Morning Ed: Media {2017.02.13.M}

Well, this is a real bummer. I can tell how good a news outlet is doing by how frustrated I am by the paywall. WSJ is in the top five.

The subject of Shy Trumpers recently came up. Steven Shepard argues that the phenomenon could mean we’re underestimating Trump’s popularity. I don’t think we are, or at least aren’t by that much. The “safe harbor” for shy Trumpers will be a Clintonesque divergence between favorables and approvals, which we’re not really seeing yet.

Noah Rothman would really like conservative critics of the media to take a step back.

Reuters seems poised to cover Donald Trump as they would an authoritarian dictator.

Every conservative media outlet that stood against Trump, or stood on the sidelines, paid a pretty significant price for doing so. Breitbart, meanwhile, reaped the whirlwind.

Bless Natalie Jackson for taking a good and righteous stand against questions like the PPP’s about the “Bowling Green Massacre.”

Does journalism need to get back to its blue-collar roots? The real question is… can it? As media becomes more vertical and hierarchal, it’s logical and probably inevitable that it would be largely populated by elites with the best connections who went to the best college that the well-to-do families they were born to could afford to prepare them for and send them.


Managing Editor
Home Page Twitter Google+ Pinterest 

Will Truman is a former professional gearhead who is presently a stay-at-home father in the Mountain East. He has moved around frequently, having lived in six places since 2003, ranging from rural outposts to major metropolitan areas. He also writes fiction, when he finds the time. ...more →

Please do be so kind as to share this post.
TwitterFacebookRedditEmailPrintFriendlyMore options

88 thoughts on “Morning Ed: Media {2017.02.13.M}

  1. 1. I’ve noticed that a lot of newspapers are getting tougher about their paywalls and at least asking for your digital data to get through. Election de Trump got a lot of people to sign up for newspaper subscriptions

    2. Not buying it. Trump won because of the freak nature of the electoral college. Saying he is secretly popular is a yearn for the days of the “silent majority.”

    3. Commentary was always anti-Trump and saw Bannon for who he is.

    4. Reteurs is probably on to something here. I am having a very hard time seeing the difference between the incompetence and authoritarian danger posed by Trump. Stephen Miller’s voter fraud stuff on Sunday makes me think we are in for some serious voter suppression.

    5. See above re Bowling Green Massacre. I don’t think it is quite 11th Dimensional Chess but I do think the Trump admin has some plans/really believes in the crazy shit they say. Even if it is not 11th Dimensional Chess, it is a distinction without a difference. They are going to justify doing some authoritarian crack downs. I don’t see why Trump voters should always get the kid glove treatment from the media while the left gets the iron fist.

    Media trust or distrust is long gone. The LGM crowd hates and distrusts the media. My mom who also hates Trump and never voted Republican seems to trust the media. I can see Scott L blast a Maureen Dowd column and then have my mom tell me how the same column really stuck it to Republicans.

    5. The media was blue collar at a time when it was not uncommon for kids to leave school before finishing high school. The cult filmmaker and author Samuel Fueller started as a crime reporter at 14. I also wonder whether you saw more blue collar reporters at the Daily News over The NY Times or Post. There was less “data journalism” in the past as well. Anyway, these guys are imaging going back to when blue collar meant Jewish lefties because that is who became reporters. Also you had Damon Runyon hanging out with gangsters. There was a amorality about the era.

      Quote  Link

    Report

    Reply
     
      • “Trump is secretly popular” the new “Obamacare is secretly popular” from 2010? The justification for the Obamacare being secretly popular was that some people didn’t like that didn’t think the law went far enough. The right metric isn’t seeing what his hardcore supporters think of him, but what the reluctant Trump supporters think and it looks like he’s been losing them. Also, is there any evidence that the non live polls performed significantly better on the general than the live person polls? If so, then we can suspect they capture something about Trump supporters. Because for all the maligning of the polls, they were closer to the mark in 2016 than they were in 2012. It’s just that in 2012, they understated Obama’s support, but still had him winning.

          Quote  Link

        Report

        Reply
         
        • The National polls were off in 2012 and less so in 2016. The state pols were off in 2016 and dead right in 2012. So it’s a mixed bag.

          If there is an issue here and now, I suspect it has more to do with response rates. Whether they answer the phone and participate. A lot if pollster types in 2016 said that was behind a lot of the fluctuation in earlier polls and is largely responsible for things like convention bumps. I thought the effect there was overstated, but there is almost surely something to it (races just aren’t that fluid).

          So it wouldn’t surprise me if now people who like Trump were hanging up at higher rates. It would surprise me if Trump were actually popular, though.

            Quote  Link

          Report

          Reply
           
          • Was it all state polls or just certain state polls that ended up being critically important? Were there state polls in 2012 as off as PA in 2016, but didn’t have an effect and were therefore ignored?

            Also, let’s remember, GWB outperformed Trump, percentagewise, in WI, PA and MI and yet he lost all three states. In fact, Bush beat Trump by over 2 percentage points in Wisconsin.

              Quote  Link

            Report

            Reply
             
            • One of the things I remember from 2012 was Silver saying “Trust the state polls way more than the national.” I thought that was wrong at the time, but I was incorrect. One of the reasons that the aggregators were wrong in 2016 but not 2012 was because that was true in 2012 and false in 2016.

              According to 538 state polling error correlated with the white population without a college degree. The second most incorrectly polled state (after Utah) was Ohio, which was heavily polled in 2012. He also overperformed in North Carolina, which was also heavily polled both times.

              The only heavily polled swing state I recall being really off was Florida (where Obama overperformed). The only two swing states where Clinton overperformed in 2016 were Nevada and New Mexico, neither by a very large margin, and only new Mexico was significant (three points).

              This is based of spreads, which is what we’re looking at. The polling was made more difficult by third party performance, but there were still problems.

                Quote  Link

              Report

              Reply
               
              • Is there a table like that for 2012? I couldn’t find one. For example, the state polls had Obama winning Co by 2, when he won by 5. That won’t stand out as “wrong” when you read afterward that Obama won CO, just like the polls said.

                  Quote  Link

                Report

                Reply
                 
        • I’d say “Trump is secretly popular” is the alternate reality people have constructed so they don’t have to think about just how rotten of a candidate Hillary Clinton was. There must be, these people think, SOME SECRET REASON why she isn’t President now.

            Quote  Link

          Report

          Reply
           
      • I would say that a grain of salt the size of Gibraltar needs to be consumed before reading the polls. Which isn’t to say that they are wrong on any one thing, but simply using them to predict the future is about as accurate as rooting through the entrails of a pig to read the signs and portents. For instance, the weighting of various demographics is suspect, we all know that the wording of various polls is often suspect and that people aren’t always truthy with the pollsters. Further, the prognosticators are often putting their own biases onto any poll reading, leaving them to often be quite Wang in the results.

          Quote  Link

        Report

        Reply
         
  2. Reuters: I’m not sure how you get those dos and don’ts equating to treating the Pres as “an authoritarian dictator.” They seem like things any media organization should be doing.

    Blue Collar Media: Not gonna happen. Those days are gone-at least for the MSM.

      Quote  Link

    Report

    Reply
     
    • Reuters: I’m not sure how you get those dos and don’ts equating to treating the Pres as “an authoritarian dictator.” They seem like things any media organization should be doing.

      I didn’t get that either. The only way I can make sense of it is if you look at the list of countries they mentioned. My whole take was meh, this is a chest puffing piece for Reuters, so they can appear to hip and relevant.

        Quote  Link

      Report

      Reply
       
  3. 5. Co-signing with my brother on returning to journalism’s blue collar roots. Its not possible. People used to read newspapers as a source of entertainment and once other options became available, they quickly abandoned newspapers. Returning journalism to its blue collar roots will also involve a really big change in hiring practices that are going to be hard to stomach for the people who run the newspapers and their readers.

      Quote  Link

    Report

    Reply
     
  4. The blue collar roots of journalism thing is one of the more persistent myths and is somewhat contradicted by the very examples they used (i.e. per Wikipedia, both Mike Royko and Jimmy Breslin had some college before starting their careers – in the America of 1950, this was the very dividing line between ‘working class’ and ‘striving middle class’.

      Quote  Link

    Report

    Reply
     
  5. Will,
    I’m mildly offended that they are characterizing PPP as a Democratic Polling Firm.

    You and this entire site chortled along with PPP when they did that #Panetta Burns thing.

    Don’t be upset that PPP is trolling — that’s what they do.

    They have the requisite “This isn’t about Trump” polling — and if the #Panetta Burns thing is hitting at around the same levels as the “Bowling Green Massacre” thing, well, what have we proved? That a good portion of people regardless of who they are and regardless of the truthfulness of the question will respond based on priors.

    ETA: Okay, so that was roughly 40% of the answering public who said they knew about #PanettaBurns.
    So 10% of Trump voters listened to Conway and thought she was telling the truth. Or something like that (listened to fox news and decided the White House wasn’t lying??)

    This is useful information, from a Public Relations perspective. Props to PPP for actually doing enough research that you can pull the real information out of the poll.

      Quote  Link

    Report

    Reply
     
  6. Fairly recently the LA Times took a different approach, refusing to display content beyond the front page to people with ad blockers running on the Times’ content. They also seem to have lowered their digital subscription price. I will almost certainly subscribe once the auto-play videos and twitchy behavior due to repeated page reformatting as ad content comes in annoys me enough. I haven’t decided whether to take it as a technical challenge or not — my intuition is that, somewhere in the mass of JavaScript they deliver, there’s a single if-statement that could be tweaked and defeat the whole scheme. A browser plug-in to find and tweak it would be amusing…

      Quote  Link

    Report

    Reply
     
  7. The Commentary piece should sneak up behind the Reuters piece and pull down its pants. Why did Reuters happen to realize that official access isn’t important 12 days into the Trump administration? What did Reuters do (or not do) in the prior 8 years that made the last administration so friendly? Isn’t the fact that Reuters is worried about legal restrictions and prosecution a sign that they’re not objective?

      Quote  Link

    Report

    Reply
     
    • If this is a partisan move, why didn’t Reuters do the same thing 16 years ago? Since Reuters and other media orgs are treating this administration differently than they treated an administration from the same party, doesn’t that indicate that this administration is sui generis?

        Quote  Link

      Report

      Reply
       
      • I don’t think Trump is one of a kind. I think the GOP has largely moved far to the right since 2008 and now Trump is the center. But there was plenty of indication in Bush II about what the GOP has always wanted to do and that is totally dismantle the New Deal

          Quote  Link

        Report

        Reply
         
        • The difference was how the Bush administration comported themselves. They were conservative and like all administrations they bent the truth, but avoided outright falsehood. Things like yellowcake or mobile weapons labs were false, but they were based on overstating probabilities of intelligence assessments. Other falsehoods were based on traditional spin. And even though the media was cool to the Bush administration post-2003, the media still engaged official channels and the administration still worked with the media, albeit cautiously.

          They way you engage with spin and exaggeration are different than the way you deal with premeditated lies. The way you deal with coolness and suspicion is different than the way you deal with outspoken hostility. That’s where the difference is. I was talking about the difference between relations due to conduct, not policy.

            Quote  Link

          Report

          Reply
           
        • Saul, again, you describe conservatives in a way that no conservative would. Or am I wrong – are you treating the “far right” as something completely different from the GOP and conservatism? From your comments, I don’t think so.

            Quote  Link

          Report

          Reply
           
          • I think you might be the exception rather than the rule. The problem with words is that people can say whatever they want. I see many odes to the importance of Burkean restraint and moderation but it all seems so pompous to me. Trump’s decisions are very popular with the Republican base. Why can’t I say they moved further to the right?

            The evidence on the ground among GOP elected officials and their base is a solid move to the right since 2008 via primarying and other methods. Trump made white identity a hallmark of his campaign even before Bannon and Miller joined in. He was an early and loud birther. His name has been connected to racism in one way or another since he entered the public eye in 1970 something when the Nixon admin sued Trump’s company (dad and son) for denying apartment applications to people of color. Why not take Trump literally and why not say the GOP are for it? They voted for it. If I said, there is no way HRC would appoint a Goldman person as Secretary of Treasury and then she did, it would be on me.

            I am not under any moral requirement to take odes of Burkeanism from bow tie wearers at face value as they puff on pipes and are impressed with themselves.

              Quote  Link

            Report

            Reply
             
            • OK, fine. Drop Burke. Tell me, on which of the three legs of conservatism have we seen a rightward shift? Social, fiscal, or defense? “Saul doesn’t like that” isn’t the same thing as “conservative”. This past election seems to have revolved around the unclaimed leg of the past 20 years, isolationism / immigration / protectionism.

                Quote  Link

              Report

              Reply
               
        • Trump seems largely indifferent to dismantling the New Deal. He’ll likely sign it if it crosses his desk, but whether you like him or not cutting government or the welfare state is certainly not what animates him. The area where he is most extreme is immigration. That… wasn’t the case with Bush.

            Quote  Link

          Report

          Reply
           
  8. The point behind internal polling is to get as accurate a picture as you possibly can of the actual lay of the land.

    The public polls? Use those to create a narrative and, if you’re lucky, you can change the results of the actual lay of the land.

    Public polling has taken a hit in the last year or so.
    I’m not sure that it’s as useful as it used to be to create narratives.

    I wonder what the internal polls actually say.

      Quote  Link

    Report

    Reply
     
    • How did the public polls take a hit? Here are the final polls for the major media shops:

      CBS/NYT – Clinton +3 (November 4)
      ABC/Washington Post – Clinton +2 (November 4)
      Rasmussen – Tie (November 4)
      Reuters – Clinton +4 (November 5)
      Gravis – Clinton +2 (November 5)
      NBC/Wall Street Journal – Clinton +4 (November 6)
      Monmouth – Clinton +6 (November 7)
      Bloomberg – Clinton +3 (November 7)
      Fox News – Clinton +4 (November 7)
      LA Times – Trump +3 (November 8)
      IBD – Trump +2 (November 8)

      If you just take this group of major polls on the eve of the election, it gives you an average of a 2.1 point Clinton lead. The actual result? 2.1 percent.

      The public polls were the only ones saying Trump would win the primary.

      The public polls for Brexit were close, but slightly off. However, the Jo Cox incident likely gave Remain a polling bump that took a little bit to roll off and therefore skewed the commentary a bit.

      It ain’t public polls that are taking the hit, it’s people’s just so stories.

        Quote  Link

      Report

      Reply
       
      • The Holy Narrative is now that polls are all wrong. That they weren’t on the general election is unimportant. Also lots of people still want to avoid the whole popular vote vs. EC thing.

          Quote  Link

        Report

        Reply
         
        • Here’s the thing: The polls were wrong. They were not spectacularly wrong and (largely) not outside the (substantial) margin of error, but they were wrong. Skewed, even, when it came to a pivotal data point (white people without a college degree). Remember when all of the aggregators gave Clinton a 90% chance of winning and Nate Silver was being called ugly things for saying there was a 35% chance Trump would win? Remember when Wang had to eat an insect? That was because the (state) polls were wrong at what they were measuring. Even though they didn’t get the outcome of Ohio wrong, they were still off by 6-8 points.

          The argument as far as Trump’s current popularity is concerned isn’t “the polls were right” but rather “even if you spot Trump a few points, he’s still very unpopular.”

            Quote  Link

          Report

          Reply
           
          • The national polls were fine and within margin of error. The problem was the probability predictor thingees which a different thing that people like because they make simple headlines and graphics. Probability models are often stupid, super complex and much harder to judge. People are confusing the two things. People misunderstand probability mostly. State polling is often harder since they have fewer polls, usually fewer polling companies and, i believe, use smaller samples.

              Quote  Link

            Report

            Reply
             
            • Until November 8th of last year, everybody knew that state polls were more reliable than national. The national polls did for okay, but they weren’t measuring the important thing. And even there, more of the last ten polls showed Clinton winning by 5 or 6 than winning by less than three.

              The whole point of aggregation was supposed to be to minimize random statistical noise. The assumption being that polls would not consistently be wrong in a particular direction. Except, of course, they were. Because, evidently, there was a persistent problem polling a particular demographic. One that supported Trump.

              As mentioned, 2012 had issues. The takeaway from that isn’t “so we shouldn’t be skeptical of the polls.”

                Quote  Link

              Report

              Reply
               
              • The polling showed the election get much closer over the last couple weeks. cough Comey cough. But that is polling, each one is a just a snapshot of that time and sample. Just from rough memory ( and really, what is the probability of me being in error on that) the margins of victory in the three rustbelt states were within typical polling margin of error.

                I’m more than happy to throw darts at probability models but heck i even thought it was stupid to have Spock or Data absurdly calculating probabilities for stuff. I mean i love when they do that, but it’s still stupid. The prob numbers for the Falcons winning when they were up in the 3 rd quarter were brilliant. Of course it didn’t take a nerd and an abacus to figure the Falcons were really likely to win but again people don’t understand probability very well.

                  Quote  Link

                Report

                Reply
                 
                • The polling showed the election get much closer over the last couple weeks.

                  All of the last ten national polls were in November.

                  the margins of victory in the three rustbelt states were within typical polling margin of error.

                  Pennsylvania was three points off, with only one of the final six polls (all November) showing a Trump win. The one was a Republican polling firm
                  Wisconsin was seven points off.
                  Michigan was four points off, with only one of the final five (all November) polls suggesting a Trump win. The same Republican polling firm.

                  All of the above are looking at 538. According to 538, it’s 4.9, 6.3, and 4.3.

                  Those are not insignificant amounts, and uniformly in the same direction. That’s not a good outcome.

                  (It’s also in-keeping with what we’ve seen internationally. While I maintain the Brexit polling was fine – impressive even, given that it was measuring a fluid election – the 2015 UK election and the Israeli election saw a similar thing.)

                    Quote  Link

                  Report

                  Reply
                   
                  • uniformly in the same direction

                    This is the thing that should have people be freaking out.

                    It’s not that half the polls said it’d be 48-49 (plus 3rd party) and the other half said 49-48 (plus third party) and the 49 people are laughing at the 48 people for getting it wrong.

                      Quote  Link

                    Report

                    Reply
                     
                  • I couldn’t find the demographic turnout numbers for 2016, but couldn’t the same skew simply be due to incorrect likely voter and demographic turnout projections. IIRC, that was basically why the national polls were so wrong in 2012. So in one election, national polls lowballed minority turnout expectations, in 2016, they overshot because they underestimated Obama as a one man minority GOTV operation.

                      Quote  Link

                    Report

                    Reply
                     
                    • There are two aspects to the turnout question. The first is demographic weighting, and the second is contact rates. The latter informs the former because they get an idea of who is likely to show up to vote depending on contact rates. So landline polls tend to favor Republicans because they get fewer Democratic responses because Democrats are less likely to have landlines. The opposite is true for online polling (real online polling, not Drudge online polling). One of the reasons that RCP and HuffPo had different numbers is that the former includes landlines and excludes online, and HuffPo did the reverse.

                      Some polls do demographic weighting, but not terribly much. Pre-election debates over polls tend to revolve around the demographics of the responses. That’s also something that aggregation is supposed to mitigate. This poll has too many whites, this other one has too few, but it evens out.

                      Except when there is, for some reason or another, one side or the other consistently participates in lower numbers. That can be the case when some really bad news hits a campaign, for example. Supporters of that candidate kind of “check out” of politics. And then, during the excitement of political conventions, they’re enthusiastic about talking about politics (and hence, convention bumps).

                      It’s possible that 2012 and 2016 was solely a matter of weighting, but I am skeptical that it is. My spitball guess is that in 2012 Democrats were disproportionately likely to say that they’ll “maybe” vote (screened out of “likely” and then voted. And in 2016, eventual Trump voters were more likely to not participate because either media/pollster antipathy or a lack of enthusiasm about their choice.

                      I don’t know those to be the reasons, but that’s my suspicion.

                        Quote  Link

                      Report

                      Reply
                       
            • Re: probability models… Look at just how late in SBLI the models gave the Falcons a 99.8% chance of victory. It’s not GIGO, or not just that. It’s hard to gauge extreme outliers and how extreme they really are when you’re working on such a small and not directly comparable sample as your precedent.

                Quote  Link

              Report

              Reply
               
              • One difference between sporting events and general elections in the US is that one is a lot more fluid than the other. Points are dictated by series of plays. Elections are gauging something where 90% of it is built in. It’s a lot easier for a football game to change on the turn of a dime than an election.

                I thought the stability of the race was vastly overstated throughout the campaign, but all indications are that Clinton was never really ahead by ten points.

                  Quote  Link

                Report

                Reply
                 
          • 90% is not equal to 100%. If your model is accurate, it should be wrong 10% of the time, which is about the same odds as rolling a 5 on two dice. If your model is never on the wrong side when it says 90%, then your model is broken. The Super Bowl LI gives us a good insight in that sometimes really unlikely things happen without there being some sort of grand media conspiracy. In the election, Hillary should have campaigned more in MI and WI, in the Super Bowl, the Falcons should have run the ball when they had it inside the Pats 25 with 4 minutes to go.

              Quote  Link

            Report

            Reply
             
            • Once you’re talking about 90%/10%, the 10%, as I view it, is that “There’s a 10% chance the polls are not accurately reflecting what’s going to happen.”

              Or, in other words, wrong. Skewed, even.

              To quote myself:

              Whereas when we analyze the likelihood of a kicker getting it through the uprights we’re dealing with a sample set of actual instances, with elections we’re basing it on models that are projections and predictions. At this stage of the game, that’s where most of the uncertainty lies. So if we talk about a 75% chance of Clinton winning, we’re talking much less about another Lehman incident and much more about the models being wrong. That’s the probability we’re assessing.

              That post was wrong in its assumptions, but right in its hedging. Clinton looked good to win unless the baseline (ie polling) was wrong. Which turned out to be the case.

                Quote  Link

              Report

              Reply
               
              • Or 90-10 could imply that there was an odd distribution of the population polled that would lead to an atypical outcome (which is what we got). We got an election where Texas was closer than Iowa. Also, the state poll outperforming national polls is a recent thing. In 2008, the national polls were dead on, while the state polls were slightly off. In 5 of the 6 elections before 1996, the national polls were closer than the state polls.

                  Quote  Link

                Report

                Reply
                 
                • Regarding State vs National, Silver also cites 2004 and 1996 as cases where there was a divergence where the state polls more accurately reflected the state of the race than the state polls. Even in his “pro-national” portion he didn’t cite any that were significantly off in the other direction as 2016 was.

                  I agree that this was an atypical outcome. That, to me, leaves me believing the pollsters (and at least some of the aggregators) are not bad at their jobs. They went with the data they had. The data they had did not accurately reflect anything but the national popular vote*.

                  It’s not just a point here and a point there. A picture of a race that could easily have gone one way or the other. They projected a race that shouldn’t have been close going (narrowly) in the other direction. While there was some bias involved, there was a reason that the aggregators, the media, and folks here read the polls the same way with the same expected outcome.

                  Nearly half of the swing states (7 of 16, I’m excluding Utah) were off by more than 4 points (Maine is 3.9), and all of them were off in the same direction. Looking at RCP in 2012, of the 12 states that came within ten points, only one (Michigan) had that large of a differential (though two more at 3.9).

                  Back to the states, a uniform skew is not the result of statistical noise. It’s not the result of chance. It’s hard to pin down what the issue was in 2012, though in 2016 we seem to have a clearer answer (non-college whites). I think Natalie Jackson makes a good defense here:

                  In retrospect, it’s easy to second-guess that assumption. There’s more error in polls than most people realize. I know that, and have done substantial research on poll error (with more in the works). But I kept looking at the consistency of the polls. They wavered in the exact margins, sure, but always showed Clinton winning in the key states that she needed to win. I saw no reason to question that the polls would be accurate overall.

                  I don’t actually think the pollsters are bad at their job. I think their job is difficult. But while it’s not their fault, it is their problem.

                  The national poll were indeed closer, but “the polls” does not mean “the national polls and only the national polls,” especially not in a race determined by states. And I do want to point out that in the final set of polls, there were more than showed a 5+ margin than one closer than three.

                  But as it pertains to this discussion, involving whether the favorables polling may be skewed, I would liken this more to the earlier polls in the race. Those were pretty volatile and not because people kept changing their mind. It was a matter of response rates. I find it very easy to believe response rates could be awfully lumpy right now. Possibly in a way that inflates Trump’s numbers, but more likely in a way that does the opposite. Most of the lead-up suggested a less tight (national) race than it ultimately was.

                    Quote  Link

                  Report

                  Reply
                   

Leave a Reply

Your email address will not be published. Required fields are marked *