Jon Letendre Posted July 25, 2016 Share Posted July 25, 2016 Unless I have screwed up the meanings of all the terms and forgotten all the stats I learned. (I swear I took the course and got an A, but alas, many years ago.) Link to comment Share on other sites More sharing options...
Michael Stuart Kelly Posted July 25, 2016 Share Posted July 25, 2016 Jon, Not to worry. William is pretty adamant that Nate Silver never alters his methodology, so if you did make a mistake, Silver's constancy will serve you well as a model and a corrective. Just give it time like Cuzin Nate does and everything works out. Michael Link to comment Share on other sites More sharing options...
Jon Letendre Posted July 25, 2016 Share Posted July 25, 2016 Oh, good, thanks. ? I did mess it up, I think, at least with the assumption that "one in X chance means run it X times for confidence of at least one hit." Not sure that's necessarily true. Maybe Bob can help me. Link to comment Share on other sites More sharing options...
william.scherk Posted July 31, 2016 Share Posted July 31, 2016 (edited) Why I love OL. Quote One wonderful thing about good faith forum discussion is that mistakes can be corrected, misapprehensions brought in line with reality; discussants can fix error, fill gaps in knowledge -- and each move on smarter or better-informed. How we achieve this Win-Win situation ... On 7/25/2016 at 11:52 AM, Michael Stuart Kelly said: On 7/25/2016 at 10:25 AM, william.scherk said: There are three kinds of 'forecasts' that 538 releases. Today's is an example of a "Now-cast," and the other two are examples of "polls-only" and "polls plus." You can read about the differences here. [...] If someone will claim that 538 has changed its methodology on the fly -- I will wonder if that that someone understands the methodologies in the first place. It suggests someone is incurious or has long finished interrogating 538's nuts and bolts. A 'fixed mindset'? Perhaps.[...] The way I understand the issue, the three models have not changed in underlying methodology since the three forecast models were unveiled. Does that make sense? William, My goodness, is that a gotcha and all? Well, I guess that depends. I meant to make clear that there are three kinds of 'forecasts' available at 538. You did not disagree that there are three kinds of forecasts: polls-only, polls-plus, and 'Now-cast' ... so what you do with that information may depend on your assessment of their value to political campaigning. There are readers besides you who may now have a better understanding of the differences in the three products; I don't presume to know your mind. Knowing the differences between the three models is useful to assessing their value -- if any -- to you. Gotcha is kind of a blob term or label. It poisons the well by denigrating any 'corrective' or 'update' -- instead of engaging with the details, far easier to tap with a label-gun. Gotcha then can be used as an all-purpose tool of evasion. Forgive me for my own presumptions, but I don't think you find the 538 forecast products to be interesting. I mean, interesting as a data-source. Why I suggest this is for two reasons: understanding the limits of the forecasts, meaning what they can and cannot show -- and using the forecasts as 'intelligence.' This 'intelligence' of course needs careful checking. That is probably the most boring part of 'Gotcha'ing, working out the details. Look at it this way, readers: if the Trump campaign 'cares' about polling, then they have employed their own polls-analysts. These analysts do not depend on a single source or a single aggregate running average. They seek information of strengths and weaknesses in their own and the opposing camp. They seek reliable information about 'how we are doing' in various states as the election unrolls. They will use 538 to augment their own internal intelligence gathering -- just as they will use the data compiled at RealClearPolitics, and at other aggregating and forecasting sites (like Front-loading HQ, Sabato's Crystal Ball, etc). Quote Let's suppose you're right to doubt. Doubt what? What is it that I am right to doubt? This is a confusing aside -- as it does not follow any of my plain statements. Quote Let's see if I can extract a few choice phrases from the [Rush Limbaugh] quote above to help my poor stoopid mook self to understand The Oracle. [1] This is a guy that the Democrats relied on for comfort... [2] Nate Silver then was analyzing polls. That's what he does. He analyzes polls; then makes percentage predictions based on his unique analysis. [3]... Nate Silver has secret ways of analyzing results and making in-depth predictions... [4] See, this is why this is special. Nobody else has a "polls-only" or "polls-plus," but Nate Silver does! Fair comment from Rush in #1. It is almost conventional wisdom. #2 is a simple kind of distinction that by now everyone reading this thread understands: 538 does not run polls. It takes polls as inputs to its forecast products. It does indeed 'analyze' polls, in several senses. It analyzes (checks for 'accuracy') polling success in the wide marketplace of public polling -- meaning PPP has a rating and Gravis has a rating, more or less objectively attained by comparing accuracy and other qualities, it analyzes the polling 'business' by giving grades to various houses. Further, it analyzes 'trends' on various scales as they emerge from polling aggregates. This is the demographic 'breakout' analysis that so consumes both the GOP and the Demorats. I can list more ways that analysis is provided -- but here note that I offer no value judgement. I am simply identifying the product-range. As an aside, Michael, sometimes I feel a bit autistic 'reading' your emoticons. So your comment on #2 is murky to me: "Oh... Now I get it! Silver looks at polls, he doesn't do polls! Who or what are you having fun with here? Is it that you always 'got it' and you never thought 538 ran polls? I don't know. Anyway, the Rush quote #3 -- Nate Silver has secret ways of analyzing results and making in-depth predictions... "Secret Ways" is a loaded term. The 'secret ways' of a 538 business product are not as entirely opaque as the loaded term implies. I will search for Rush discussing the published methodologies of the forecast trio. At a conceptual level, the methodologies are not 'secret' -- it is relatively easy to understand what distinguishes the Nowcast from the other two. I mean to say that the methods are examinable, to a reasonable degree. Not 'secreted' ... This is borne out by Nate Silver's mea culpa for his famously-wrong analyses of the Trump phenomena. As he explains, it was the non-math analysis,, the 'feeling,' the seat-of-the-pantys impressions and unverifiable mental 'weight' of expectations that fouled his product. (Oops... Secret ways? Now I don't get it. ) The smiley gets in the way. If you read the 538 explanations of methodology in the forecasts (especially the differences between), then you get your own understanding. Sometimes a single phrase, however loaded, does no work in argument. [4] See, this is why this is special. Nobody else has a "polls-only" or "polls-plus," but Nate Silver does! Not true. Frontloading and Sabato and 270toWin all use a 'polls-plus' instrument that adds or lightens weights according to variables such as Presidential approval rating, econometrics of different stripes. That is why the polls-plus can be more interesting and predictive than polls-only. In any case, think of how an Objectivist would offer a competing and better product than the Hillary-Zombies at 538. A Roark in the field of electoral forecasting. He (and the O'vish) can learn from and overcome mistakes made in the field. They want to offer as accurate a forecast as possible, and they want the information they trade in to be more reliable than the competition. And if They were We? We would need to 'take in' every last sufficient polling product made by others and add them into our design. We'd need to accurately assess things like "which party is up or down in which areas of Pennsylvania" -- according to valid and reliable soundings, history, contingency, expectations, electoral 'theory' and design. (Now I think I get it... but maybe not... you know... hmmmm... Yeah. I am glad that we have competing products to compare. And plenty of time to spare before November 8th. Michael, your arguments over the year have shown ambivalence about polling. I suppose that you are not too curious about the ins-and-outs of the various forecast products out there. You are sure in your own forecast, which is made at the highest level, the binary level. Me, I find that you figuring Trump will win is the real 'secret way' in a sense. Your instruments of measurement are in your heart, I think. Polls don't matter on that level. One thing we all do is get down to the level of analysis in a Presidential race that makes sense to us, that feels right, that is reasonable in scope. Nate Silver and the Gang of Hate at 538 are doing what we all do: taking in 'soundings' and interpreting them in forecasts. Personally, I am most interested in the states that are showing significant differences from the last two Presidentials. The electoral college arithmetic depends on these states, so I am most interested in what these state polls show just before and after the debates between Trump and Clinton. And my analysis of what they show is informed by my understanding of fundamentals, of demographics, historical comparisons. What's your secret, Michael? Quote I really like this part: And last week, Nate Silver was telling all of Democrats (summarized), "Don't sweat it! Right now, we got Trump at maybe 10 to 15% chance of winning the White House." . . . In his 'Now-cast' election model for who would win if ballots were cast [today], Silver gave the Republican nominee a 57.5% chance of winning the presidency," up from 10 or 15%. So Guru Nate Silver says Donald Trump's chances of winning jumped by 30-40% in one week? Last week Trump wasn't even in the running, but this week he's winning? Last week Trump was a statistical nobody, but this week he's everything? One week? Exactly. You are doing the fact-checking on Rush Limbaugh. One week he [Trump] jumped by 30-40% at 538? One week? From 10 or 15% to 57.5% chance? The answer is of course ... Now-cast; see the line as it moves across the months since June ... Polls only: Polls-plus: Quote That's one hell of a jump for one hell of an algorithm, especially since nothing else in the news reflected such a sharp jump (not even the RNC convention). And, of course, The Oracle didn't, he wouldn't, he couldn't, never ever ever ever change a single parameter in his methodology. What is clear is that no argument is made that Nate Silver/538 altered the methodology of the three forecasts on the sly -- since they were first unveiled to the public. If somebody wants to make a persuasive case that something hinky went down, I am open to it. Argument by sinister speculation is not as persuasive as a warranted claim. Unless an absence of evidence is taken as evidence, there is zero warrant backing a claim that some unseen tinkering accomplished the resulting trends and surprises in the 538 products. Looking back at the snapshots of the three interactive forecast pages in the spoiler, it seems Rush's Argument** is based on a weird and otherwise inexplicable one-week jump in the metrics (in the spoiler above). Rush is arguing along the lines that Nate Silver is if not corrupt-as-only-a-Democrat-can-be, from which inference can be taken that mysterious fingers are being applied surreptitiously. This is arguable. What is not arguable is that Rush misstates the 'weird' swing. It is not weird that the race has 'tightened up.' Any Red Hat enthusiast might be heartened by the general impression that Mr Trump is Winning, so I find it odd that Red Hat good news (derived from public opinion up to this point) is viewed with such suspicion. It seems more Occam-likely that forecast probabilities changed because inputs changed. Exactly how a rational personal mental calculus would change in light of fresh inputs. How does one get the true-est input 'soundings' to mix into the secret ways of forecasting? It's an individual thing, I think. Depending on the level of analysis (popular vote, trends, historical comparison, state-by-state 'swingers' and 'leaners,' etc), what everyone needs are valid inputs and valid assumptions. I believe that one can choose the most objective means to reason out the details. To those who want to set aside 538 products and opinions, that is a perfectly sane thing to do. It is one's own 'predictions' that are really on trial, not the motley of more-or-less rigorous predictions from the predictions houses. They will at best supplement our individual predictive certainty. In the end, incorrect predictions by OLers will be the ones under critical review and discussion post-November. My prediction today: a tight-ish race in national numbers, estimates of total popular vote, until at least September. In the swing-states, no real prediction other than if-then, if-else, else-if. I watch Pennsylvania, Ohio and Florida most closely. I assume that the least predictive polls are the last month's batch. I predict that the outcome of the election will be easier to estimate the closer we get to November. If Mr Trump is going to win enough electoral votes, then by mid-October it will be well-apparent that he will take Florida, Pennsylvania and Ohio. Else if ... he is not positioned in the hard-slogging county-by-county, precinct-by-precinct get out the vote in that trio, if his numbers lag Clinton's in must-win combos of states, then Mr Trump will lose to Mrs Clinton. And then we will talk about who predicted what and when, and why it matters, if it matters, and what it means. No hearts will be broken. ______________ ** TBA Edited July 31, 2016 by william.scherk Added screen-caps of 538's interactive forecast calendar graph; TBA- Rush's argument. Link to comment Share on other sites More sharing options...
Michael Stuart Kelly Posted July 31, 2016 Share Posted July 31, 2016 On 7/31/2016 at 2:55 PM, william.scherk said: Michael, your arguments over the year have shown ambivalence about polling. William, That's not exactly true. I am ambivalent about polls when pollsters stop trying to predict what happens and try to use polls to manipulate public opinion. That's most polls and pollsters these days. So the real issue with polls is story-wars, not predictions. Just look at how much crow they all had to eat about the primaries. We don't need to poll the pollsters to know they were wrong. Michael Link to comment Share on other sites More sharing options...
Michael Stuart Kelly Posted July 31, 2016 Share Posted July 31, 2016 8 minutes ago, william.scherk said: Who or what are you having fun with here? Is it that you always 'got it' and you never thought 538 ran polls? William, Did you forget? You once made a big production out of schooling me on this. Michael Link to comment Share on other sites More sharing options...
KorbenDallas Posted August 1, 2016 Share Posted August 1, 2016 WaPo is getting in on the predicting fun: "Donald Trump has a totally plausible path to 270 electoral votes:" https://www.washingtonpost.com/news/the-fix/wp/2016/08/01/donald-trump-has-a-totally-plausible-path-to-270-electoral-votes/?tid=pm_politics_pop_b Quote That map gives Trump 273 electoral votes to Clinton's 265. He wins. It's not crazy, right? Link to comment Share on other sites More sharing options...
BaalChatzaf Posted August 1, 2016 Author Share Posted August 1, 2016 1 hour ago, KorbenDallas said: WaPo is getting in on the predicting fun: "Donald Trump has a totally plausible path to 270 electoral votes:" https://www.washingtonpost.com/news/the-fix/wp/2016/08/01/donald-trump-has-a-totally-plausible-path-to-270-electoral-votes/?tid=pm_politics_pop_b Ohio is the key state here. Link to comment Share on other sites More sharing options...
william.scherk Posted August 2, 2016 Share Posted August 2, 2016 19 hours ago, KorbenDallas said: WaPo is getting in on the predicting fun: "Donald Trump has a totally plausible path to 270 electoral votes:" Interesting details at the link ... For all of the worry about the Blue Wall and Trump's historically poor numbers among Hispanics, the path laid out by the [NY]Times for Trump to get to 270 electoral votes is entirely plausible. Here's what the electoral map looks like if Trump wins those three states and nothing else changes from the 2012 election. 17 hours ago, BaalChatzaf said: Ohio is the key state here. Winning Ohio is not enough in the plausible scenario. Florida and Pennsylvania will, if flipped, bring the Trump EC numbers over 270. See embedded map below. I bet that the Trump campaign is spending money on internal polls (private) in these three states. And I bet the party will gear up its Get out the Vote machinery. I recommend (again) the 270toWin interactive Electoral College maps. Calculations are very simple -- win a state, win that state's electors. By keeping an eye on credible and reliable polling data as we get closer to decision, the plausible Trump victory in PA, OH and FL may move past 'possible' to 'probable.' Click the map to create your own at 270toWin.com -- Link to comment Share on other sites More sharing options...
KorbenDallas Posted August 2, 2016 Share Posted August 2, 2016 52 minutes ago, william.scherk said: Interesting details at the link ... I'm wondering if WaPo, Nate, and others are factoring this into their estimates: "Tim Kaine promises bill to legalize illegal immigrants in ‘first 100 days’" http://www.washingtontimes.com/news/2016/jul/25/tim-kaine-promises-bill-legalize-illegal-immigrant/ Quote PHILADELPHIA — A new Clinton administration would pursue a bill to legalize illegal immigrants in “the first 100 days” of her tenure, vice presidential candidate Tim Kaine told Spanish-language network Telemundo in an interview Monday [July 25, 2016], presenting a deep contrast with Republicans. Link to comment Share on other sites More sharing options...
william.scherk Posted August 2, 2016 Share Posted August 2, 2016 37 minutes ago, KorbenDallas said: I'm wondering if WaPo, Nate, and others are factoring this into their estimates: "Tim Kaine promises bill to legalize illegal immigrants in ‘first 100 days’" It depends on how the issue percolates into voter consciousness, and then depends on public polling that can capture a shift in opinion. If this is a big 'decider' kind of issue -- if the Trump/GOP campaigns drill on the message -- then a change in polling will be evident, and the Clinton-Kaine ticket will suffer. My Spanish is not as good as my French, but I could follow the conversation at Telemundo. FoxNews Latino has also covered the interview/s (which occurred July 25 or so). FoxNews Latino: Quote Wasting little time in exploiting his Spanish skills, the presumptive vice presidential nominee for the Democratic Party Sen. Tim Kaine told the two top Hispanic networks that a Hillary Clinton administration would tackle immigration reform within the first 100 days in office. The senator from Virginia, who spent a year in Honduras as a young man, gave interviews in Spanish to Univision and Telemundo on Sunday in his hometown of Richmond. Both aired on Monday. Kaine said that he would work hard to make Clinton’s promise to make introducing an immigration reform bill in Congress a priority. “Hillary will do that in the administration’s first 100 days,” Kaine said in Telemundo. “With my Senate background with colleagues and party officials, I will help with that effort.” “Senators are already talking about how, after January, if Hillary is president, we should work on immigration reform quickly,” he told Univision. -- I have a couple of responses to MSK's thoughtful clarifications in the pipeline Your browser does not support iframes. Link to comment Share on other sites More sharing options...
william.scherk Posted August 3, 2016 Share Posted August 3, 2016 On 8/1/2016 at 2:56 PM, KorbenDallas said: WaPo is getting in on the predicting fun: "Donald Trump has a totally plausible path to 270 electoral votes:" It's fun for almost everyone, no? I missed this on ABC: semi-quasi 'forecasting' of If Election Today. It features a round-table including Greta from Fox and Cenk from the TYT online network. I hadn't seen Cenk do his schtick on big network opinions shows before. He certainly had a pointed, if not nasty message. But of note are how easily the not-predictions are made. It's fun. It's too early, which makes for the extra zing. This is Greta's plot: Don't give up on the segment because of the stupid title -- Cenk becomes the outlier, with a nod to his political wisdom from Greta at the end: - from the TYTN Youtube page for the video: Published on Aug 1, 2016 · 1 day ago Cenk Uygur host of The Young Turks recently made his first appearance on This Week with George Stephanopoulos on ABC News. Cenk appeared on the powerhouse roundtable with Jonathan Karl, Greta Van Susteren, Alex Wagner and Kristen Soltis Anderson. Greta's reasonable point is that the debates will play a bigger part in persuading the persuadable / undecided than the day to day hoopla of polls right now. Link to comment Share on other sites More sharing options...
Michael Stuart Kelly Posted August 4, 2016 Share Posted August 4, 2016 2 hours ago, william.scherk said: I hadn't seen Cenk do his schtick on big network opinions shows before. He certainly had a pointed, if not nasty message. William, I saw that when it happened. You forgot to mention that Cenk was the only one saying Trump would win if the election were today. You did say he was an outlier, but you didn't say what he was outlying on. Cenk didn't like saying Trump would win, he hated it, but he was the only one to say it. btw - I do agree that the debates are going to count big with undecideds in this election. I'm not sure how big that number will be by then, though. There's a lot of mud coming down the pike and if a crapload more sticks to one candidate over the other, there will be a smaller undecideds pool. Michael Link to comment Share on other sites More sharing options...
Wolf DeVoon Posted August 4, 2016 Share Posted August 4, 2016 To paraphrase an old football coach, if I ever need a brain transplant I'll want one from a reporter, that way I know it will never have been used. (comment at Zero Hedge) Link to comment Share on other sites More sharing options...
william.scherk Posted August 4, 2016 Share Posted August 4, 2016 (edited) 43 minutes ago, Michael Stuart Kelly said: You forgot to mention that Cenk was the only one saying Trump would win if the election were today. You did say he was an outlier, but you didn't say what he was outlying on. It could be deduced that as outlier to all other state plots thus so to Greta's plot. But I mostly meant to leave some suspense. It was interesting to see the participants quick play of variables. As to details of Cenk's "Election Today" Now-cast outlier, I figure he could very well be right about the configuration of a Trump win, it is close to conventional wisdom that Florida, Ohio and Pennsylvania are key, lock, tumbler to a Trump victory. 97 days to go. Fun. Hoopla. What-ifs. In a big noisy churn. Cenk's plot of Election Today. Those three key states we talked about are in Trump hands. I see FoxNews has released a poll that sees Trump down 10 pts to Clinton in a three-way with Johnson. Rigged! We need to find out what is going on. Edited August 4, 2016 by william.scherk Link to comment Share on other sites More sharing options...
Brant Gaede Posted August 4, 2016 Share Posted August 4, 2016 The media seem to be getting hysterical about Trump. The latest is about the Republican Powers That Be are concerned about Trump dropping out. (Trump is descending into Nutsville.) What are we gonna do? What are we gonna do? Yeah, right. Trump does seem to be preparing his ego for a soft landing. --Brant Link to comment Share on other sites More sharing options...
Wolf DeVoon Posted August 4, 2016 Share Posted August 4, 2016 12 minutes ago, Brant Gaede said: The media seem to be getting hysterical about Trump. The latest is about the Republican Powers That Be are concerned about Trump dropping out. (Trump is descending into Nutsville.) What are we gonna do? What are we gonna do? Yeah, right. Trump does seem to be preparing his ego for a soft landing. --Brant Salem talker Michael Medved full tilt attack mode, "Never Trump" interview guests five days a week, absolutely certain that Trump is mentally unstable. I think it's funny. So-called "conservatives" will do anything to stop Trump because he doesn't need them, doesn't give a crap about them. If he wins, they're sunk. Link to comment Share on other sites More sharing options...
BaalChatzaf Posted August 4, 2016 Author Share Posted August 4, 2016 5 hours ago, wolfdevoon said: Salem talker Michael Medved full tilt attack mode, "Never Trump" interview guests five days a week, absolutely certain that Trump is mentally unstable. I think it's funny. So-called "conservatives" will do anything to stop Trump because he doesn't need them, doesn't give a crap about them. If he wins, they're sunk. That is the same kind of crap that was used against Barry Goldwater back in 1963. The canard that Goldwater was insane was spread about. Link to comment Share on other sites More sharing options...
KorbenDallas Posted August 4, 2016 Share Posted August 4, 2016 21 hours ago, william.scherk said: I missed this on ABC: semi-quasi 'forecasting' of If Election Today. It features a round-table including Greta from Fox and Cenk from the TYT online network. I hadn't seen Cenk do his schtick on big network opinions shows before. He certainly had a pointed, if not nasty message. But of note are how easily the not-predictions are made. It's fun. It's too early, which makes for the extra zing. Liked the video, haven't seen him on a main network either 18 hours ago, william.scherk said: Cenk's plot of Election Today. Those three key states we talked about are in Trump hands. I see FoxNews has released a poll that sees Trump down 10 pts to Clinton in a three-way with Johnson. Rigged! We need to find out what is going on. I'm liking this map, and Cenk even better, hah.. I noticed he flipped NV, in 2012 it was blue but from 2000-2008 red. Here is what Trump said in Feb after winning the primary there, "You know what I'm really happy about, because we've saying it for a long time? Forty-six percent with the Hispanics! Forty-six percent! Number one with Hispanics!" I don't think he'll be that popular with Hispanics in the general election. The wall + deportation vs. no wall, legalization will become a big issue in the debates, one I don't think they'll vote for. Edit: Noting flipping NV back to blue doesn't change Cenk's outcome for Trump, but I think the Hispanic vote will be an issue nationally and could change the map more.. Link to comment Share on other sites More sharing options...
william.scherk Posted August 5, 2016 Share Posted August 5, 2016 (edited) 4 hours ago, KorbenDallas said: I noticed [Cenk] flipped NV, in 2012 it was blue but from 2000-2008 red. Here is what Trump said in Feb after winning the primary there, "You know what I'm really happy about, because we've saying it for a long time? Forty-six percent with the Hispanics! Forty-six percent! Number one with Hispanics!" I thought at the time that it was reasonable to think that there are a relevant number of Hispanic Republicans in Nevada. And I thought it reasonable that Trump would be thrilled at getting near-half the votes in the Nevada caucus outing from that potential pool. Good news. Then I thought, what are the absolute numbers who voted for the GOP candidates as a whole, then I went to look for the numbers on the Democrat side. I figured that if there were eg. six thousand Latino votes implied by entrance polls, what mattered was the comparison of which party got what from that demographic at caucus, with that comparison balanced by the relative size of the likely general electorate. Exactly how many people are we talking about? And that is what I figured Donald Trump was asking his staffers: exactly how many people are we talking about? Are they representative of the larger population of likely voters? -- roughly 66k GOP caucus-goers. Of them, an estimated 1500 Hispanic participants. That is a three percent or thereabouts. The sample size in the entrance poll was inexactly 100 to 200 of those 66k GOPers.** Does anyone else remember investigating this good news, and how it would mean great things for Trump and 'the hispanics'? Quote I don't think he'll be that popular with Hispanics in the general election. The wall + deportation vs. no wall, legalization will become a big issue in the debates, one I don't think they'll vote for. The last reasonable poll found him with miserable national Latino numbers compared to Romney. But I don't think that this is set luck fudge, even in Florida. The underlying taste for an outsider is still sharp within the GOPish and Independent electorate, at least in my crystal ball. This is one of those cash-out moments when we probably best say, "See you back here late September." Or we can dig deeper into trends in the three key states. Why can't Trump slowly move into head-to-head poll numbers as the weeks go by, in the trio? His campaign has as many days to run as would two long Canadian campaigns. Things can move, people in the aggregate can shift allegiances over that long a period. Of course, in some sense, Mr Trump must beat the Hoopla. Even with a hostile, traitorous barrage of hate-Hoopla from large-audience legacy media, his millions of supporters are mostly unmoved. What about Colorado? Anyone see that teetering into the Blue? -- I am now cracking my file on Regression to the Mean. ** -- from the Cursed Ogre/Oracle's lair: Did Trump win Hispanics in Nevada? You can be sure that Trump will tell us he did! There was a lot of nerd-fighting over who won the Hispanic vote in the Democratic caucuses in Nevada, and we suspect there will be some over the Republican caucuses as well. Indeed, the entrance poll had Trump beating Rubio 45 percent to 28 percent among Hispanics. But keep in mind that the sample size on that result is somewhere between 100 and 200 people. That means the margin of sampling error for the Hispanic subgroup is near +/- 10 percentage points (or even higher). Perhaps more importantly, just 8 percent of Republican voters were Hispanic (or 1 percent of the Nevadan Hispanic population), and they are not politically representative of the larger Hispanic community. Edited August 5, 2016 by william.scherk Not exactly Added paragraph with links and numbers, from Nate Link to comment Share on other sites More sharing options...
william.scherk Posted August 5, 2016 Share Posted August 5, 2016 From the Oracle's Lair -- Trump’s Campaign Focused On Attracting Unlikely Voters, A Memo Shows -- an intriguing article that looks at the Trump campaign's efforts to find and bring to vote the 'unlikely' voter during the primary and caucus season, to find those individuals who may have been turned off or rarely voted in the past --- and turn them on to Trump. The article quotes from one of the campaign's data-workers (for lack of a better term), and cites a memo from early in the race. The thinking involved strikes me as pretty smart. The 'identification' of the 'unlikelies' as individual contactable assets is good political housekeeping, and in a way cousin to MIchael's notions about The Unseen. There is a persuadable cohort, eligible to vote, and willing to get back in the game in 2016, able to be energized by the Trump phenomenon. Here's a brief excerpt. Those who ignore the Oracle's opinion and campaign analysis can skip it, sanely. It may only underline something already known: hidden GOTV strengths that will be exploited over the next 94 days. Key term: "Low Propensity. In an effort to more accurately discern which potential voters Trump would be most likely to appeal to, Braynard commissioned polls — even though his candidate had previously denied the need for such internal data work. “I don’t have pollsters,” Trump said on “Meet the Press” in August of 2015. “I don’t want to waste money on pollsters. I don’t want to be unreal. I want to be me.” Braynard used voter issue models formulated by the data firm HaystaqDNA and sold by L2 to help refine his sense for who a likely Trump voter might be, conducting his own analysis to determine which characteristics made people likely to support Trump. “These are voters who do not think ISIS can be defeated with hashtags and do not think that the solution to BLM [Black Lives Matter] murdering cops is for cops to do a better job listening,” Braynard said. “Pat Buchanan, I think, accurately described these people as conservatives of the heart — these are law and order people who are rich in the wisdom of the Old Testament.” With Haystaq, Braynard said he was able to determine with up to 80 percent likelihood whether or not a person was a Trump voter. Once the campaign had identified that pool of voters, it pursued the low-propensity number among them with intensified phone contact and information about where and when to vote. Link to comment Share on other sites More sharing options...
KorbenDallas Posted August 12, 2016 Share Posted August 12, 2016 Ouch, "Exclusive Fox News Latino poll 2016 results" http://latino.foxnews.com/latino/politics/2016/08/11/exclusive-fox-news-latino-poll-2016-presidential-election/ [...] 2016 Presidential Vote Preference Among Latino Registered Voters Now May 2016 Hillary Clinton 59% 58% Donald Trump 17% 21% Gary Johnson 16% 9% August 7-10, 2016 Latino Registered Voters + or - 3.5% Pts. Opinion Of… Hillary Donald Clinton Trump Favorable 55% 15% Unfavorable 41% 82% August 7-10, 2016 Latino Registered Voters + or - 3.5% Pts. Opinion Of Hillary Clinton Now May 2016 Favorable 55% 56% Unfavorable 41% 41% August 7-10, 2016 Latino Registered Voters + or - 3.5% Pts. Opinion Of Donald Trump Now May 2016 Favorable 15% 23% Unfavorable 82% 74% August 7-10, 2016 Latino Registered Voters + or - 3.5% Pts. [...] Link to comment Share on other sites More sharing options...
KorbenDallas Posted August 13, 2016 Share Posted August 13, 2016 Larry Sabato has the race 347 to 191 for Hillary, he mentions the Hispanic vote. Not looking good for the #TrumpTrain---I wonder if they sell those delicious taco bowls on it: Watch the latest video at video.foxbusiness.com Link to comment Share on other sites More sharing options...
william.scherk Posted August 13, 2016 Share Posted August 13, 2016 13 hours ago, KorbenDallas said: Larry Sabato has the race 347 to 191 for Hillary, he mentions the Hispanic vote. Not looking good for the #TrumpTrain As I said above, I think this is one of those cash-out moments. "See you back here in late September." From my point of view, it is simply too early for any of these models to predict the outcome in November. Neither 538 nor Sabato can do it. Somewhere in late September, long after any 'convention ball bounce' has deflated, that is when I think we could begin to make confident predictions. And even then ... so much can change in the remaining 85 days or so. We may see changes in the electoral mood caused by unforeseen events. Or, so says my cousin Cassandra. That said, it does appear that Hispanic support for Mr Trump is stuck under that of Mitt Romney and John McCain at this point in the earlier contests. Can that level of support change? Will it change? It is hard to know right now. I am guessing Yes. Back to Nate Silver and the Ogres at 538, here's an example of their Fantasy Politics, from yesterday: What a Clinton Landslide Might Look Like ... Link to comment Share on other sites More sharing options...
william.scherk Posted August 17, 2016 Share Posted August 17, 2016 Trump pivots 360 degrees. Innovative. https://t.co/PbLbMVehuL — Nate Silver (@NateSilver538) August 17, 2016 Link to comment Share on other sites More sharing options...
Recommended Posts
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now