So what the h*ll does this mean??? Hillary is now down AGAIN on the 538 site......

[a kennedy]

Is it because the flower woman is coming to the debate???

http://projects.fivethirtyeight.com/2016-election-forecast/?ex_cid=rrpromo

[The Velveteen Ocelot]

1. Of course not.

The polls don't move that fast - it takes days, not hours, especially for state polls. What we might be seeing on 538 is still some leftovers from last week. Anyhow, whether or not she comes she is small potatoes; if she has any effect at all it will be to HRC's advantage.

[Farmgirl1961]

2. But it doesn't make any sense

There was slow but steady movement upward...then as of yesterday it started sliding. If it was leftover from before, she would have continued to slide downward.

[BadDog40]

7. A new PA poll came in from Muhlenberg College

That showed Hillary leading by 2 points, same poll 10 days ago had her leading by 8 points. I wouldn't put to much into it.

http://projects.fivethirtyeight.com/2016-election-forecast/pennsylvania/

[Dawson Leery]

3. Nate did some consulting work for the GOP last month.

He was telling what they can do to win. He can't be trusted.

[PJMcK]

14. That is not correct (n/t)

[oberliner]

24. This is a lie

Delete it.

[Demsrule86]

41. I have concluded that too...

Trump is winning less vote from a variety of groups than Romney, and Silver consistently had Obama with above a 90% chance...Silver is playing footsie with the GOP and slanting his data with all sorts of GOP enhancing models. I believe Sam Wang and the new gun in town Nate Cohn will prove more accurate this year. They both have her up to a probability of about 70%. Also to be fair, Nate's numbers do not include Thursday's polls which were good for Hillary.

[Dawson Leery]

54. +1,000,000

[SaschaHM]

4. They added a poll that had Clinton up by 2 in PA. That's why there is a slight shift downward.

[DarthDem]

5. No Need to Worry About Minute Shifts

. . . and summon the very concerned.

Silver would claim some reasoning for the number shifting and then say he can't explain it because the computer runs itself.

The real reason for the tracker changing so often is to drive clicks and generate revenue for ESPN.

[MFM008]

6. I'm done with checking him.

Period.

[Loki Liesmith]

8. closer poll than usual in PA

Hillary only up by 3.

[Farmgirl1961]

9. Is that just one poll or several?

It seems like she was up by 7 or so just a day or two ago...

[Loki Liesmith]

11. 1 poll

Muhlenberg.

She's still up by about the same margin. This is just sampling error.

[Farmgirl1961]

13. I sure hope so...

Still -- best to have to Dems in PA working extra hard to GOTV.

[Farmgirl1961]

10. I find that so peculiar

I've always thought that Philadelphia, with a couple of million people, is largely democratic in nature.

[Loki Liesmith]

12. It is

If you have a jar full of marbles, half white and half black and pick a handful of ten marbles.

Sometimes you will get 8 white marbles and two black.

[Farmgirl1961]

15. I sure hope that the jar doesn't have half white and half black -- I mean orange marbles!

n/t

[unblock]

51. if you knew the first thing about statistics,

you'd know you should be working with skittles, not marbles.

[bettyellen]

16. Western PA is very conservative- they don't call it Pennsetucky

For nothing!

[Farmgirl1961]

17. Yes, I know. Is that where the votes were from?

n/t

[bettyellen]

18. I'm pretty sure the statewide polls pick people spread throughout the state...

[Kilgore]

30. All of PA is not the same

Check out this link to an article in the Atlantic.

http://www.theatlantic.com/politics/archive/2016/09/why-democrats-in-western-pennsylvania-are-voting-trump/499577/

[Demsrule86]

43. Sorry

PA is fools gold for the GOP...they have to win the Philly burbs and they can't...especially not with Trump.

[geek tragedy]

19. Washington Post just produced a bad national poll

Clinton +2 LV, tied RV

[LeftRant]

20. You shouldn't poll watch if it makes you nervous.

It almost always tightens at the end, and we're just not going to know until election day. There's also not much we can do about it other than fret unless you go volunteer or contribute money.

[LenaBaby61]

21. You shouldn't poll watch if it makes you nervous.

I'm not normally one who "panics" over polls. Heck, I really don't take them seriously until after Labor Day and after a 1st Debate. That's always been my way to stay calm etc.

However, you gotta wonder HOW Hillary can go from having a 7-8 or so point lead this week in Pennsylvania and as of today only be up 3 points. Nothing this week happened that I saw that was in a negative vein for her. In fact, tRump had a not so great week, yet he's only not 3 points behind her? I have to also admit that I haven't looked inside of the internals of this poll, so maybe I'll do it and maybe it will make sense. I mean really some of these polls have internals that are laughable. I do truly believe that some--of these pollsters/networks who employ certain pollsters want a horse race meme working so you'll watch their networks for the "Latest" breaking news on this presidential horse race is narrowing by leaps and bounds and where Hillary is no longer the lock she was earlier right after a very successful Democratic Convention (And oh by the way, watch US so we can bring in lots of 'nice' ad dollars/potential revenue to our stations/networks).

[Loki Liesmith]

27. You forgot to average

Hillary did not have an 8 point lead in PA last week.

She is not leading by 3 this week.

It's always been 5.

Average. Or take the median. Never rely on a single data point.

[LenaBaby61]

29. That's right.

"Average. Or take the median. Never rely on a single data point."

Thanks Loki

[LeftRant]

22. Indeed, they benefit from a close race.

But I don't think you can compare poll changes over 1 week with much certainty. The algorithms are best-guesses and always a work in progress, sample sizes are usually not huge (costs a lot to do), and there are many variances you can't account for. I would look more for trends over several weeks to see confirmable changes.

But I still think they always tighten up at the end. 2012 felt a little bit like a nail biter.

I continue to think that unless HRC screws something up bad or an Event of some kind happens, she's going to beat him. He's just so gross. There must be more than 50% of voters who can see that. Surely...

[Demsrule86]

23. Why do you post negative stuff about Hillary?

"Hope Hillary's voice holds up" And you mention her 'coughing'. And of course letting people know about the Libertarian Forum. And" why does Hillary have horrible comments on her twitter? I have never seen a positive post from you concerning Hillary.

http://www.democraticunderground.com/12512320662

The link for the coughing post is above...

And now you add a post about Gennifer Flowers.

[Kaleva]

26. Bingo!

[Demsrule86]

38. I remember posts sometimes.

Do some really think we don't understand what they are about? In this election with a candidate as bad as Trump, it is unbelievable.

[abetterkid]

34. Good catch...

I've seen ALOT of underhanded shade throwing coming from certain folk.

[Demsrule86]

39. Exactly...so sick of it...

We are not stupid and remember previous posts.

[oberliner]

25. Things often tighten like this

It is the ebb and flow of the elections season.

[democrattotheend]

28. Several bad polls today

Just when it looked like she was turning things around, things seem to have started turning back. For example, the Morning Call poll that had her up by 8 last week has her up by 2 now: https://assets.documentcloud.org/documents/3113757/2016-Muhlenberg-College-Morning-Call-late.pdf

And the latest ABC national poll also has her leading by only 2, compared to 8 last time.

I don't understand why things seem to be sliding again. Maybe the attacks in NYC and the rioting in Charlotte are making people think twice about Trump's ideas?

[Loki Liesmith]

32. The odds of getting an out of sample poll are about 1/20

1 out of every 20 polls is going to give you a result that is out of trend. When it's fairly close, you will get more counterintuittive results..

The more polls taken on an interval, the more likely this is.

Lots of people are releasing their polls pre-debate so they can announce big changes after the debate.

We are most likely ahead. Let's just see what happens tomorrow.

[Persondem]

35. You seem like a good person to ask ...

I have done some stats analyses but it sounds like you have lived stats so...
If you get a few minutes, take a look at this page from 538. It's a detailed look at how the forecast for Florida is put together with a list of their polls considered and their weights. Be sure to scroll down to see the explanation for how they came up with their forecast.

http://projects.fivethirtyeight.com/2016-election-forecast/florida/

I am curious as to what you think.

[Farmgirl1961]

31. I don't get it either

It's not like Hillary had a bad week. Trump did. Does Charlotte, NY, NJ and WA play into his hands?

[doc03]

33. Chuck Todd was just talking about how Trump is doing worse among

most groups of white voters than Romney did. Romney lost, Trump isn't getting the Latino or black vote and he is doing worse among
white voters than Romney how the fuck are the polls so close?

[LisaL]

36. It doesn't make any sense to me either.

If Trump isn't doing as well as Romney with whites, he sure as hell isn't doing as well as Romney with minorities. So why are the polls so close?
Are they assuming turnout will be low among democrats?

[democrattotheend]

37. That's what I am thinking

Some pollsters are basing their turnout model on 2004. Even if Trump underperforms Romney among every subgroup, he can still win if Democratic voters don't turn out enough.

[Demsrule86]

40. So the polls are rigged to help Trump...just like in 2012.

No reason to believe this election will be like 2004 which was very close and before we got Virginia.

[democrattotheend]

42. Turnout model based on 2004

Because 2004 was arguably more similar than 2012 or especially 2008 because people in 2004 didn't like either candidate and we don't have a dynamic, inspiring candidate who is expected to drive up turnout levels beyond what they typically are for Democratic constituencies.

You make a good point about Virginia though. It wasn't a swing state in 2004 so turnout will probably be higher there now than back then.

It's not just the public polls that are showing Trump having a better shot now than he did before. The fact that more Republicans like Cruz are coming on board shows that their internals are also showing movement in their direction. Back in August it looked like the party might give up on Trump and focus down ballot.

[Demsrule86]

44. I disagree with you completely

First of all, Hillary Clinton is not disliked among Democrats and some independents. I would remind you that she won the primary with millions of votes. And consider that in 04 Virginia was rock solid for the GOP. This is no longer the case. Trump is so awful...he will drive up the number of Democrats voting...the model is off and it is misleading at best and out and out fraud at worst. Polls show an overwhelming number of voters do not think Trump is qualified to be president. Given Trump's poor showing with white college-educated voters, Black voters, Jewish voters, Asian voters and Hispanic voters, he will not win. His numbers are much worse than Romney's were, and Romney lost badly.

[doc03]

47. I do think turnout will be very low this year. The Democrats I know

are just not excited by Hillary Clinton. In the past Democrats haven't turned out when they don't have
charismatic candidate like Obama or Bill Clinton. I have said this many times I haven't seen one Hillary
sign or bumper sticker yet this year in this area. I have never experienced an election with less enthusiasm
than this one and I have seen many clear back to JFK.

[Demsrule86]

45. I would also like to add that 2012 was more like 2004 than this year despite Virginia.

In both cases you had an incumbent president who was not that popular at the time of the election...Bush had the fall out from the 2000 election where he was really selected and not elected and Obama from the beating he took from his own party in 09 which lead to the election debacle in 2010...and his inability (unfair though it was ) to move a progressive agenda forward. My own son did not vote in 12...he just didn't see the point. This year, he told me would crawl across broken glass to vote for Hillary, and he was a Bernie supporter. So I believe 2012 was more similar to 2004 than this election is...and with Trump' numbers among various groups he can't win.

[triron]

46. As a side note

in New Mexico we have had an uptick in applications for citizenship from legal immigrant residents due to the deportation rhetoric coming from the Trump campaign. I would imagine that is happening in other states as well.

[BlueMTexpat]

48. Ignore it and get

yourself out to work like hell to get Hillary elected!

[RAFisher]

49. I'm so sick of these 538 conspiracy theories.

FiveThirtyEight goes into great detail how it works. But everytime it gets closer people come up with some unsubstantiated reason why it's getting closer.

[Foggyhill]

53. His "details" are hand waves; his model is what I call social science "math"...

His model is ridiculously arcane considering how little he's got to base it on
and can spit out whatever he wants it to spit out with mere tweeks in inputs
(which he also controls and whose motivation for inclusion is just as arcane).
Got the pile of university engineering, mgt and communication (including polling) coursework
to have this opinion and I'm comfortable with it.

[MyNameIsKhan]

50. This is purely a statistical model

So many polls coming out so they can sell themselves outside election cycle ... The polls have margin of error...

No worries we have 90 percent chance to win

[La Lioness Priyanka]

52. It means the zombie apocalypse is about to occur

Such drama