IOWA: The HuffPost/Pollster Charts show accelerating movement to Bernie.

[reformist2]

First chart is since the start of the year, second chart is from June 1 onward.

Look at those last few dark blue dots... Hillary is dropping so fast the trend line can't even keep up.

[restorefreedom]

1. it bears repeating

trend lines matter

[reformist2]

2. Especially when only 30% are following the race closely.... I'd love to see a breakdown that way.

[HereSince1628]

8. We need to remember that polling trends aren't processes with physical momentum

that can be confidently assumed to continue. Which isn't to say that public opinion can't have any component of momentum, just that momentum in public opinion isn't governed by physical laws.

People have a tendency to relate graphs to physical experiences, opinion isn't such a process

The trend in opinion tells a story locked in the history of the data that reveals it. That's not true for the projection of the trends in public attitude into the future. The future still has freedom to be influenced by events that effect/drive it.

Things to keep in mind when projecting include...

--what's happening to the size and leanings of the undecideds.

--what's happening to the pool of candidates ... actual or percieved adding or shrinking is going to matter.

--whether the last trend included some unusual effector on public opinion... hit with unfairness or making a faux pas.

--the presence of events that influence psychological and philosophical resistance to opinion changing... indictments or claims of impropriety etc.

[kenn3d]

3. RealClearPolitics displays the IA trends as even more dramatic.

RealClearPolitics displays the IA trends as even more dramatic.

http://www.realclearpolitics.com/epolls/2016/president/ia/iowa_democratic_presidential_caucus-3195.html

And notice that by their calculations the spread is now zero.
The 2 most recent polls in IA reported Sanders +1, and Sanders +10 respectively.

RCP and HuffPollster sites composite data from mostly the same major pollsters, but they each include or exclude some polls differently.

[abakan]

5. And the peasants rejoiced! nt

[bobbobbins01]

7. Uh oh, they crossed the streams

Feeling the Bern!

[kenn3d]

9. RCP and HuffPollster sites composite data from mostly the same major pollsters...

Here is the HuffPollster Iowa Democratic Caucus chart (Customized*)

*To include only the A and B rated pollster data (19 polls from 9 major polling firms).
*To exclude Automated/IVR/Online polls.
*To exclude Partisan and Sponsored polls
*To include Registered and Likely Voters
*Date Range 5/1/2015 to present.


I'm sure valid arguments might be presented as to the value or validity of such customized composite charts. I was only intrigued by the fact that widely referenced composite charts from different sites variously include/exclude specific polling data for whatever reasons. And many who challenge the results of various polls will often point to online or automated polling as being less credible or certain pollsters having reputations for inaccurate methods or historically poor results.

I find it very interesting that customizing the HuffPollster default dataset to disallow what might be considered as lower quality or suspect data, brings the resulting graph into much closer agreement with the RealClearPolitics default chart.

No flames please.
Anything can happen, so support your candidate and GOTV!

[Fawke Em]

4. Is 5 recs still the line to beat for the Greatest Page?

If so - K and R to the GP!

[Catherina]

6. Beautiful charts n/t

[sabrina 1]

10. A recent CNN poll confirms this trend. Bernie is within 10 points NATIONALLY

to Clinton now. And all withine less than four months.

And without Corporate funding or any support from the Dem Party.

He truly is the People's Candidate!

[DCBob]

12. There have been several polls since then and all show Hillary gaining ground.

[kenn3d]

14. Well....no. But of course, anything can happen...

The polls you've listed are from the HuffPollster National Democratic Primary composite chart dataset. The current default (all inclusive) plot is shown below.

Note: Using this all inclusive poll data:
On April 29th when Sanders announced his candidacy the spread was +55 Clinton.
On July 29th, (90 days later) the spread was +37 Clinton.
Today (45 days later) the spread is +20 Clinton.

Only 3 of the 8 polls in your list however, are top-rated (Grade A or B) live polls. (The rest are internet and/or automated polls.) These 3 are highly regarded but different pollsters (CBS/Times, ABC/Post, and CNN), each with unique methodologies, so no trend can/should be drawn by comparing them.

Here is the chart customized to show only top-rated pollster data, no internet or automated polls, no partisan or sponsored polls, and only registered "likely to vote" responses.

Note: Using this "customized" data:
Current spread is +13 Clinton

Despite its A rating the CNN poll might be an outlier... nevertheless, there really is no evidence here to support "Hillary gaining ground".

Of course, anything can happen...

GOTV!

[DCBob]

15. I was responding to the person who cited the CNN poll as if was something of significance.

That's a relatively stale poll and what I posted were all the polls since then.

Im not sure what your concern is with that?

[kenn3d]

16. And I was responding to you both...

The CNN poll was indeed part of something of major significance. Although it may not have so much significance in and of itself, and it truly does not define the current composite statistics or spread. That fact is really not due to any effect of the subsequent polls you listed however.

It's probably worth noting that we're still asking a lot to expect national polling to indicate much beyond name recognition at this point. Hillary is almost universally known and not surprisingly ahead in the national polls, but it's obvious that Bernie is gaining on her fairly rapidly now as he becomes more widely known. These polls will reflect more genuine voter choice as campaigning expands beyond the earliest primary states.

I sure wouldn't call a poll released just 10 days ago, "stale" though. Things aren't changing quite that fast at the national level right now. All of the polls in your list were released within a 1 week period, (which is a very narrow time slice indeed), and the newest of them is already 4 days old btw. But while the CNN poll reported the best raw numbers yet for Sen.Sanders, and noticeably better for him than recent polls from other major pollsters, it was actually in line with the trajectory predicted by the previous CNN polls taken over the prior 2 months. That upward slope is easily seen in the default graphs, and is almost as steep as Clinton's downward slope.

Anyway, here is the more important point: You can not legitmately draw trends directly between data from the CNN poll and data from the polls taken by other pollsters. The methodologies are different and the raw numbers produced are not directly comparable. I think it is important to understand this, (but I'm really not concerned). I was merely pointing out that no legitimate statistical trends can be drawn by comparing the polls in your list. Only 2 of those 8 polls were by the same pollster (Ipsos/Reuters) which is an internet poll, and the results delta reported just 1 week apart was only 7 with a MOE of+\-3. And, only 25% of the respondents in those polls identified as Democrats. This yields no significant change, even between those weekly Ipsos polls much less to the numerically overwhelming trends shown in the composite data (even if you give internet poll data equal weighting to the live polling done by CNN, ABC, CBS and the several other top-rated pollsters included in the dataset). The rest of the polls on your list simply are not statistically comparable either to the CNN poll or to each other, and are only useful in determining the longer term composite trends which can be readily seen in the charts.

So, the point of my post was to illustrate that the statistically legitimate trends which can be drawn by considering sufficient data composited over time, are clear and indisputable. And, that when qualifying the dataset further by culling the less reliable polls, those trends become even clearer in showing that they do not in any way whatsoever support your belief in "Hillary gaining ground".
Quite extraordinarily to the contrary in fact.

Still... anything can happen of course...
(And you of course, can still believe anything you wish to about the meaning of these polls.)

Peace

[DCBob]

17. Thanks for the thorough discussion of statistics..

I am a scientist and quite familiar with statistical significance. I have designed experiments, collected and analyzed data and published in peer reviewed journals. However, when it comes to discussing polls on message boards I take some liberty in using the term significance. Of course we cant make any actual scientific determination if those recent polls indeed show Hilary gaining ground but I do believe she is. When you see multiple polls showing the same trends then I think one can make that statement with some confidence.. at least on a message board.

[MoveIt]

11. But I hear the voters are hungry for gaffes and demand Biden!

Of course those are lies, borne out by polling.

[DCBob]

13. Of course Iowa and NH are not representative of most of America.

But it does appear he will do well in at least those two states.