Thursday Morning Graph Wonkery

Normally I don’t bother checking Kevin Drum’s graph work because he has done it professionally for longer than I have, and he seems quite good at it. The rare exception is a post last night that I think gets a pretty important point wrong.

Recently a British firm tracked the most influential voices in the health care debate. Here is the graph, with influence on the vertical axis (higher is better) and positive/negative opinion of the British NHS on the horizontal. Kevin added the ‘liberal blogosphere’ venn circle in post.

blog_market_sentinel_healthcare

Kevin makes two conclusions concerning netizens like us. He finds the influence of Twitter depressingly high, and and he finds the liberal blogosphere’s influence depressingly low. Here is why I think that neither conclusion holds water.

About Twitter, it is slightly unfair to compare an individual news outlet such as the Guardian to Twitter, which is basically a communication medium. Since the British firm almost certainly used online linkage and quoting to estimate influence* and Twitter is used by practically everyone who has both a pulse and an internet connection, graphing the influence of Twitter strikes me as about as valid as graphing the influence of email.

About the liberal blogosphere, that dot cluster leads me to the exact opposite conclusion from what Kevin reached. The higher dots, presumably Kos, TPM and HuffPo, sit uncomfortably close to the nearest major news outlet. If I were a WaPo writer trying to pretend (like they always do) that online liberals don’t matter I would not like that graph at all. More fundamentally, Kevin commits the multiplication fallacy by comparing the average influence score of a liberal blog (the center of Kevin’s oval) with the influence of any individual news outlet. Kos may not have the influence of the Washington Post (yet), but established liberal blogs outnumber commercial news outlets by more than one factor of ten. If you want to treat the liberal blogosphere as an aggregate entity, then on aggregate the liberal blogs have a hell of a lot of influeence. If you multiply the number of dots down there by their average influence (assuming the Y axis values are scalar, which is probably wrong), the ‘liberal blogosphere’ data point looks like it could give Twitter a run for its money. Given how those dots can coordinate on an important issue, this graph tells me that liberal blogs have come awfully close to being a new wing of the fourth estate.

BTW, where is the conservative blogosphere? Maybe the analysts did not feel like using a log scale.

(*) Full disclosure – I did not dive into the internals because, unfortunately, I am not paid to blog and what I am paid to do is kicking my ass right now.

***Update***

Also, did you notice Stephen Hawking made it halfway up the Y axis? He is one guy who made something like two public comments on the matter. That suggests that there is a ton of untapped potential for celebrities (other than Jon Voigt) who want to spread the word.

***Update 2***

Another thing. Is it true that the New York Times presents a dramatically more positive picture of the NHS than the liberal blogosphere does? Paging Brent Bozell…

Share On Facebook
Share On Twitter
Share On Google Plus
Share On Pinterest
Share On Reddit

27 replies
  1. 1

    I’d like to believe you on the bloggy influence. (I do believe you re Twitter. I think you are right on there.)

    I certainly agree that blog influence is greater than implied in the graph above (and I’m working on a post that talks about the phenomenon by which a seemingly formal presentation of data passes for an actual meaningful argument too often in our discourse right now (perhaps it was ever thus…).

    But my own recent experience with book publicity (also to be blogged about when what I get paid for stops kicking my ass) suggests that the blogospheric influence still has a way to go to match the MSM, even when eyeballs and or multipliers point towards a greater impact.

    Long story short — most of the MSM has lost the ability to move the book trade (exceptions being The New York Times Book Review, The New Yorker, and NPR); the blogosphere has a lot of voices and powerful link connections — but my experience is the wide, or at least detectable, notice on the blogosphere has less influence than one short hit on Ira Flatow’s show. The old way of creating a conversation about a book is gone, mostly; the new media have not yet managed to replace it.

    The problem with this anecdote: books are themselves a relatively minor medium right now. Politics is where the blogosphere has truly cut its teeth. The major player set out to have influence and they clearly have. But I think that the appearance of a mass medium with all its power to move the dial in practices is still running ahead of the fact of power on the ground.

    FWIW.

  2. 2
    NobodySpecial says:

    As the quote says, ‘When a hundred thousand assholes speak with one voice, even the King must pay heed.’ Or something like that. The funniest part is that the blogs are starting to coalesce into seperate regions – I see a lot of people that read Kos who also DON’T go to FDL or Huffington Post much. I’m sure the example cuts both ways, and what you may get is a more scalar effect than might be considered.

  3. 3

    For the technically minded, this means that we crawl the internet looking for pages which are about the topic, then we track mutual references between people, institutions, entities mentioned in the context. The resulting structure gives us a mathematically verifiable measurement of “authority” in the context.

    I question their use of the term “authority,” and the conclusion drawn. Pretty graph, but mostly useless. “Authority” is not a concept that can be tracked easily through their method.

  4. 4
    Napoleon says:

    I thought Drum/Market Sentinel’s conclusions are complete BS.

    @Thomas Levenson:

    I have often wondered what kind of bump some of the more popular shows on NPR would give a book or musician and have assumed it was appreciable since NPR actually has a huge audience of almost exactly who you would want to talk to if you had a book or slightly out of the mainstream music.

  5. 5
    Suicidal Zebra says:

    That’s bad news for us bloggers. On the other hand, is Twitter really the most influential medium out there? Seriously? And in what way is the UK Conservative Party influential in an American debate about healthcare? For that matter, does trawling the web for references to the NHS really tell us anything at all about how Obama is doing on healthcare anyway?

    I am, tbh, with Kevin Drum on this. The graph just doesn’t make much sense when it comes to tracking the influence of organisations on US political debate. And lets me honest, when it comes to Twitter everyone and their dog has an account but relatively few people make use of it.

  6. 6
    ChrisB says:

    Something of an attack on the public option from the usual suspects this morning. Everyone on Morning Joe is sure that Obama abandoned it last night while CNBC had on a stream of corporate honchos opposed to it.

    All the more reason to demand it.

  7. 7
    RSA says:

    The report is apparently not yet available (or maybe it is, but not for free). Still, when I look at this chart, I see a red flag: Why would twitter and youtube lie exactly on the y-axis with a sentiment of zero? How are they even measuring sentiment on Twitter and YouTube, where in the first case the text fragments are very short (my impression is that some information retrieval approaches have difficulty with very short text, though it’s not my area), and in the second case you’d really want to get sentiment by analysis of spoken words on video? Not to mention Tim F’s aggregation issues.

    And then I see in Keven’s link to marketsentinel the following:

    Twitter ranked highest of all, but as – like YouTube – it is used interchangeably by both camps, we ranked it neutral.

    The marketsentinel post quotes various other media sources, like Fox News and Huffpo and Yglesias, but they treat Twitter and YouTube purely as media, without considering the content. That’s leads to obvious apples-and-oranges comparisons. That is, Washington Post has a measured value of zero for sentiment (however they’re measuring) but some other points are just put there by default.

  8. 8
    RSA says:

    The report is apparently not yet available (or maybe it is, but not for free). Still, when I look at this chart, I see a red flag: Why would twitter and youtube lie exactly on the y-axis with a sentiment of zero? How are they even measuring sentiment on Twitter and YouTube, where in the first case the text fragments are very short (my impression is that some information retrieval approaches have difficulty with very short text, though it’s not my area), and in the second case you’d really want to get sentiment by analysis of spoken words on video? Not to mention Tim F’s aggregation issues. And then I see in Keven’s link to marketsentinel the following:

    Twitter ranked highest of all, but as – like YouTube – it is used interchangeably by both camps, we ranked it neutral.

    The marketsentinel post quotes various other media sources, like Fox News and Huffpo and Yglesias, but they treat Twitter and YouTube purely as media, without considering the content. That’s leads to obvious apples-and-oranges comparisons. That is, Washington Post has a measured value of zero for sentiment (however they’re measuring) but some other points are just put there by default.

  9. 9

    @Napoleon: Made a big difference to me — from four digits to low threes in about an hour on Amazon ranking after 20 minutes or so with Ira.

    More generally — when I talk to publicity people in the book business, including the firm I hired, there is a pretty rock-solid consensus that NPR is the gold standard for serious-book attention.

  10. 10
    Napoleon says:

    @Thomas Levenson:

    As I think about it I actually recall listening to you and Ira on the podcast version of your interview while at the gym. It was interesting.

  11. 11
    Xanthippas says:

    Full disclosure – I did not dive into the internals because, unfortunately, I am not paid to blog and what I am paid to do is kicking my ass right now.

    Generally, this explains my approach to blogging lately as well.

  12. 12
    cybrestrike says:

    If you understand the twisted lenses of how the Village perceives media influence, then it makes total sense.

  13. 13
    harlana pepper says:

    that graph just makes my head hurt

    time for another thread for stupid people

  14. 14
    jibeaux says:

    Well, Stephen Hawking was in a fairly unique position on this, in that he was thrust into the middle of the debate via a horrible and patently false argument that he didn’t have a lot of choice but to respond to. It became a pretty elegant illustration of the absurd, false, insulting, hyperventilating “arguments” of anti-reformers and their living, breathing, genius, counterarguments. If he’d just gone out there without anyone having mentioned his name previously and voiced his support for NHS, it wouldn’t have had nearly the punch.

  15. 15
    linda says:

    reactions to joe wilson continue to pile up; this one is especially on point — via washington monthly:

    Bruce Bartlett noted this morning, “He’s become the new Sarah Palin of the Republican Party, where one’s popularity is in inverse proportion to one’s stupidity — the stupider a Republican is these days the more popular he or she becomes.”

  16. 16
    sal says:

    I know hundreds, at least, of people with a pulse and internet connection but don’t know a single person who uses Twitter. This may be an example of the “Villager” fallacy wherein a local case (most of the people I know use Twitter) is mistaken for a general case (“…Twitter is used by practically everyone who has both a pulse and an internet connection”).

  17. 17
    liberal says:

    Tim F. wrote, Paging Brent Bozell…

    LOL! Good one.

  18. 18
    Shawn in ShowMe says:

    More fundamentally, Kevin commits the multiplication fallacy by comparing the average influence score of a liberal blog (the center of Kevin’s oval) with the influence of any individual news outlet.

    He did it that way to compare apples to apples. If you reduce the liberal blogosphere to one data point so that it looks more impressive on a graph, then an apples to apples comparison would require reducing the MSM to one data point.

    Liberal blogs are doing just fine in the influence department considering their youth and lack of message discipline. DKos and Huffpost will be two clearly defined data points by the end of Obama’s first term, for better or worse.

  19. 19
    Kirk Spencer says:

    There’s a bit of a lie going on here but you have to trace back even further to find it.

    The above chart is produced by market Sentinel. Now what’s amusing is they also post their chart in the Economist. If you go there you might notice something very interesting.

    Mark Rogers of Market Sentinel says that Obama has lost because the highest cite is Twitter and the second highest cite is Fox News. He’s lying. In his report to the economist Fox News is THIRD. Between twitter and Fox is the NHS itself. Obviously favorable, and more positive than Fox is negative.

    There’s also another error, but this is in procedure. What is counted is how often a site is cited. What is NOT considered (as far as I can see) is how often the cite is referenced so as to rebut it. I see, as an example, frequent posts that link to Fox News on Daily Kos and here – posted so as to point out how idiotic or wrong that person was.

    Add Tim’s point – which applies to the whole set of links, not just blogs – and it becomes pretty obvious this isn’t useful at all except as a conversation point.

    And I don’t think I’ll be recommending this company to any clients based upon their performance here.

  20. 20
    gnomedad says:

    You would think the “American Medial Association” would be a centrist group.

  21. 21
    Geoff says:

    “If you want to treat the liberal blogosphere as an aggregate entity, then on aggregate the liberal blogs have a hell of a lot of influence.”

    Yes, but the audience of liberal (and conservative) blogs probably overlaps considerably. As a result, a dozen liberal blogs don’t reach 12 times the audience of a single liberal blog. I don’t think this provides that data you would need to determine overall influence there.

    Also, this chart (http://4.bp.blogspot.com/_Shss.....raffic.jpg) from March indicates about 6M people using Twitter. The demographics also skew heavily to 18-49 year olds, with almost no one younger and few people older. Since that’s somewhere around 2% of the US population, and not even a representative group of it, I’m with sal.

  22. 22
    CalD says:

    Two questions:

    1. What country are we talking about here? Wherever it is, British news sources and the UK Conservative party seem to be amongst the most influential opinion-makers. How did the Telegraph get in there, for example? People in this country don’t even read our own newspapers.

    2. How is it possible for something (e.g., Twitter or YouTube) to be highly influential in a discussion without having any apparent affect (positive or negative) on sentiment? Or perhaps were they’re talking about the relative sentiment of the source itself, not its impact on the discussion… in which case, I guess the question pretty much still stands…?

  23. 23
    Perry Como says:

    Full disclosure – I did not dive into the internals because, unfortunately, I am not paid to blog and what I am paid to do is kicking my ass right now.

    I don’t get paid to blog, but I do get paid to write tools that do data visualization of tracking sentiment on the web. This graph is not that informative.

  24. 24
    Ginger Yellow says:

    The main problem with the graph, if you ask me, is the use of the NHS debate as a proxy for the overall debate. That’s obviously going to overstate the importance of institutions like the NHS and the Tory party and understate the importance of people arguing about different kinds of mainly private sector solutions.

  25. 25
    Bony Baloney says:

    Not to be a dick, but has Hawking ever advanced a falsifiable hypothesis? He’s an astrophysicist. He could say the Galaxy is bordered by neon-yellow Peeps, and who could prove it one way or the other? All we really know is that filming a teacup smashing against the floor and then running the film backward looks really cool, especially if you’re on acid.

    Einstein was an ingrateful butthole whose place in Dante’s Inferno is being stuck in Nazi Germany designing Jew-murdering weapons for all eternity, but say what you will, the atom bomb did in fact work as intended and advertised. Has Hawking demonstrated matter-transference beams while I wasn’t paying attention, or is he essentially famous for being famous, like Tiny Tim?

  26. 26
    distantharbor says:

    Where is the ‘other’ Sarah? Sarah Silverman did and AWESOME viral video during the election ~ bring her back to get the same crowd on board with Health Care Reform.

  27. 27
    Ginger Yellow says:

    “Not to be a dick, but has Hawking ever advanced a falsifiable hypothesis? ”

    Well, his most famous hypothesis, that black holes “evaporate” through particle-antiparticle generation near the event horizon. It has yet to be directly observed, but a recently launched instrument is looking for the predicted gamma ray bursts from evaporating stars, and the LHC may (or may not) generate black holes which may (or may not) decay through Hawking radiation leaving a telltale signature either way.

Comments are closed.