# Doubting the statistical analysis at 538 **URL:** \ **Category:** Politics & Elections\ **Created:** [September 2, 2016, 10:09pm UTC](https://boards.straightdope.com/t/doubting-the-statistical-analysis-at-538/764819 "2016-09-02T22:09:00Z")\ **Posts on this page:** 20\ **Page:** 2
**Author:** ![adaher](https://avatars.discourse-cdn.com/v4/letter/a/dec6dc/32.png) [@adaher](https://boards.straightdope.com/u/adaher)\ **Post date:** [September 3, 2016, 12:11pm UTC](https://boards.straightdope.com/t/doubting-the-statistical-analysis-at-538/764819/21 "2016-09-03T12:11:53Z")
> [@JKellyMap](#): > > I agree that the folks in these polls who have expressed an intention to vote for Gary Johnson should be looked at very carefully. I’m a Hillary supporter, and this is just about the only thing that worries me somewhat. Two reasons not to worry TOO much: 1. Silver expected Johnson to lose support by this point but he hasn’t, so now Silver’s model has him keeping most of his polled support. 2. The reason for this is that Trump is the least libertarian Republican candidate ever. It’s really tough for libertarian-leaning conservatives to justify a vote for him. I’m more worried about the true undecideds. Seems to me that of the two, Clinton is the more known quantity, so I have to think that the undecideds are waiting for Trump to give them a reason to get off the fence. Of course, my dream result is that they all break for Johnson and he finishes above 15%.🙂 ---
**Author:** ![asahi](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/asahi/32/8693_2.png) [@asahi](https://boards.straightdope.com/u/asahi)\ **Post date:** [September 3, 2016, 12:21pm UTC](https://boards.straightdope.com/t/doubting-the-statistical-analysis-at-538/764819/22 "2016-09-03T12:21:15Z")
Not looking at any hard data, the most accurate calculus would have to take into account past voting behavior in major statewide elections over the past 5 years, including gubernatorial and senatorial races, as well as the election of 2012. The problem with probability is it deals with the here and now. A prediction analysis in September doesn’t factor terrorist attacks, race riots, natural disasters, and scandals in October, or a freak snowstorm on election day. Most pollsters agree that Clinton is clearly winning but she’s also vulnerable. Both candidates are probably a little immune to controversy at this point, but not completely so. This race will probably come down to centrist white voters (educated mainstream white women in particular), and it will ultimately come down to which voters are more motivated to vote for their candidate on November 8. If Clinton goes into the final day of campaigning with a 3-5 percent lead in the polls, she probably wins the race comfortably. If it’s a 1-2 point lead, on the other hand, anything can happen. I may be wrong but I get the feeling that Trump’s voters are more committed to voting for a return to the 1950s and against the Clintons, and I think they will surprise pollsters on election day. But will there be enough of them, and will they be scattered in the right places? ---
**Author:** ![DSeid](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/dseid/32/20194_2.png) [@DSeid](https://boards.straightdope.com/u/DSeid)\ **Post date:** [September 3, 2016, 12:45pm UTC](https://boards.straightdope.com/t/doubting-the-statistical-analysis-at-538/764819/23 "2016-09-03T12:45:18Z")
> [@JKellyMap](#): > > I agree that the folks in these polls who have expressed an intention to vote for Gary Johnson should be looked at very carefully. I’m a Hillary supporter, and this is just about the only thing that worries me somewhat. One way to look at them is to see what changes between the polls that include that option and the two-ways. Given no Johnson choice Clinton goes up more often than not. Not so sure that there is much reason to believe that Johnson supporters and undecideds will break big time Trumpward. > [@adaher](#): > > Wang and Silver had it out in 2014 and Silver took him to school. Wang relies too heavily on the equivalent of a Nowcast. Silver projects into the future, which you kinda need to do 60 days out. Silver attacked Wang but my read of that interaction was far from Wang being schooled. And FWIW Wang schooled Silver during the primaries. While Silver was highly confidently going on about how Trump’s collapse was soon to be and then continued with horse race narratives about contested conventions Wang was early on saying that the die was cast for Trump as the winner on the GOP side. (Both agreed on Clinton.) In terms how they each did predicting mid-term results in the final predictions, Wang and Silver both called them about the same … interestingly enough with Wang giving greater uncertainties than Silver did. Both are projecting into the future and both look to the past for guidance as to how the future may vary from conditions today. The problem though with learning from history is deciding which aspects of history are relevant for this stretch of future. Silver is reluctant to say that this patch of future will look like the past, maybe because he was so overconfident on Trump’s collapse in the primary, maybe because he incentivized to have a horse race narrative. And maybe this time will be different. Wang OTOH looks at the polls and sees that the polls have been behaving in a low variance election manner, as they have in presidential cycles since 1992, and thus has confidence that they will continue to behave in that manner. In any case this time they each represent extreme values in the spread of uncertainty vs confidence. Since I see more danger from overconfidence I’ll go with Silver but I really do think Wang is right. A Silverian hedge. ---
**Author:** ![Lord\_Feldon](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/lord_feldon/32/341_2.png) [@Lord\_Feldon](https://boards.straightdope.com/u/Lord_Feldon)\ **Post date:** [September 3, 2016, 12:57pm UTC](https://boards.straightdope.com/t/doubting-the-statistical-analysis-at-538/764819/24 "2016-09-03T12:57:09Z")
> [@CarnalK](#): > > The electoral map is too tough on Trump. The electoral map is tough on Trump _because_ he’s losing, though. It’s not really a separate thing, except in ultra-close, practically tied elections like 2000. ---
**Author:** ![adaher](https://avatars.discourse-cdn.com/v4/letter/a/dec6dc/32.png) [@adaher](https://boards.straightdope.com/u/adaher)\ **Post date:** [September 3, 2016, 1:00pm UTC](https://boards.straightdope.com/t/doubting-the-statistical-analysis-at-538/764819/25 "2016-09-03T13:00:02Z")
Silver covered that too. Even if it’s tied, that doesn’t mean Clinton’s going to win because Electoral College. ---
**Author:** ![DonLogan](https://avatars.discourse-cdn.com/v4/letter/d/51bf81/32.png) [@DonLogan](https://boards.straightdope.com/u/DonLogan)\ **Post date:** [September 3, 2016, 1:27pm UTC](https://boards.straightdope.com/t/doubting-the-statistical-analysis-at-538/764819/26 "2016-09-03T13:27:08Z")
Clinton has 273 in the bank. CO, NH, PA, VA. That’s it. Absent the biggest gaffe since the revival of the debates, it’s over. Clinton will get 2 “Johnson voters” for every 1 Trump gets. ---
**Author:** ![JKellyMap](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/jkellymap/32/15880_2.png) [@JKellyMap](https://boards.straightdope.com/u/JKellyMap)\ **Post date:** [September 3, 2016, 1:28pm UTC](https://boards.straightdope.com/t/doubting-the-statistical-analysis-at-538/764819/27 "2016-09-03T13:28:21Z")
> [@DSeid](#): > > One way to look at them is to see what changes between the polls that include that option and the two-ways. Given no Johnson choice Clinton goes up more often than not. Not so sure that there is much reason to believe that Johnson supporters and undecideds will break big time Trumpward. I wasn’t aware of that – thanks. I’ll sleep a little better tonight. ---
**Author:** ![CarnalK](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/carnalk/32/486_2.png) [@CarnalK](https://boards.straightdope.com/u/CarnalK)\ **Post date:** [September 3, 2016, 1:37pm UTC](https://boards.straightdope.com/t/doubting-the-statistical-analysis-at-538/764819/28 "2016-09-03T13:37:07Z")
> [@adaher](#): > > Silver covered that too. Even if it’s tied, that doesn’t mean Clinton’s going to win because Electoral College. On [Sabato’s Crystal Ball map](http://www.centerforpolitics.org/crystalball/2016-president/), even if Trump keeps all the “lean Republican” and steals all the “leans Democrat” States, he finishes with 265 electoral votes. So I’m betting he does not realistically have a 30% chance of winning because Electoral College. ---
**Author:** ![adaher](https://avatars.discourse-cdn.com/v4/letter/a/dec6dc/32.png) [@adaher](https://boards.straightdope.com/u/adaher)\ **Post date:** [September 3, 2016, 1:52pm UTC](https://boards.straightdope.com/t/doubting-the-statistical-analysis-at-538/764819/29 "2016-09-03T13:52:02Z")
THat’s with current polling. If he is 1 point ahead, there are a lot more lean Republican states. A popular vote win and an electoral college loss is always very unlikely. If Trump wins the popular vote, he almost certainly wins the Electoral College as well. ---
**Author:** ![DSeid](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/dseid/32/20194_2.png) [@DSeid](https://boards.straightdope.com/u/DSeid)\ **Post date:** [September 3, 2016, 1:54pm UTC](https://boards.straightdope.com/t/doubting-the-statistical-analysis-at-538/764819/30 "2016-09-03T13:54:07Z")
> [@DonLogan](#): > > Clinton has 273 in the bank. CO, NH, PA, VA. That’s it. … Well that _is_ it: Silver (polls-only) gives PA a 27% chance of going Trump and in a scenario in which PA has gone Trump then assuredly OH and FL have as well. Is PA, with an RCP average of Clinton +6.2 in four-way polling, 6.0 in two-way, a 27% Trump win? ---
**Author:** ![CarnalK](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/carnalk/32/486_2.png) [@CarnalK](https://boards.straightdope.com/u/CarnalK)\ **Post date:** [September 3, 2016, 2:04pm UTC](https://boards.straightdope.com/t/doubting-the-statistical-analysis-at-538/764819/31 "2016-09-03T14:04:58Z")
> [@adaher](#): > > THat’s with current polling. If he is 1 point ahead, there are a lot more lean Republican states. Yeah, but I gave him every single “lean Dem” state and he still comes up short, so “there would be a lot more lean Republican states” isn’t much of a rebuttal. That’s where those extra lean Republican states would come from. ---
**Author:** ![DonLogan](https://avatars.discourse-cdn.com/v4/letter/d/51bf81/32.png) [@DonLogan](https://boards.straightdope.com/u/DonLogan)\ **Post date:** [September 3, 2016, 2:29pm UTC](https://boards.straightdope.com/t/doubting-the-statistical-analysis-at-538/764819/32 "2016-09-03T14:29:09Z")
> [@DSeid](#): > > Well that _is_ it: Silver (polls-only) gives PA a 27% chance of going Trump and in a scenario in which PA has gone Trump then assuredly OH and FL have as well. > > Is PA, with an RCP average of Clinton +6.2 in four-way polling, 6.0 in two-way, a 27% Trump win? No, in my model, it’s 0%. ---
**Author:** ![suranyi](https://avatars.discourse-cdn.com/v4/letter/s/e36b37/32.png) [@suranyi](https://boards.straightdope.com/u/suranyi)\ **Post date:** [September 3, 2016, 5:07pm UTC](https://boards.straightdope.com/t/doubting-the-statistical-analysis-at-538/764819/33 "2016-09-03T17:07:15Z")
> [@DonLogan](#): > > Clinton has 273 in the bank. CO, NH, PA, VA. That’s it. Absent the biggest gaffe since the revival of the debates, it’s over. > > Clinton will get 2 “Johnson voters” for every 1 Trump gets. I really hope that’s true but I’m starting to worry. How are those 273 counted? Michigan and Wisconsin, for example, are looking far from certain for her at the moment. ---
**Author:** ![DonLogan](https://avatars.discourse-cdn.com/v4/letter/d/51bf81/32.png) [@DonLogan](https://boards.straightdope.com/u/DonLogan)\ **Post date:** [September 3, 2016, 6:43pm UTC](https://boards.straightdope.com/t/doubting-the-statistical-analysis-at-538/764819/34 "2016-09-03T18:43:04Z")
> [@suranyi](#): > > I really hope that’s true but I’m starting to worry. How are those 273 counted? Michigan and Wisconsin, for example, are looking far from certain for her at the moment. Trump is not going to win MI or WI any more than Clinton is going to win AZ or GA. I haven’t changed my thinking in the past 3 weeks or so, I still have the final electoral map result to be the same as 2012. You pre-answered the following question by characterizing MI and WI as being in play, but since you did ask about 273 - look at this map and tell me which blue state Trump is going to win. > **[270toWin - 2024 Presidential Election Interactive Map](https://www.270towin.com/maps/Eb7ox)** > > Create your own forecast for the 2024 presidential election ---
**Author:** ![asahi](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/asahi/32/8693_2.png) [@asahi](https://boards.straightdope.com/u/asahi)\ **Post date:** [September 3, 2016, 6:52pm UTC](https://boards.straightdope.com/t/doubting-the-statistical-analysis-at-538/764819/35 "2016-09-03T18:52:48Z")
> [@DonLogan](#): > > Clinton has 273 in the bank. CO, NH, PA, VA. That’s it. Absent the biggest gaffe since the revival of the debates, it’s over. > > Clinton will get 2 “Johnson voters” for every 1 Trump gets. It is absolutely not over. Clinton is definitely out in front, but over it is not. And if her campaign and supporters genuinely believe that, it’ll be over for Clinton. Hillary Clinton has a natural advantage over Trump due to the increased diversity of the electorate, but Hillary Clinton is incapable of pulling away from Trump. She’s just not likeable enough. She will have to fight in this election and work hard to earn the victory. ---
**Author:** ![asahi](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/asahi/32/8693_2.png) [@asahi](https://boards.straightdope.com/u/asahi)\ **Post date:** [September 3, 2016, 6:55pm UTC](https://boards.straightdope.com/t/doubting-the-statistical-analysis-at-538/764819/36 "2016-09-03T18:55:45Z")
> [@DonLogan](#): > > Trump is not going to win MI or WI any more than Clinton is going to win AZ or GA. > > I haven’t changed my thinking in the past 3 weeks or so, I still have the final electoral map result to be the same as 2012. > > You pre-answered the following question by characterizing MI and WI as being in play, but since you did ask about 273 - look at this map and tell me which blue state Trump is going to win. > > [270toWin - 2024 Presidential Election Interactive Map](http://www.270towin.com/maps/Eb7ox) Hillary Clinton is not Barack Obama. ---
**Author:** ![iiandyiiii](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/iiandyiiii/32/7924_2.png) [@iiandyiiii](https://boards.straightdope.com/u/iiandyiiii)\ **Post date:** [September 3, 2016, 6:57pm UTC](https://boards.straightdope.com/t/doubting-the-statistical-analysis-at-538/764819/37 "2016-09-03T18:57:18Z")
> [@adaher](#): > > Hillary Clinton stands today at 41.5% of the vote in a four way race. Trump may be doing worse, but with that many undecideds and third party voters there’s just no way you can put such high odds on Clinton. 75-25 is about right. There are still 10-20% of voters out there that can be persuaded. And Clinton seems to have stopped trying, instead hunkering down. The “total” (i.e. the 41.5%) is almost entirely meaningless. It’s the difference that is actually predictive. Obama’s total in '08 and '12 was well below the numbers he actually received on election day. ---
**Author:** ![DonLogan](https://avatars.discourse-cdn.com/v4/letter/d/51bf81/32.png) [@DonLogan](https://boards.straightdope.com/u/DonLogan)\ **Post date:** [September 3, 2016, 7:02pm UTC](https://boards.straightdope.com/t/doubting-the-statistical-analysis-at-538/764819/38 "2016-09-03T19:02:44Z")
> [@asahi](#): > > Hillary Clinton is not Barack Obama. She ain’t running against a war hero Senator or a governor who saved the Olympics and implemented the original Obamacare, either. ---
**Author:** ![asahi](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/asahi/32/8693_2.png) [@asahi](https://boards.straightdope.com/u/asahi)\ **Post date:** [September 3, 2016, 7:24pm UTC](https://boards.straightdope.com/t/doubting-the-statistical-analysis-at-538/764819/39 "2016-09-03T19:24:58Z")
> [@DonLogan](#): > > She ain’t running against a war hero Senator or a governor who saved the Olympics and implemented the original Obamacare, either. No, she’s not, but this is 2016, not 2012. We probably would be wise to avoid relying on precedents to predict what might happen this year. There has been nothing predictable about this election. Sure, some states like New York, California, Massachusetts are in the bag. But assuming that more centrist states that have recently elected republican governors and senators are going to vote for Clinton just because they voted for her party in 2012, 2008, 2000, or 1996 is an assumption that is not safe to make. ---
**Author:** ![elucidator](https://avatars.discourse-cdn.com/v4/letter/e/8edcca/32.png) [@elucidator](https://boards.straightdope.com/u/elucidator)\ **Post date:** [September 3, 2016, 8:09pm UTC](https://boards.straightdope.com/t/doubting-the-statistical-analysis-at-538/764819/40 "2016-09-03T20:09:17Z")
> [@asahi](#): > > …She will have to fight in this election and work hard to earn the victory. She will. She might be faulted on many counts, but “lazy” ain’t on the list. Plus, she needs, and \*we \*need, not just a win but a total clobbering. Which can be done. [Previous page](https://boards.straightdope.com/t/doubting-the-statistical-analysis-at-538/764819.md?page=1) [Next page](https://boards.straightdope.com/t/doubting-the-statistical-analysis-at-538/764819.md?page=3)