Tuesday, March 25, 2014

"Data" the buzzword vs. data the actual thing



I'm a big Nate Silver fan, but let me join the chorus of people looking at his new "data-driven" blog site and saying "WTF?". As far as I can tell, it's barely data-driven at all!

For example, take this post about how climate change is not increasing the cost of natural disasters. The blogger, Roger Pielke, notes that natural disaster losses have slightly decreased since 1989 as a percentage of world GDP, and concludes that climate change is not causing (and will never cause!) increased losses. The post has been much maligned by professional climate scientists for having crappy and misleading data, but put that aside for now. Let's focus on the idea that this post represents "data-driven" journalism at all. It doesn't. 

The "data" in the post consists of one annual time series with a sample size of 23. That's too small to do any sort of statistical analysis on, but then again, the post doesn't do any statistical analysis. It shows a trendline, and from that trendline it draws broad, sweeping conclusions about the effects of climate change. How is that any more "data-driven" than what any blog does? Every newbie blogger and his dog draws a trendline and extrapolates it - and if the blogger is worth his salt, he'll at least have the common decency to qualify his extrapolation with "if this trend continues", which Pielke does not.

Furthermore, Pielke's analysis is just sloppy. What happens if you strip out earthquakes, tsunamis, volcanoes, etc. from the data? What if you extend the trend back 40 years instead of just 23? How does the recent trend compare with the trend from before climate change started significantly affecting global temperatures? And so on.

And the economic theory behind the conclusion is even sloppier. What about the costs of mitigation - levees, reinforced buildings, relocation of crops, and the like? That won't show up in loss measurements, but it represents a real cost to the economy. And what about variance? Aren't people risk-averse ? As Paul Krugman pointed out the other day, if you try to pretend you're just looking at data without any theory, you've just ignored your hidden theoretical assumptions.

OK, the Pielke post sucks, but that's just one post. Let's look at a few others. 

How about this post by Ben Casselman? This one has barely more data in it than the Pielke post! It shows a single monthly aggregate time series (of voluntary job quits), observes that quits are lower now than in 2008, and concludes that A) the economy may be becoming less dynamic, and B) wage gains may be suppressed going forward. This is less sophisticated than the average econ blog post. Well, at least Casselman used the word "may".

And how does this post extract any information from the data? Where is the data analysis connecting quits to dynamism, or to wages? There is none. Instead, Casselman links to a bunch of Wall Street Journal articles, and one speech about "dynamism" by Dennis Lockhart of the Atlanta Fed. The linked articles are all taken to support Casselman's central thesis. This is pure hedgehoggery, not foxiness

Or take this Casselman post on long-term unemployment. Compare the amount and detail of the data, the sophistication of the analysis, or careful thought about the data's implications to any of the following Matt O'Brien posts on the same topic: Post 1, Post 2, Post 3, Post 4. Casselman and O'Brien reach the same conclusion, but in terms of "data-driven journalism", Casselman's post is nowhere near O'Brien's league. 

Or take this post by Andrew Flowers on whether the labor market is slack or tight. Compare this to the average econ blog post on the topic, in terms of data quantity, data quality, data analysis, and data interpretation. Again, no contest.

Looking at a bunch of other posts, you can see that this is par for the course. And not one of the posts attempts a single quantitative prediction, which is what Silver has famously thrilled the world by doing in the past.

In sum, this so-called "data-driven" website is significantly less data-driven (and less sophisticated) than Business Insider or Bloomberg View or The Atlantic. It consists nearly entirely of hedgehoggy posts supporting simplistic theories with sparse data and zero statistical analysis, making no quantitative predictions whatsoever. It has no relationship whatsoever to the sophisticated analysis of rich data sets for which Nate Silver himself has become famous.

The problem with the new FiveThirtyEight is not one of data vs. theory. It is one of "data" the buzzword vs. data the actual thing. Nate Silver is a hero of mine, but this site is not living up to its billing at all.


Update: Just to be clear, I remain a Nate Silver fan, and I don't mean to get involved in (or start) any feuds. In fact, if I didn't have so much respect for Nate Silver, I'd never have written this post at all. In fact, I think all the criticisms of the new 538 site are based on the idea that Nate Silver is exceptional and can do better than the middle of the pack. It's the curse of high expectations.

97 comments:

  1. Sure looks like Nate thought the Econ & Global Science fields were populated with the rookie bloviators and shills that he encountered in politics.

    But at least in sports, he'll be able to churn out a steady stream of falsifiable predictions and look more prescient than the partisans or bookies who cater to local prejudices, the same sort of soft comparables that he's used to from the likes of Karl Rove's tub-thumping.

    And if people actually care about the odds of who wins, or how valuable a player's 3-pointer attempt is when his team is behind by 5 with 2:00 to play, rather than just enjoying the athleticism and excitement of a contest, then FiveThirtyEight will stay on its feet long enough to re-purpose its other areas of claimed expertise.

    ReplyDelete
    Replies
    1. Anonymous3:28 PM

      Thoughtful post, Walt French. Staying on its feet and repurposing are useful concepts for judging FiveThirtyEight at the moment. Silver earned his reputation in sports and election-campaign analysis, both of which are forms of entertainment. (Who's likely to win? Place your bets!) But, for example, economics and climate change have real human implications. Shallow work in these areas will result in poorly informed citizens, with adverse consequences. I'm pulling for Silver, as I am for Ezra Klein, because a democracy that produces positive outcomes depends entirely on a well-informed electorate.

      Delete
    2. Anonymous5:32 AM

      Did Nate become "uncool " among progressive bloggers after or before he predicted a Democratic debacle in the approaching Senate elections? Inquiring minds want to know. But I'm too lazy too actually do the research. At least these links are interesting, though:

      http://www.slate.com/articles/news_and_politics/politics/2014/03/nate_silver_s_2014_senate_prediction_the_fivethirtyeight_whiz_forecasts.html
      The FiveThirtyEight whiz forecasts a Democratic defeat in 2014, and liberals are furious.

      By David Weigel
      http://m.nydailynews.com/1.1731917
      Democrats are balking at Nate Silver's latest prediction that Republicans have a 60% chance at winning the U.S. Senate in the midterm elections.

      Delete
    3. Sorry, but Dave Weigel's whole "who's ignoring the data NOW, huh" schtick doesn't work (proving, once again, that Weigel's reputation for thoughtfulness is wholly undeserved). Krugman's been voicing his disappointment with the 538 site since its first few days, so no, he hasn't "turned" on Silver because of what Silver's numbers show might happen this fall. Nor have any of Krugman's criticisms been about Silver's continuing electoral analysis.

      Look, this isn't complicated. Silver's claim to fame is his ability to aggregate and interpret poll results in a way that is very accurate in predicting presidential election results. That actually is data-driven analysis. That's not what his new site is doing, and it's not hypocritical, or whatever, for people to point that out.

      Delete
    4. Anonymous10:50 AM

      "Did Nate become "uncool " among progressive bloggers after or before he predicted a Democratic debacle in the approaching Senate elections?"

      I think that most Democrats with any awareness of how midterm elections tend to work understand that that's a real possibility, and understood that long before Silver said anything about it.

      - Guggenheim Swirly

      Delete
    5. I'm not seeing how to reply directly to AC/5:32 but your post pretty well breaks Occam's Razor, plus a whole host of other tenets of good analysis.

      Your, and other spinmeisters' attempts to position this as sauce for the Liberal ganders is more of the same hijacking of the national dialogue that Silver's analysis skewered by unskewing the tub-thumping fundraisers in 2012.

      My complaint here is with the weak tea in what I've seen lately. E.g., the “climate change” piece is a complete distraction from the overall story. It sets up a strawman argument that a trend to more serious storm damage should have been evident based on data back to the '80s if we're indeed going to be increasing average temperatures in the next decades and centuries.

      And even with such a ridiculous premise to knock down, it doesn't: the trend line may have a zero slope, but the confidence range around the slope includes some truly scary scenarios. (Not that careful scientists would attempt to extrapolate so wildly from an unrepresentative base.)

      That's not serious data-analysis and predictions; it simply discredits the writer in the minds of people who have ever done similar analysis in any field and may well confuse the general public about the extent/nature of the climate debate. I have NO idea why 538 dragged such a red herring across the trail of understanding our world.

      Meanwhile, the question of how many midterm seats the R's will capture is what it is, and Politico can engage in the mud-slinging about whose ox is being gored. The concern about the quality of FiveThirtyEight.Com is about how egregiously it's violating its own promise to readers.

      Delete
    6. Joel Rosenbaum1:44 PM

      Nothing screams "credibility" like an anonymous post sourced in hearsay.

      Delete
    7. Good point. And why Noah would still be a fan after this, I don't know. He didn't even discuss Pielke's known climate "skeptic" history.

      Noah must be drinking the Nate Silver marketing Kool-Aid.

      Delete
    8. Anonymous11:57 AM

      Have to say I found Silver's analysis of 2014 Senate races premature and simply his opinion rather than being data driven. He freely admits that handicapping elections this far in advance is more about his "special sauce" and less about poll aggregation. What's a data guy to do when there's no data? Me, I'd just be quiet.

      Delete
  2. David6:30 PM

    You've talked about this some in your assessments of Macro, but maybe some of the reasons that "data-journalism" has thrived in some areas and not in others is because of the varied availability of rich, useful datasets in different domains. Moneyball came to baseball instead of football because it had the most informative data. Likewise, data-journalism is going to do a lot better with election prediction where there are lots of available data than in lifestyle. If I remember, his Oscars predictions have bombed pretty strongly

    ReplyDelete
    Replies
    1. Anonymous11:50 PM

      Ya, sure. I buy that this is going to be the case with things like lifestyles. But all of the posts referenced above are about the economy... Macro economics isn't exactly dealing with a glut of data, as far as I am aware.

      Delete
  3. Anonymous7:35 PM

    Walt French: Did you read that horrible sports piece on steals in basketball?

    ReplyDelete
    Replies
    1. Anonymous2:05 PM

      That piece almost made me throw my laptop through the wall.

      Delete
    2. You mean the one where the author said that steals are undervalued then produced the magic picture to show it was worth 9 times a basket (or was it point? doesn't matter it was basically asserted)?

      That was awful.

      Delete
    3. Anonymous7:00 PM

      Zach, your comment is the sloppy one, sorry. That blog post said something in the lines - if a player with 2 steals/game is replaced by a player with 1 steal/game, the negative effect on the overall success of the team would be similar to replacing a 19 ppg player with 10 ppg player (all other things equal). Said this way it makes a lot of sense.

      My comment on this blog post: I am afraid some people are angry with Nate Silver, not because his new blog is so bad, they are just not happy with the political correctness (from liberal point) of some of the http://fivethirtyeight.com/ conclusions/opinions/predictions.

      Delete
    4. Anonymous8:34 PM

      "... I am afraid some people are angry with Nate Silver, not because his new blog is so bad, they are just not happy with the political correctness (from liberal point) of some of the http://fivethirtyeight.com/ conclusions/opinions/predictions."

      Bollocks.

      Delete
    5. KJMClark9:31 PM

      Well, if you're talking about the election conclusions so far for the Senate, the man's doing a public service. No, we don't like the message, but unlike the head-in-the-sand conservatives, we're *absolutely* listening to the messenger.

      I couldn't care less about sports. If you're talking about economics, I can get much better analysis on every day's Calculated Risk and linked blogs. If you're taking about climate change I can get much better analysis from RealClimate or even ClimateProgress than what I've seen at fivethirtyeight. So far, from my perspective, 1/4 is too low a signal/noise ratio. Still have high hopes for Ezra Klein's new venture - hopefully the Vox thing, Wonkblog, and something from the Times will force Silver to do better.

      Delete
    6. "we don't like the message, but unlike the head-in-the-sand conservatives, we're *absolutely* listening to the messenger"

      KJMClark here captures a measure of epistemological closure:

      When Silver persisted in advancng what SEEMED* to ReCons to amount to a 'prediction' of Obama winning re-election in 2012, ReCons mostly loudly turned off their already low-voltage animalyzers and began acting like pod people from Invasion of the Body Snatchers, - pointing at and moaning to those among the random fluctuations in daily tracking polls that pleased them.

      Now, on the launch of Nate's NewsnArk, ReCons are projecting at DemProgs some sort of equivalent response - despite that DemProgs are NOT disputing 538 on the current projections for 2014, but instead are referencing a few of his hires for particular particularly sloppy work.

      Indeed, as with Noah in this post, it's not even that DemProgs are disputing 538's bottom line on some of this sloppiness: rather, they're saying that the sloppiness itself renders 538's bottom line conclusions into magic box guesswork and supposition.

      [*As Silver pointed out time-and-again in 2012, 538 was NOT 'predicting' a win for Obama: rather, it was publicizing the percentages of likelihood produced by feeding aggregated polling data into the 538 model.]

      Regardless, I suspect many if not most of the wingosphere pushing this meme about DemProgs attacking Silver as the messenger is motivated by a combination of the usual cynicism we associate with electoral politics as a zero-sum game OTOH and situational co-relation between expectations and fund-raising success.

      Delete
    7. No. Frankly I'm neither very interested nor able to evaluate / second-guess any conclusions in baseball. I enjoyed Moneyball the book and the movie but that's about as far as such things go for me.

      Media-based analysis, whether data- or religion-based, needs a way to hook readers. A delicate balance of being provocative for eyeballs, versus wrong for tune-outs.

      Delete
  4. Anonymous8:19 PM

    Krugman & a number of other people have been complaining about the new 538, but none of them actually said what I think the problem is. This post really nails it. This is exactly what I've been thinking: no real data, no real insight.

    ReplyDelete
    Replies
    1. Glad I could give voice to your thoughts...

      Delete
  5. Casselman definitely got called on his (lack of) use of data in comments to the quits/wage increases post. One hopes that if enough of that happens, someone over there will learn something. (If not, I suspect I won't bother to drop by much.)

    ReplyDelete
  6. Ben Casselman here. Appreciate your critique--I mean that entirely non-snarkily. We're a new site, and we're still figuring out many things, among them how to be rigorous without being impenetrable. I don't think any of us would claim we've gotten that balance perfectly on every article, but we're working on it, and substantive criticism will help us get there faster.
    In that spirit, I won't offer a point-by-point rebuttal. But I will note two things: First, we've made a conscious choice to distinguish between blog posts, which are intended to be quick-and-dirty, and features, which are intended to be more in-depth. (More about how we're thinking about this here.) The "quits" post you reference is in the former category--I saw what I thought was an interesting tidbit at the bottom of a Journal story, and wanted to call it out.
    The long-term unemployment post, by contrast, was a feature. The story was based in part on Alan Krueger's new BPEA paper, but also based on my own analysis of the CPS microdata (much of which I'd done before Krueger et al. published their paper). I have great respect for the work Matt O'Brien and others have done on long-term unemployment (among many other issues), but I think we presented new evidence in a way that was accessible to a broad audience. In emphasizing the latter (accessibility), we may have downplayed the former (the original research we did); as I said, that's a balancing act we're still working through.
    I hope you'll keep reading--and keep criticizing--as we evolve.

    ReplyDelete
    Replies
    1. Thanks for dropping by!

      Normal blogging is fine, I just wonder what the difference will be between 538 and any other such site? There are a million chartbloggers out there, and chartblogging is not going to get any better than it is.

      I thought what was awesome about Silver's election and sports stuff was that he had models that were pretty good, and could just blog about their results. It's obviously impractical to build new models for every blog post, but why not do stuff like that?

      Delete
    2. Ben Casselman9:42 PM

      I'll leave the "vision thing" to Nate, who I think has done a pretty good job of laying out what he wants the site to be in his intro post and in various interviews. Predictive modeling will certainly be a piece of what we do, but only a piece, and probably not the biggest one.
      I think our ideal piece, when we get it right, will work at multiple levels--it will not only make sense to the non-wonk but actually draw them in and hold their attention (we've invested heavily in editing talent for that reason), while at the same time adding value to the expert or near-expert reader. Will every story clear that bar? I'm sure not. But that's what we're aiming for.

      Delete
    3. Ben, I went back and re-red the long-term unemployment piece, and I'm still not sure how to react. It seems to me to be a nicely-reported piece, making useful references to other research. But you'd have to tell me what has been *added* to that research. Maybe there's some original work *behind* what's posted, but, without seeing that work, it's hard to assess its validity. Your comment above references your own analysis of CPS microdata (and good for you for doing that), but I can't tell, just from reading the piece, how that analysis informs your piece, how that analysis was performed, specifically what it tells us.

      Unless this is still not the longer analysis where we're supposed to learn about your methods and conclusions. And I really don't think I'm being unfair here.

      Delete
    4. Ben -

      Good luck! Just to offer my two cents, I think you guys are going to have to make it clear what your value add is relative to Scientific American (on the science side) and The Atlantic/Bloomberg View/Business Insider (on the economics side). Those outlets are already substantially more data-driven than what I've seen on 538, and reach very wide popular audiences.

      Silver's value-add over Politico bullshitters and sports scouts was clear.

      Delete
    5. Yeah, I agree with Noah. What really made Silver good in Presidential election blogging was that he had a good model. He had a smart theoretical model of the effects of polls, and how to weight the various ones, unemployment, GDP, etc.

      It wasn't just a naive equally weighted average of all polls, and here's a chart. Also, to really be good, Silver and crew will need top academic experts in statistics, ultra-smart people in a very hard field that have studied it for decades intensely.

      And Silver's competition is people like that, for example statistician Sam Wang of Princeton, who also was a big election forecaster in the media.

      A non-academic can add to what an academic does, but he better make sure with something like statistics especially to make sure to consult the academics, and work with them, or the mistakes can be regular, and often horrible. The same goes, of course, for academics in any field the data's from, from climate science to economics.

      Delete
    6. I mean this seriously for Nate and his crew, order a PhD level first year statistics book (or econometrics, which is statistics for economics) -- and see how hard it would be to learn everything in that book (and all the prerequisite math). But that stuff is really important real world; it's not just mental masturbation at all, and it takes basically a decade at least of intensive, 40-100 hours a week of study by super smart people to understand.

      Please order this book:

      http://www.amazon.com/Econometric-Analysis-7th-William-Greene/dp/0131395386/ref=sr_1_4?s=books&ie=UTF8&qid=1395817731&sr=1-4

      It was the top econometric first year PhD student book when I was a student in the early 2000's, probably still today. Or you can get one an edition or two old for a fraction of the price. Really see what these people know.

      Delete
    7. I mean, isn't Silver's 538 prediction model essentially a black-box? The cool part about it was that it worked, not the explanation of its guts. Still, I think there are ways to explain those complex guts without resorting to calculus-based hardcore statistics (say, something like Peter Kennedy's "A Guide to Econometrics").

      If they are dumbing down the methods themselves so they are explainable to the lay-person rather than doing the necessary analysis (which might be complex) and finding better ways to explain it, it's an insult to our intelligence.

      Delete
    8. Anonymous12:20 AM

      I think it's too bad that predictive modelling won't be central to the new 538. It was what was interesting about the old 538. Not entirely clear that I, or anyone else, wants to read something else at 538. Especially when a lot of it seems sloppy at the get go. Maybe if you sort out the sloppiness issue in time for the election modelling to draw folks back in, it will work out, but the web is not a merciful place, it doesn't provide time to "figure things out". When Silver started the original independent 538, it was great from the get go, and the move to the NYT didn't involve any major glitches either.

      Delete
    9. Anonymous8:00 AM

      As an avid reader of Krugman, 538, and Roger Pielke Jrs' blog -- and as a climate scientist/meteorologist, I have a couple of beefs with your post here (particularly as it relates to Pielke)...

      Just like in economics, climate science has a couple different viewpoints within it when the conversation turns to policy implications compared to when just discussing 'the science'. There actually aren't too many areas of disagreement between a scientist like Pielke and one like a Keith Briffa (the Earth is warming, CO2 is a warming agent and is responsible, warming will continue, etc. etc) ... But when using data to isolate/favor policy, there's the rub. It just about "always" causes controversy when people advocating different policy approaches (or evaluating the feasibility of them toward their desired goal) "critique" each other using data. In economics, Krugman's policy opinions aren't the only ones on the table, but he puts forth data and modeling to continually convince/remind his readers that his are the best. Then, when criticizing "opponents" if-you-will, invariably there comes the charges of 'cherry-picking', 'data-manipulation', and even lying. This can be expected across the board. Krugman is not immune to it as a giver or receiver, and so are us all.

      As it relates to Climate Science though, it is a different bird entirely. Depending on the variable you choose, you may only actually have 23-data points with which to determine a powerful conclusion. Sometimes even just a single set of tree rings (among thousands of samples) deliver the most important clues of how temperature has changed across centuries. Also, with climate science, policy implications are always about the future (even distant future), whereas in economics you have the benefit of near-term validation/correction. I consider your criticisms of Pielke's use of data to be premature, but only because I already understand the larger context from which he makes his points (there are many papers on the subject, even addressing all the valid concerns you bring up).

      Yes, it would have been nice if he already anticipated what others that disagree with him would bring up in regards to 'data' that might change the game--- but even Krugman doesn't do this (or rarely he does, but it's quite awesome when he does). Nate Silver does this often in the arena of politics and polling, which is why he's always been a must-read. The "but why didn't you also talk/think about this" mantra more belies confirmation bias or ideological disagreement if using it as a prejudging, especially when it comes to data pointing toward policymaking. You may find, if you do a little more research (which may be a lot more space than a Krugman or Pielke column can provide in full), you'll find that few climate scientists actually can (or do) disagree with Pielke's data, even as they 'know' they're still right about many foundational aspects of climate change and the continued anticipated rise in global temperature. It's the fact that the data can be used to mitigate demans that certain favored policies are the 'only way' and 'must happen' that has them irked the most. The same fate befalls Krugman as well, and people who agree/disagree with him. To make the jump that "the Pielke post sucks" without any real material of your own (instead conjecture), weakens your own message ironically in the same way, allowing it to be only instantly accepted by those already predisposed to do so. (1/2)

      Delete
    10. Anonymous8:05 AM

      (2/2)

      So in sum, I do agree with you that Roger's piece would have benefitted from him doing all the work of his detractors as well as his own and presenting it all--- Nate's initial advertisement for the positions he wanted to fill included that very thought, but those kinds of expectations have NOT been continually met by someone like Krugman either. And, your conclusive criticism of the data girding Pielke's post (what he includes and what else is out there) is incorrect, in my opinion.

      Delete
    11. He did NONE of that work, besides your 23 data points nonsense neglects the fact that people want at least 30 years of data before they draw a strong conclusion and besides that they want some basic physical basis for the behavior. That is why it is called climate, not weather. You are confusing the two.

      Delete
    12. Nathanael4:38 PM

      Ben, you ask questions: "among them how to be rigorous without being impenetrable"

      The answer to your biggest question is
      "Don't be afraid to admit when you don't have anything useful to say".

      Or, more concisely, if you don't have the data, shut up.

      Pielke has a bad reputation among his colleagues because he's full of shit. http://thinkprogress.org/climate/2014/03/19/3416369/538-climate-article/ In his case, he's probably operating in bad faith. Nate should have the sense to shitcan him. His colleagues have ripped him apart quite comprehensively -- with the data which he ignored.

      The Laskow article -- I believe that Laskow is operating in good faith, but she doesn't know enough to write about the topic she writes about. She needs to learn more biology before she can actually write about the topic.

      (Her bad article: http://fivethirtyeight.com/features/can-evolution-outrace-climate-change/ )

      If you ask someone who actually knows something about the stuff raised in her article, they'll tell you something like this:

      Some species will evolve quickly and adapt to climate change. So what? That doesn't help HUMANS. Climate change still sucks for humans. We evolved in a particular climate. We do NOT evolve quickly (long generation time -- upwards of 18 year generations for humans, versus minutes or seconds for bacteria and some algae).

      Humans depend on a particular set of food chains. Some of the plants and animals in those chains may evolve into new species -- probably ones we can't eat. Others will die off. Either way, we're screwed.

      If 538 is going into the business of climate change denialism, it needs to be shut down. I thought Nate Silver had better standards than this. I can get him better science writers just by pointing him at ScienceBlogs.

      Delete
  7. Good statistical analysis is hard. That's why there's a highly skilled profession dedicated to the subject. Pundits may have many reasons for not performing rigorous data analysis, but at least one of them is that it is hard. It's also something you can't do flying solo. You need a subject matter expert to help structure how you approach an analysis, and how you interpret the results (not to mention how you collect the data). I think lots of people would be interested in a newsmagazine-like item that pursued careful, informed, and sophisticated statistical analyses of newsworthy data. But it takes more than a plan to be data-driven to do this. You need training in the methods, you need experience collaborating with subject matter experts, and you need to know something about the subject matter yourself. "Data-driven" seems these days to mostly be a marker that the author is comfortable with Microsoft Excel.

    ReplyDelete
  8. Anonymous11:12 PM

    Isn't the problem here that journalists can't be jacks of all scientific trades at once, and any attempt to churn out daily, weekly or even monthly pieces that second guess the work of scientists who have invested many hours of research and analysis in some topic is doomed to superficiality?

    It's one thing to call out the analytic blunders of random pundits, but it's presumptuous for people who are not oneself climate scientists - or demographers, or medical scientists, or whatever - to offer off the cuff conclusions about the truth in these fields. Analytical expertise might be a stand-alone skill, but it only generates in scientific knowledge in conjunction with a base of information with which one has developed working experience. A PhD in statistics does not make for an instant economist, climate scientist, etc.

    Perhaps more worthwhile objective for analytically sophisticated generalists in journalism would be to report on the work other people are doing, and use their quantitative facility to explain that work in a manner that is not dumbed-down, but which approaches the reading expectations of popular journalism.

    ReplyDelete
    Replies
    1. I thought the whole point of 538 was to hire people who *are* experts, not journalistic generalists. If that's not happening, what *is* the point?

      Delete
  9. "As Paul Krugman pointed out the other day, if you try to pretend you're just looking at data without any theory, you've just ignored your hidden theoretical assumptions."

    Kruggie didn't take the opportunity here to use his classic "Accidental Theorist" line, a play on the big 80's movie "The Accidental Tourist". Krugman titled a fantastic 1997 Slate article, "The Accidental Theorist" (and later a compilation book). Quoting:

    "And Greider's theorizing is all the more speculative and simplistic because he is an accidental theorist, a theorist despite himself--because he and his unwary readers imagine that his conclusions simply emerge from the facts, unaware that they are driven by implicit assumptions that could not survive the light of day."

    At:http://www.slate.com/articles/business/the_dismal_science/1997/01/the_accidental_theorist.single.html

    Recognizing implied theorizing (ultimately from a set of hopefully benign assumptions about the world linked with unbroken logic) is one of those deeply intelligent, intuitive conclusions that really smart academics will come to understand. Although, you could still be deeply intelligent and not see this. There's lots of things that can keep a deeply intelligent mind busy and elsewhere focused.

    ReplyDelete
  10. Also, I've said this before, it's very different when you do a blog yourself for no money. When there's a lot of money on the line, and prestige, and deadlines, either de facto or imposed, there's going to be a lot of pressure to constantly come up with whiz-bang, cool, or contrarian stuff, so a lot of pressure to really knowingly BS something up, instead of being truth-based and non-misleading, and if that means you haven't come up with something whiz-bang cool in a while, oh well, this is not paid and it's my blog.

    ReplyDelete
    Replies
    1. But wasn't Silver getting paid at the NYT?

      Delete
    2. It was probably a performance based contract and after exceeding expectations he realized he was leaving a lot of $ on the table by staying with the NYT . Leaving was a good move

      Delete
    3. Whether it proves to be a good move remains to be seen. I actually think Nate has overplayed his hand but no doubt if his own site bombs he'll find a new perch easily enough. His political analysis was obviously light years ahead of the bullshitters at most political sites but he wasn't alone. I actually thought Wang at Princeton (who was somewhat critical of Silver's model on the grounds of its complexity) was superior.

      Delete
    4. What Silver did was not usually political analysis but electoral analysis. The confusion of the two on the part of so many reporters is what led to his success.

      Delete
    5. Gutless Wonder8:46 PM

      well is a nice set of data from dateless times, and despite finding the data a "noble failure," described it as "so unobtrusively, flawlessly written that even at its most puzzling it comes as perilously close to poetic beauty as any contemporary set of data I've ever read.....it's really a nice date

      Delete
    6. Probably, he was paid at the NYT. But it wasn't hard to have regular hot stuff that was also good when you were focused on forecasting elections and sports, and you knew that well and had good models, and just needed posts for one small segment of a section of the paper.

      Here, he's going for a huge, very broad site with a huge demand for new content every day, and there's a lot of incentive to come up regularly with whiz-bang and contrarian stuff. If he doesn't drive in the traffic he goes out of business. If he does, he's a rich and prestigious man -- enormous incentive. I was just at the circus, and I was thinking how a business so heavy in fixed costs can make the owners bankrupt or rich overnight by even a modest increase in attendance. The same is generally true for a venture like Silver's.

      It's not just on TV you can jump the shark.

      Delete
  11. The NYT has published plenty of stuff Paul Krugman doesn't like. only when someone strikes out on their own and is a success does Krugman attack. Seems motivated by jealousy. If Roger Pielke posted his story on his own blog without 532 would Krugman care as much? probably not. Krugman is trying to draw attention to his own languishing brand by turning against him

    ReplyDelete
    Replies
    1. Anonymous12:39 PM

      This is one the stupidest things I've read in a while.

      Delete
    2. Are you serious? Krugman whether you agree with him all the time or not is probably the most followed columnist at the NYT. Apart from his tenured perch at Princeton he and his wife are the authors of one of the most widely used econ textbooks in the US university which I think is on it's third edition. At a guess he's probably got the highest price tag of any economist (with the possible exception of Bernanke) singing for his supper at seminars.

      Delete
    3. Anonymous8:50 PM

      When was the last time you saw a comment by someone with "truth", "enlightenment" or "understanding" in their moniker that wasn't Dunning-Kruger-ed to the hilt?

      Delete
    4. Anonymous9:01 AM

      Very sad that wherever you go you read these personal attacks that have nothing to say about the merits of the case. So, you hate Krugman. Does that really contribute to this discussion?

      Krugman's comments from the start mirrored my personal reaction to 538. I, too, was a great admirer of Silver, and I was devastated to see the shallow, pointless posts on the new 538 website that made it look a lot more like Business Insider than anything else. And Krugman's point is quite compelling: when you claim to be data-driven and you have no theory, then you will be implicitly theorizing without being aware of it, thus winding up being sloppy and undisciplined in your theorizing.

      Delete
    5. Bill Ellis9:09 PM

      I think you meant to post that on Breitbart.com.

      Delete
  12. Nate Silver is like a great QB who constantly trashes other NFL players and coaches and then decides one day, "I'm tired of all these crappy NFL teams, I'm starting my own NFL team!" Then he tries to get investors and tries to build up an entire NFL franchise from scratch within a year. Imagine how well that would turn out.

    ReplyDelete
  13. RAstudent10:07 AM

    I kind of feel a touch let down by the new 538 as well, although i would say thats because my expectations were so high. Silver's election and NCAA basketball predictions were top notch. I never paid much attention to his baseball stuff because watching baseball is like watching paint dry. His NFL stuff is ok but my own stuff out performs his and I am a nobody.

    I do find it funny that you are ripping into Nate about data after writing in a post several weeks (maybe months now?) that Richard Florida was a blogging hero of yours. Talk about superficial BS. All you ever get there is correlations with a caveat that they are not causal yet that are interpreted that way 2 paragraphs later. Why? Because he doesn't even understand correlations and his stats people are a bunch of bull shitters.

    What is with all the Silver bashing lately? I think people's expectations were to high and it reeks of jealousy for sure. He does need to fire the climate change skeptic though. That guy sucks.

    Love your blog.

    ReplyDelete
  14. After reading the first few posts of the new FiveThirtyEight, I've decided not to bookmark it and instead wait for the inevitable shakedown period to end and the dead-weight to be jettisoned, before looking into it again and seeing whether there is any actual data to drive the "data-driven journalism" there. Right now I'm waiting for Ezra Klein's Vox to launch to see if the other wunderkind can get off the ground with his "explanatory journalism" concept without falling out of the air just as it clears the runway.

    ReplyDelete
  15. RA: "..it reeks of jealousy for sure.."

    You know, there is a (subtle but real) distinction between disappointment and jealousy. Those who see jealousy in these criticism's of Silver's latest might want to look up the term "projection". I wish Silver well, but at first blush it looks like he has been infected with the WaPo virus.

    ReplyDelete
    Replies
    1. RAstudent12:34 PM

      Good point and he may have been. I am just not sure i understand all the haters. While it hasnt been what I expected from him, there are many people making a lot of money putting out worse stuff.

      Delete
    2. Anonymous9:51 PM

      Yes, many people make a lot of money putting out worse stuff, but that does not stop it from being a disappointment when it comes from someone who had earned your respect from great electoral and basketball work.

      I hope 538 succeeds, and wildly so. I am shocked (and I don't use that word casually or often) at how poor the articles have been so far. I keep reading them almost as if I'm rubber-necking at an accident on the freeway. I repeat, I would be overjoyed if they all turned into a great pieces tomorrow and lasted that way for decades.

      Nate desperately needs to hire adults if he means to do any sort of serious analysis. There are just not that many phenoms like Nate Silver. You generally need experienced people if you are going to fill out a roster like Nate is attempting to do. Put bluntly, he's failed miserably at selecting his roster.

      Delete
  16. Anonymous12:44 PM

    What worked great for Silver before was the fact that the journalistic content was derived from underlying models (as already mentioned in the comments). When you finally have the model framework, you can create a lot of journalistic content derived from this.

    I (perhaps naïvely, but Silver's comment about media pre-538 set the bar high) though this would be an underpinning of the new 538, but right now it looks like a blueprint of other sites which boils down to finding a few data points, create a graph or two, and make typical journalistic inferences, i.e. narratives (thought those guys hated that).

    I'd love to see their staff (being allowed to?) do more thorough research. Derive their journalism from models, research, and analysis that is less shallow than graphing some aggregates or time series and concluding, or misusing linear regressions. If need be, create two-part features. One "mainstream" and one with more details, output and method.

    Perhaps this will be less fruitful economically (but similar to Grantland, this should not be as imperative as usually. This should be a great benefit for 538 and was a factor that increased my expectations before the launch). Output would also be less. But at least it could be something new and exciting, not simply a retread of the type of content that is already out there...

    ReplyDelete
  17. Nate Silver's skill is in prediction. As I've written about (http://www.separatinghyperplanes.com/2014/01/correlation-does-not-imply-causation.html), you don't need causality to be able to make good predictions. But these new 538 bloggers are all about making causal claims, not mere predictions. You can't do causal inference without a convincing theory and identification strategy.

    538 is aiming to do way more than they can possibly do. They want research-quality articles on a multitude of topics, and they want to publish at a rate of several per day. Even the best research institutes with thousands of researchers and billions of dollars can't handle that volume.

    ReplyDelete
    Replies
    1. "538 is aiming to do way more than they can possibly do. They want research-quality articles on a multitude of topics, and they want to publish at a rate of several per day. Even the best research institutes with thousands of researchers and billions of dollars can't handle that volume. "

      You can; you just recruit experts for contributors. That's why people are mad at him for hiring Pielke.

      Delete
  18. What I was hoping for with 538 was some really cool analysis of the current talking points of the day plus some other more general analysis of trends in the economy. So, to attain this there would be great looking graphics which elucidated not obfuscated the analysis. The analysis would be sophisticated - there would be be meta-analysis of aggregate data on top of FRED (for example). In general there would be a model developed of the economy and a library of analyses built up that could be called on to develop an article - around the critical questions of the day. It would be like wonkblog on steroids.

    However, as other people have pointed out we didn't get that - here's hoping they improve, they can hardly get worse.

    ReplyDelete
  19. "climate change is not causing (and will never cause!) increased losses"

    Where does he say that? And who's the "professional climate scientists" who debunked the Pielke post - the two links provide none.

    ReplyDelete
    Replies
    1. Happy to provide you some help on that. Try this for a start. Any you might enjoy this

      Delete
    2. From your link, “While we can already detect trends in data for global hurricane activity considering the whole life of each storm, we estimate that it would take at least another 50 years to detect any long-term trend in U.S. landfalling hurricane statistics, so powerful is the role of chance in these numbers.”

      Thanks for the link and let me recommend it to others as pretty good writing, with no problems to me (just a concerned citizen with some analytical chops, non-expert).

      Delete
  20. Here's Pielke's post (with link to paper) on normalized US hurricane losses: http://rogerpielkejr.blogspot.com/2012/11/normalized-us-hurricane-damage-1900.html

    ReplyDelete
  21. Some what on topic related to cost of remediation. Following Katrina the US spent $14B to fix the flood control systems in the New Orleans area. I don't know how much if anything was spent by Governor Jindall who seems to think government spending is evil. Don't know if Pielke figured this into his analysis or not.

    ReplyDelete
    Replies
    1. " I don't know how much if anything was spent by Governor Jindall who seems to think government spending is evil."

      Oh, I'll be that Jindall spent every federal dollar he could get. And every state dollar that he could get. And probably some of the money over which he had authority might have been 'wasted' on flood control, rather than going straight into his donors' pockets.

      Delete
  22. Anonymous4:11 PM

    Gotta say, this is the most stinging rebuke of Silver of them all. Oh sure Cowen and Krugman shitting on him for producing a mediocre product after so much hype is one thing -- they are old! they are competing platforms! but having the Nate Silver of the economics blogosphere slam down the Nate Silver -- AND THEIR NAMES ARE ALMOST IDENTICAL JUST LIKE THEIR VAGUE APPERANCES -- must really sting old Nate. He even had one of his underlings pop in here and try to defend the incompetence. O well, at least he has all of that dirty money from ESPN to keep him warm.

    ReplyDelete
  23. I like the idea and concept of Mr. Silver and Co.'s new bloggy, web-mag. I consider it to be a more sophisticated pop-culture men's site, without selfies of attractive females, alcohol aggrandizement, and annoying attempts at making videos go viral.
    538 is not a refereed academic publication or a Woolridge econometrics text. It's simply more noise (technically speaking). However, amidst the multitudes in the blog-o-sphere who feel they perpetually have to prove how they are, indeed, 'The Most Intelligent Man in the World,' 538's offers predominately male readers subject-matters that are a nice cross-section of topics from wonkie interest to straight inconsequential whimsical fun.

    538 is science, pure sociology.

    ReplyDelete
  24. Anonymous5:00 PM

    Why does anyone think Nate Silver is good at statistics? He is the worst sort of regression monkey, one who actually doesn't know anything about the underlying assumptions he is making. But at least before he was just forecasting. Now he's trying to make causal arguments, which are way above his paygrade. The guy is a charlatan and should be publicly castigated until he admits it.

    ReplyDelete
    Replies
    1. Anonymous9:56 PM

      Is this false flag? If not, seek help. Immediately.

      Delete
    2. I don't know who actually made it up, but "regression monkey" made me chuckle.

      Delete
  25. Anonymous5:15 PM

    Noah,

    How well does Ben Casselman's piece "Missing: Up To 4 Million Workers" meet the standards you hope FiveThirtyEight achieves?

    ReplyDelete
  26. I thought Silver's reaction the the DSCCs memo/campaign pitch was particularly bad. The memo was a response to Silver's analysis that the Republican's are currently have a better chance to retake the Senate than the Democrats do to keep it. The memo stated that Silver does good work, his predictions aren't always correct, but they do show the challenges ahead:

    "We don't minimize the challenges ahead. Rather, we view the latest projection as a reminder that we have a challenging map and important work still to do in order to preserve our majority."

    When they later used Silver's projection in a fundraising e-mail, Silver wrote that they were being "hypocritical" and said they "can't have it both ways":

    http://fivethirtyeight.com/datalab/having-it-both-ways/

    Apparently, if you say an analysis shows that you have a tough road ahead but isn't proof that you'll lose, you're a hypocrite if you later say that it...shows you have a tough road ahead but isn't proof that you'll lose.

    ReplyDelete
  27. I've thought the value in Nate Silver's election modeling wasn't the model itself, but the fact that he did the leg work. Going out,collating a bunch of different polls and organizing them into a single coherent data set.(with estimates of quality and bias)

    There are a few polls of polls at places like real clear politics, but they're pretty naive averages of a small set.

    ReplyDelete
    Replies
    1. As Eli wrote on Rabett Run Data without a good model is numerical drivel. Statistical analysis without a theoretical basis is simply too unrestrained and can be bent to any will. A major disaster of the last years have been the rise of freakonomics and "scientific forecasting" driven by "Other Hand for Hire Experts" and, of course, Pielke is the first one out of the Roladex on that

      When data and theory disagree, it can as well be the data as the theory. The disagreement is a sign that both need work but if the theory is working for a whole lot of other stuff including things like conservation of energy as well as other data sets, start working on the data first.

      Delete
  28. Anonymous9:29 PM

    Noah - I completely agree with you. This morning, after reading Walt Hickey's 538 piece on Payscale's ROI on college rankings, I felt compelled to post there not just about how fundamentally flawed the "analysis" was but also how pieces of this low quality, branded with Nate's valuable reputation, actually undercut Nate's stated objectives around data journalism.

    Top quality mathematical models can be applied to a wide variety of problems to provide valuable insights. Nate made his reputation by building terrific models and using those models to both debunk shoddy work and to provide new insight gleaned by analyzing the data well. I have been a huge fan of Nate's since 538's early days, but thus far, I have been extraordinarily disappointed with the work on the new 538 site. I hope that he can right the ship, but I am not particularly optimistic. Nate's strength comes from his deep understanding of modeling. I do not believe that his team has that knowledge.

    My entire professional career has been making mathematical models and leading teams to do so, first in transportation, then in finance. After 30 years of hiring people to build mathematical models, I have found time and again that it is not difficult to teach a modeling expert to build a model for a particular domain, but it is incredibly hard to teach a domain expert to build a good model.

    The new 538 seems to have shifted from modeling and its resulting analysis (which built Nate's reputation) to journalism with some data (which may undermine Nate's reputation).

    ReplyDelete
  29. Comment in bad taste (Rebut this! (with data please)):
    What is Silver's business plan?
    Where does he get his money?]

    I'm not saying "sellout" but his reputation is the type of thing that could attract the Big$$ with an agenda.

    ReplyDelete
  30. I did want to mention 2 things about Nate Silver. One, he DID NOT accurately predict who won all 50 states in the 2012 election, as he claims. He accurately predicted a few days out the outcome in 49 states. 46 of those were a snap if you followed the polls closely. Nate Silver INACCURATELY said that Romney was going to win Florida. Then, on the morning of the election Silver saw what was happening from his sources in Florida ON ELECTION DAY MORNING and CHANGED his Florida prediction to Obama. That is NOT a prediction, it was already election day, he cannot be counted as getting that right when he sees election day information and switches. ***** My 2nd point, as to the false claim that global warming/ climate change is not increasing the cost of natural disasters. I have heard Silver in interviews indicate in subtle ways that he really does not believe global warming is much of anything. He seems to share the Republican attitude that man cannot affect the climate, that its all just natural variance. I cannot give you a quote, I did not DVR these interviews. Maybe someone has recorded them.

    ReplyDelete
    Replies
    1. no no no
      everyone knew, or should have known, that of the 50 states, only about 10 were in play
      was there any doubt whatsoever of how Texas, Georgia and Alabama, or NY MA or CA would vote ?
      of course not
      So to accurately understand the probabilitys and statistics, you would have to go back and look at what states were in play
      one time honored method - perhpas skewed this cycle by romney - is where the candidates were going and where they were buying ad time; I've lived in MA for 20 years, and we get very, very, very little in the way of advertising - I mean, who is gonna waste money on MA ? we are gonna go Dem
      That was one of the odd things about 2008; Obama carried normally safe GOP states (expanding the map as they say

      Delete
    2. Silver didn't "change" his Florida prediction. He called it a toss up. The number generated by his model changed from ever so slightly favoring Romney to ever so slightly favoring Obama, but that was meaningless. The bottom line was that 49 states were somewhere between set in stone and reasonably certain, but one was anyone's guess

      Delete
  31. In reference to what I said in my above post, that I remember Nate Silver expressing Republicanesque or at least Republican-lite global warming denial, but in a subtle fashion, I was correct. I could not find his quotes on TV, but I did find this link. Nate Silver plays the game of elevating a scientist (who is one of the 3% who do not recognize the facts about global climate change) to an undeserved status of one with legitimate issues about climate change. You will not catch Silver making outrageously false statements ridiculing the entire global warming concept, he has to keep up his image that he is merely a completely non-partisan judge of factual data. Which he is not.
    http://delong.typepad.com/sdj/2012/11/michael-mann-vs-nate-silver-on-climate.html

    ReplyDelete
  32. I call it data anecdotalism.

    ReplyDelete
  33. Anonymous11:29 AM

    Want some more "data driven" evidence on the coming costs of climate change? The insurance industry, which is raising premiums and limiting coverage for situations where they think climate change will have an impact. Here's a recent Times story, maybe Nate's blogger could look around a little more before posting such a trivial piece:

    http://www.nytimes.com/2013/05/15/business/insurers-stray-from-the-conservative-line-on-climate-change.html?pagewanted=all&_r=0

    ReplyDelete
  34. Quote
    take this post about how climate change is not increasing the cost of natural disasters. The blogger, Roger Pielke,
    Unquote

    Dr Pielke says that disasters are not increasing, and gives a link to the IPCC as a source.
    IF you follow the link, you wind up not at the data supporting the claim, but at the IPCC homepage - it is up to the reader to try and find on the huge IPCC cite the relevant data
    Lazy and sloppy; every discipline has technical basics; for wonkblogging, one of the basics is precise links

    Almost as bad as economists who give "www.bls.gov" as a source

    ReplyDelete
  35. Today Silver publishes a list of favorable, unfavorable, and neutral Krugman posts about fivethirtyeight and/or Nate Silver, and since the motivation would come from how negative Krugman has become some Silver left NYT, I read all of the negative ones that themselves linked to Silver's posts Krugman was taking issue with, and one I'd read on fivethirtyeight I read and was puzzled by.

    I found Silver's writing on economics jaw-droppingly uninformed, as though he'd taken the same Econ 101 class I took thirty years ago, didn't read the whole thing, and hasn't tried to follow the discussions since.

    I'm not an economist either, but the numbers he looks at without context, and the models he uses to explain them were really freshman explanations of tax policy that the Wall Street Journal editorial page likes to recite. (He mentions the Laffer curve as not really being applicable to the level of taxation in one.)

    It just had me stunned to see what a statistician who is not an economist doesn't think it's relevant, what with numbers hee compares, what he ignores or more likely doesn't know to consider that are also numbers. Just painful. The criticisms are really accurate and deeply unfortunate.

    The articles I've read on the new site don't invite me back as a reader of journalism and current events, or interesting ways to think. They were vacuous and uninformative, universally less than half-baked.

    I'll watch his polling analysis closely.

    ReplyDelete
  36. Largely agree. I hope the early issues will be ironed out. No, this isn't about anger because of his mid-term prediction. That, actually, was one of the better articles they've done so far. What made it stand out for me was the article reporting on the BLS study showing the difference between earning for college graduates and high school graduates. Great study, and the reporter was simply relegated to saying "Hey, great study"! I'm willing to be patient and give the site time to find it's niche, though. There's been a sprinkling of good material, hopefully it will increase (probably) as time passes.

    ReplyDelete
  37. Anonymous2:29 PM

    Somewhere, Jill Abrammson is smiling.

    ReplyDelete
  38. Anonymous3:50 PM

    <> Since a sample size of 23 is too small to be meaningful, I wish Noahpinion and Readers of Noahpinion could at least draw their inference from an analysis of at least 23 articles on Nate Silver site before passing a judgment ...

    ReplyDelete
    Replies
    1. Haha good point, good point...

      Delete
    2. But the point is, a sample size of 23 is only too small for certain kinds of statistical analysis and tests and data, with certain kinds, or lack of, apriori information.

      A crucial and little understood point is that it really depends on the situation and on your apriori or supporting information.

      If two people are walking through the woods and one eats a berry and immediately keels over, should the other say, hey, it's only a sample of one, and since I haven't eaten since lunch,...

      Obviously not; given the aprori information that most people have on basic biology, and the similarity of humans in basic bodily behavior, it's extremely likely from just that sample of one that those berries are extremely dangerous.

      And in fact, this isn't just my conjecture. I've studied some really advanced PhD statistics, and many times they've talked about medical samples less than 10, but because the results of those 5 or 6 were so extreme, it was very conclusive evidence that something big was going on.

      Suppose, for example, you sample a roulette wheel, and 23 times in a row it lands on 15 black. Would you say, it's only a sample of 23, that doesn't tell you anything? If the sample was even just three, all 15 black, that would be very strong evidence that the wheel was highly biased.

      Delete
    3. This is actually a very good example of what Krugman was saying about the importance of modeling, and implied, or "Accidental", modeling. We have really important basic knowledge about how the human body works and how similar all human bodies are in basic ways -- that's a model. But it comes from a ton of supporting data and evidence that we have going in, or apriori, to our berry sampling -- Ignore that model, and interpret the sample naively, at your own peril.

      Delete
  39. I'm a bit uncomfortable with a blanket statement such as 'As Paul Krugman pointed out the other day, if you try to pretend you're just looking at data without any theory, you've just ignored your hidden theoretical assumptions.'

    Science, esp. Physics, has not always been about hypotheses and hypothesis testing. If you go back far enough, you find a contrary sentiment in scientists such as - oh - Newton and Maxwell for example.

    Newton eschewed hypotheses altogether -- *hypotheses non fingo* ('I feign no hypotheses'). That is, he at least pretended to approach his data empirically and with no theory at all in mind, that wasn't suggested by it. He did allow what we would call today semi-empirical models (motivated curve fitting).

    Maxwell is maybe the last scientist of stature to oppose, in print 'The Method of Hypothesis' as a novelty and danger for science. Read the last few chapters of his 'Matter and Motion' which is very readable even today.

    Oh, and the one time he did use the Method of Hypothesis, albeit with many caveats and strictures, he ended up with a classical molecular theory that was wrong, and proven so by early Quantum Mechanics.

    So, far from 'there's always a theory' -- the two most significant physicists of the last 400 years, who formulated about half of what we know and maybe more, felt differently from you and Krugman about how to approach data.

    I'm aware I'm overstating my case a bit here -- but the contrary view should be looked at, and the fact that physics, as a paradigm science (whatever you think it's relation to economics might be...) should at least be known to have a controversial methodology dispute, when it comes to hypothesis testing with data -- one that was well known a century ago, and had a name, the Method of Hypothesis, to pigeon hole the discussion. The discussion is not closed, and needs to be revisited in several contexts.

    ReplyDelete
  40. I think part of the problem with this whole matter is that science isn't really meant to be a debate. It's not supposed to be a smug snarkfest, taking place across blogs on the internet either. Not to sound like a jerk, but most of the people in the discussion sound like they don't have a very rigorous education in science, or that they left it behind a long time ago so they could "celebrity minds" who write articles for major online news sites.

    And when you treat science like something it's not, as the comment above mine points out, you start say stuff that doesn't make sense, or is stridently ignorant, or is just plain wrong. The people who seem to be writing this stuff, or reveling in it, don't seem to have an appreciation for this fact, don't seem to realize that they're dealing with more delicate and thoughtful material than they treat it. Experiments can be wrong, statisticians can outsmart themselves (frequently), the nature of the things being studied can be completely misunderstood, pop science lovers and scientific neophytes don't seem to bring much understanding of that to the discussion. They seem to delight in a few online articles as being their secret weapon to being smarter and more "data driven" than everybody else. But empirical reality isn't a debate between Fox News and MSNBC. And it just feels like everyone would be a little smarter if these enthusiasts realized you don't have scientific literacy, or an appreciation for how the experimental process works, by just reading a lot of scientific news stories over the internet.

    ReplyDelete
  41. Checkout here worlds famous collection of coolest website imadude.com. This cool website is only for you. Imadude.com provides latest funny picture, videos and much

    more.For more info feel free to visit on our website http://www.imadude.com and enjoy.

    ReplyDelete
  42. If you want to see some additional godawful "analysis" from 538, look through Harry Enten's stories. More often than not, the data he cites is just plain wrong. He either doesn't know how to cross-reference, or just has no problem flatly lying to his readers.

    One such example was when he stated that the GOP, as of late April, was in the same statistical polling position as they were in 2010.

    The problem? The poll he cited, Pew Research, didn't even perform polling at this point in 2010. And an aggregate of generic ballot polls showed a 1.9 point drop for the GOP from their 2010 figures.

    Regardless of what your political affiliations are, I think that's a poor enough performance (one of many) to warrant asking what in God's name Nate was thinking by hiring this team of boobs?

    ReplyDelete