Obama made Glenn Beck "a better man"

Discussion in 'Politics & Current Events' started by HeHasTheJazzHands, Nov 9, 2016.

  1. HeHasTheJazzHands

    HeHasTheJazzHands greg rulz ok

    Messages:
    20,702
    Likes Received:
    1,966
    Joined:
    Jun 18, 2011
    Location:
    Louisiana
    Surprisingly, a lot of online polls swung more towards Trump. But the alt-right tends to hijack a lot of polls.
     
  2. Randy

    Randy Ooh, Degrasse Tyson-son Super Moderator

    Messages:
    19,787
    Likes Received:
    2,406
    Joined:
    Apr 23, 2006
    Location:
    The Electric City, NY
    Election is way over and we're still doing the partisan, knowingly deceptive word smithing? Come on, man. Nearly every poll had Hillary ahead, some (more than any poll favoring Trump) by large numbers. "FiveThirtyEight predicted how it model could be wrong if things went the other way" is hilarious. Id love to be able to bet on a team and still win the bet as long as I recognize the other team might win, too. Wow.

    Since we partially "went there", the whole "fake news" thing is hilarious "sorry, not sorry" excuse making for the echo chamber in journalism. Snopes exists because fake stories have persisted for a long time, notably going back all the way to George W. Bush. I read a tone deaf article in NYT this morning about how deceptive fake news is because of logos and domain names. I'm pretty sure I remember conservatives buying those Re Re Re Re: emails as gospel and not bothering to fact check them over a decade ago. People have just shifted away from email and onto Facebook, so that's the only real change in how fake news gets spread. And that arguably less brainwashing of a format because friends or friends of friends or family can be of different political leanings, see the post and stick barbs in it; the old chainmail format made that essentially impossible.

    Id also like to note that Obama is a black man, with a "funny sounding" name (which includes a middle name reminisct of a Muslim leader we deposed) had every possible fake story thrown at him from Genesis and he still won a primary and two general elections. The argument the Dems lost the race over face news or racism is just "feel good" self assurance bull.... that resolves nothing and will forecast another loss if we choose to be guided by it.
     
  3. Randy

    Randy Ooh, Degrasse Tyson-son Super Moderator

    Messages:
    19,787
    Likes Received:
    2,406
    Joined:
    Apr 23, 2006
    Location:
    The Electric City, NY
    Regarding Beck, the guys a shill. He hopped on the Republican and Christian Right bandwagon because it was a silent majority group who were perpetrating this storyline were they were being persecuted for their beliefs ("We can't say Merry Christmas anymore!"). You essentially had a large group of people playing the role of the victim, and someone like him sees that as an opportunity to cash in. He pushes his "I feel your pain and I'll help you through this" malarkey.

    Same deal with 2016 Democrats. We still outnumber Republicans nationally, and we're coming of a pretty clean 8 years with our guy in office. We're no underdog, we just lost in a raw deal. Easy for a guy like him to swoop in and use the same routine, while we're busy licking our wounds. The guy's an opportunistic social climber. "Ive got the cure for whatever ailes you" snake oil salesman.

    Also, if we're going to niggle over where the hipster trend is in booze, I think its drifted over to small batch distilleries. Might be too bleeding edge though. /hipsterAF
     
  4. celticelk

    celticelk Enflamed with prayer

    Messages:
    4,382
    Likes Received:
    330
    Joined:
    Apr 13, 2011
    Location:
    Ann Arbor, MI
    Funny thing about statistics: low-probability events actually happen. They just happen with less frequency than higher-probability events, by definition. The difficulty comes when you have an event that, by its nature, can only occur once - like, for example, a presidential election (every election is a unique event by virtue of its historical circumstances): does the occurrence of a low-probability outcome actually signify that the model was wrong, or just that Trump got lucky? And FiveThirtyEight was giving Trump something like a 1-in-3 chance, which is not that remote - less than winning a coin toss, but better than rolling a 1 on a standard die. And the election was *actually* close: flip 55,000 Trump votes in MI, WI, and PA to Clinton, and she wins those three states and the election. (For perspective, Trump earned more than a million votes in MI alone.)
     
  5. thraxil

    thraxil cylon

    Messages:
    854
    Likes Received:
    223
    Joined:
    Dec 28, 2008
    Location:
    London
    Yeah, I'm kind of tired of seeing fivethirtyeight slagged like that. I think that perhaps they did a poor job of conveying how statistical models like theirs work to non stats-geeks. That, combined with their having been astonishingly accurate in a few previous election cycles led too many people to see "70% Clinton, 30% Trump" from them and interpret it as a foregone conclusion that Clinton would win. Even Nate Silver and the other editors got caught up in that thinking in their interviews and more editorial articles. But their numbers and models were pretty solid overall.
     
  6. Randy

    Randy Ooh, Degrasse Tyson-son Super Moderator

    Messages:
    19,787
    Likes Received:
    2,406
    Joined:
    Apr 23, 2006
    Location:
    The Electric City, NY
    Ah, that makes a lot more sense when you flesh it out that way. That I can agree with.

    My one caveat would be that, while FiveThirtyEight MAY have been explicit in stating the stats didn't guarantee the outcomes, the entire rest of the media were not as explicit. Not even close. And that's not conjecture based on poor reading of stats, thats based on 24/7, opinion and analytical news coverage with talk heads continually stating the results were all but locked down completely. I don't think it would be unfair to say there WAS a narrative saying consensus believed Hillary would win, period. An average person can only hear that repeated so many times before they believe it.
     
  7. celticelk

    celticelk Enflamed with prayer

    Messages:
    4,382
    Likes Received:
    330
    Joined:
    Apr 13, 2011
    Location:
    Ann Arbor, MI
    No argument there.
     
  8. Demiurge

    Demiurge Intrepid Jackass

    Messages:
    3,832
    Likes Received:
    281
    Joined:
    Dec 25, 2005
    Location:
    Worcester, MA
    Maybe it was just a matter of time before projections based on polling would fail, and, given Nate Silver's past successes, 538 would be the biggest giant to fall.

    Statistics class is long behind me so please correct me if the concept isn't as significant, but I don't really see participation bias brought up that much. Do pollsters forget that not everyone likes participating in polls & surveys and that those who do might be specifically motivated? There's hand-wringing that the sample sizes weren't large or diverse enough, but the sample population itself might be skewed.
     
  9. Randy

    Randy Ooh, Degrasse Tyson-son Super Moderator

    Messages:
    19,787
    Likes Received:
    2,406
    Joined:
    Apr 23, 2006
    Location:
    The Electric City, NY
    Ehhh, theres some self defeating logic going on in there. You say you're not happy with the bad rap FiveThirtyEight is getting, because their numbers were "solid" but then you mention their editor was going on TV projecting results. In a perfect world i suppose we could separate raw numbers from the opinion of those collecting them, but this isnt a perfect world and he belies his credibility or his sites credibility by jumping into crystal ball reading. You can't blame anybody for criticizing that.

    Also, theres this "FiveThirtyEight had it right while everyone had it wrong" simply because they were supposedly explicit in the fact "over 51% probability =/= guarantee". Thats great and all but FiveThirtyEight doesn't run the polls, some independent and mostly other media outlets do. To say FiveThirtyEight did anything exceptional by just reporting everyone else's numbers is fetishizing FiveThirtyEight. If we're opening up anybody who statically had Trump above 0% chance of winning as being right, then technically the whole media had it right. I'm not ready to give anybody that much credit.
     
  10. AngstRiddenDreams

    AngstRiddenDreams Filthy Casual

    Messages:
    3,467
    Likes Received:
    311
    Joined:
    Mar 7, 2011
    Location:
    Seattle
    I think the problem lies in the average American having little understanding of how numbers, let alone statistics work. Situations like this remind me of the time A&W (fairly sure) released a 1/3 pound burger that consumers rated as tasting better than a McDonald's quarter pounder. But when asked which they would purchase, the majority of people said they'd stay with the quarter pounder because 4 is bigger than 3. I'd say fractions are significantly easier to understand than statistics, but maybe I'm skewed as a math major.

    It would awesome if there could briefings on television news station where people get explained, "80% odds aren't a guarantee. The polls used to generate these statistics are highly unreliable as bots have been found to cast poll votes, among other issues that make them non-representative. People are highly unpredictable. Voter turnout is highly unpredictable."
     
  11. Smoked Porter

    Smoked Porter SS.org Regular

    Messages:
    385
    Likes Received:
    57
    Joined:
    Nov 28, 2014
    Location:
    Texas
    Yep. Everybody loves a good redemption story, but I'd have to see a decade (at a minimum) of consistent logical thinking and points from the dude to maybe believe he's a changed, sane man.
     
  12. thraxil

    thraxil cylon

    Messages:
    854
    Likes Received:
    223
    Joined:
    Dec 28, 2008
    Location:
    London
    The distinction I was trying to make was between their model and its results, and the editorializing that the staff engaged in on the side. The former were solid, the latter got swept up in the same wishful thinking that the rest of the media did.

    fivethirtyeight doesn't just "report everyone else's numbers". They aggregate pretty much every poll out there and analyze them based on past performance, sample size and demographic targeting and then use them to build their own model that incorporates various things like certain demographics across different regions that tend to vote as a block. Then they run thousands of simulations based on those models and show the results.

    The editorializing usually took the form of "for Trump to win, he needs to flip all of these states that are leaning towards Clinton; he might get one or two, but it's unlikely that he'll get all of them". But that logic was predicated on the results of those states being statistically independent, while 30% of their own simulation runs had them all flipping together.
     
  13. Drew

    Drew Forum MVP

    Messages:
    26,339
    Likes Received:
    1,875
    Joined:
    Aug 17, 2004
    Location:
    Somerville, MA
    Fivethirtyeight's projected result was wrong this election, no doubt. But, their probabilistic forecast wasn't necessarily off. Basically, they gave Trump a one-in-three chance of winning at a time where many of their peers were more like 1-in-10 to 1-in-20, citing the uncertainty in the polls. They forecast a Clinton victory, but they also forcast a way less certain outcome than their peers.

    And that's the thing with probability - speaking as a financial professional who spends a lot of time playing with numbers and not as a frequent FiveThirtyEight reader, when working with small sample sizes (which, with one election result, is what we have) it's challenging to evaluate the prior forecasts. A 1-in-3 probability may very well have been exactly right coming into the election, but with only one result, that looks like a "miss" if Trump wins, and "way too conservative" if Clinton wins. If you could rerun the election as a hypothetical using different voter universes that when randomly sampled would give you the same polling data and expected errors as what had been previously obtained, you'd have a bigger sample size to work with, and you very well might see two out of three of those universes consistently elect Clinton, with one out of every three breaking towards Trump. The problem is, we only have one universe of voters, one result, and no good way to go back and evaluate the forecasts made beforehand. Technicallly speaking, even the guys projecting a Clinton win with 99% probability can't be proven to have been wrong just because Trump won, because a 1% probability event can still be expected to happen one time out of one hundred.

    This is all very tangential, but I think it's important. I'm not even trying to defend FiveThirtyEight here, as they made plenty of errors this cycle (and their mea culpa on getting the Trump nomination is worth re-reading here). I'm just defending statistics, and saying that if an event is forecast to have a one-in-three probability of happening, that doesn't prove the forecast was wrong if it actually happens.
     

Share This Page