nation is also left out on here.
posted by HeroZero at 2:44 PM on April 29, 2010

Fake ask metafilter: are there any browsers that do not expose my browsing history to arbitrary web sites in this manner? How is this not a critical security issue?
posted by Wood at 2:58 PM on April 29, 2010 [1 favorite]

Wood, take a look at the CSS history stuff you can find from enn's and my links; it is considered a security issue.
posted by hattifattener at 3:01 PM on April 29, 2010 [1 favorite]

Or you could just surf in "Private Browsing" mode all the time. I do this and these sites couldn't find anything on me.
posted by The Winsome Parker Lewis at 3:03 PM on April 29, 2010

The Atlantic (62% conservative)
Rolling Stone (63% conservative)

Perhaps that has to do with Rolling Stone being culturally associated with baby-boomers, and a significant proportion of those having moved to the right as they got more of a stake in the status quo? I wonder what it says for Playboy.
posted by acb at 4:01 PM on April 29, 2010

Hmm. cssfingerprint says the best odds are that I'm young, black and rich, whereas I'm actually middle-aged, white and middle-class heading toward poor. Apparently I go to the wrong websites.

I'm apparently an 18-34-year-old Asian high-earner. (Though there's a slight tendency towards the 55+, largely due to looking up European train travel websites.)
posted by acb at 4:03 PM on April 29, 2010

Agreed that the Canadian sites are massively distorted. The Globe & Mail, which is generally considered centre-left, came in as 87% conservative. CTV news, which is center to center-left, was also in the high 80s. How did they mess those up so badly?
posted by hiteleven at 5:00 PM on April 29, 2010

On the other hand, the gender guesser guessed my gender correctly, so it wasn't a wasted trip to Slate.
posted by hiteleven at 5:00 PM on April 29, 2010

Blackvoices.com visitors are 63% conservative

On the front page today: "Ludacris' Condom Ads" and "Is the War on Drugs Finally Over?"

Whitehouse.gov readers are 58% conservative

although I've signed up to get all the views of the administration beamed into my frontal lobe via satanic microchip, so maybe only conservatives actually use the website.

What this seems to be teaching me is one of the following: 1) the labels 'conservative' and 'liberal' are considerably different from 'Republican' and 'Democrat', 2) the only people you can really count on to fall completely in line are Fox News viewers, who give Bill O' Reilly's website a creepily complete 100%.
posted by Valet at 7:05 PM on April 29, 2010

It appears that both Safari and Firefox are moving to fix this about a decade after it was first reported.

http://blog.mozilla.com/security/2010/03/31/plugging-the-css-history-leak/
posted by Wood at 9:03 PM on April 29, 2010

Rush Limbaugh: 99% Conservative, 1% Liberal

Looks like it was calibrated properly.
posted by tybeet at 6:15 AM on April 30, 2010

I was surprised to see that metafilter's own metafilter has more daily visitors than salon.com.

Yay team!

(Michael, we're bigger than U.S. Steel.)
posted by Trochanter at 6:26 AM on April 30, 2010

Boston Herald (48% conservative)
Boston Globe (59% conservative)

This is obviously complete balls. Good idea, though. I'd like to see a more plausible version.
posted by CunningLinguist at 7:55 AM on April 30, 2010

The Globe and Mail was considered Conservative?

Things have shifted since the 90's when the conservatives took a nose dive, back then it was definitely a conservative paper. These days I'd have to say it generally sasheys to the left on social issues (decriminalization of marijuana fer instance), but it's still oh so fiscally conservative.
posted by squeak at 9:49 AM on April 30, 2010