If I may play the diplomat for a moment…I think there’s plenty of evidence that both forces are at work. The 538 time charts, for example, show a longer-term trend of Republicans coming home to roost, punctuated by (and partly driven by, or at least occurring during) ups and downs clearly related, at least in large part, to short-term news cycles.
In other words, for Trump, it’s been a series of “two steps forward, one step back.” The timing and direction of individual “steps” has a lot to do with news cycles, while the overall pattern of 2-1+2-1+2… reflects the longer-term trend toward a close election as usual.
That’s as much of an outlier as the LA Times poll. Looks like it’s Clinton +2, although she continues to lose a little ground in Nate Silver’s Nowcast.
First off, remember that the LA Times thing is a tracker - not great absolute number but reasonably good for trends. When the aggregate was about Clinton +3 it was Trump +2, like it is today.
And the last NBC/WSJ poll was beginning of August, RV only. At that point aggregated polls were running Clinton +6ish and it was RV Clinton +9. Now its RV result is Clinton +5.
So being Silverish here, correcting for house effects they are both reasonably read as correlating with an aggregated result of Clinton +3ish (+2 if you want).
Not sure about reading primarily as Republicans return (not to say it is not a factor). Picking the start of the analysis as the height of Clinton’s post convention/Trumps Gold Star attack bounce is a bit contrived and disingenuous. For the two weeks before the GOP convention the aggregate was running mostly Clinton +2 to 3. Trump got his bounce into positive and then Clinton got a larger and more lasting bounce that faded and was pushed down more by the week of 9/11 to the Friday 9/16. New news cycles and her back on the trail is seeing it fairly quickly return to the preconvention level.
Total Johnson/Stein is staying 10 to 11ish overall from last half of July to now. In the two-way aggregates total not saying Clinton or Trump has dropped maybe by 3ish. It may be that a few GOP-leaning undecideds have decided they would prefer a Trump presidency to a Clinton one but the polls also shifted from mostly RV polls to LV polls over that time so I am not so sure of even that.
Looking from before the conventions, which began some major news cycle volatility in each direction, there is no consistent trend.
I’m still betting that 2 weeks out from Clinton’s return to the trail (coincident with the Birther bit turning the page on the news cycle) will see the aggregates getting back closer to Clinton +5.
I understand that a daily addiction to new polls is not healthy, but I’m unable to resist:
National (Rasmussen): Trump +5 (538 gives it a C+ and a Republican lean)
National (LA Times): Trump +2 (continuing to go back towards HRC)
Illinois (Emerson): Clinton +6 (she should be doing better)
Wisconsin (Emerson): Clinton +6 (yay!)
Virginia (Roanoke): Clinton +7 (double yay!)
North Carolina (Siena): Tie (triple yay!)
Colorado (Franklin and Marshall): Clinton +9 (quadruple yay!)
All added to 538 this morning; overall polls only back up to 58.8 for Hillary.
Illinois, perhaps, but for this set of polls, VA, NC, CO, and WI seem in line with roughly a 2012-sized win for Hillary. But it’s just one set of polls – obviously not enough to conclude with any certainty.
I was thinking of Maine and Georgia, which have had quite a few polls showing a close race. Actually, one of Maine’s electoral votes looks pretty red, with the state as a whole looking awfully swingy. Jersey is close too, but that state likes to tease Republicans every year. Not sure what’s wrong with polling in Jersey, maybe voters there are just smartasses who lie to pollsters.
Sure, maybe, but I don’t think your eyeballing the trendline counts for much. Even if there was zero news, you’d get these little noise bumps in a poll trendline. And there’s always something going on in the news cycle. So attributing the noise to (some subset) of news stories strikes me as a kind of polling phrenology. Since there’s nothing to contradict the partisan return thesis, I don’t see why that isn’t a sufficient explanation for what we observe.
Again, it is only a “return” from Clinton’s post-conventions peak and to what it was in the two weeks that preceded the conventions.
I’ll resort to the trackers, the Reuters/Ipsos and the LA Times/USC. Again, absolute numbers with trackers are questionable but they do trends fairly well as they are consistent populations and consistent techniques from point to point within each tracker.
August 7th LV four-way Reuters/Ipsos had 24% not choosing either Clinton or Trump and most recent on the site September 11th had 21% not choosing one of the two. Go back farther to July 17 and the number was 23%. Pretty stable.
August 7th LA Times/USC had about 11% not choosing either and current is 12%. July 17 was 15%. Not quite whipsawing around much.
Are you hypothesizing that the GOP partisans were there before the conventions, came more to Trump immediately after the GOP convention, and then went away in the prolonged bump that followed the Democratic convention, only to gradually somewhat return since? If not then “the partisan return thesis” fails as sufficient explanation.
OTOH we have a clear association of the volatility seen with news cycles. The pre-conventions slide for Clinton with the emails release. The Trump bump up during the GOP convention. The prolonged Clinton rise in association with a very well run Democratic convention followed by Trump attacking the Khans. His gradual move back as he shut up for a bit and Clinton’s drop with her pneumonia and being off the circuit. And now her apparently (too early to say for sure) beginning to move back as the news cycle moved on again.
Given the high degree of correlation of the polling shifts with the news cycles and the lack of evidence that there has been significant partisan return I don’t see why the news cycles moving about a polarized stable race are not sufficient explanation for what we observe.
The first of those is wholly artificial. You’re fitting your narrative to the polls. I could tell a story of which news cycles were important that doesn’t match the polls, and it would equally be disproof of your hypothesis–the point being that post hoc stories about which news cycles moved the polls aren’t very persuasive because there’s ample stock of both plausible poll-moving moments and noise in the polls. It’s the equivalent of looking at moves in the stock market, and looking at what Barack Obama ate for breakfast, and concluding that when he eats pork the economy soars.
As for the second part, I’m not as steeped in the numbers, but my understanding is that Trump hovered in the high-seventies GOP support pre-conventions, dipped down to low-seventies during Hillary’s bump, and then has steadily climbed his way back to low-to-mid eighties. Do you agree with that basic premise or do I have the facts wrong?
Looking at Huffington Post’s aggregates, I can report the following data:
Trump’s GOP support was 80.9% on July 24th. He was -2.5% from Clinton in the 4-way.
Trump’s GOP support was 77.4% on August 16th. He was -5.3% from Clinton in the 4-way.
Trump’s GOP support is 80.4% today. He is -2.3% from Clinton in the 4-way.
In fact no it is not fitting the narrative. As the news cycles occurred the directions were anticipated with the unknown being how big and how long. GOP convention will cause a GOP positive move. The size was less and the duration less than normal but the shape was as predicted. The Democratic convention causes a bump up and most expected his Khan attack to hurt him. How how and how long was more than normal but the post Democratic convention bump would have been shocking to NOT see. And everyone expected some drop from Clinton’s illness. This is not post hoc.
Of course the convention bumps aren’t post hoc. But they also aren’t what we’re talking about. You think the conventions are driving the polls from August to September?
As for Clinton’s illness, there is no corresponding period in the polls.
Right. But the point is that when he has 80.9%, he is about where he is now in the polls.
Doesn’t look like it. Looks like the opposite to me.
The most recent poll of Maine was tied overall, 2nd district* Trump+11, 1st district Clinton +11 (since two and similar in size). That is significantly at odds with last time relative to national standing. So are relative standing in a number of states, more to Trump’s benefit than Clinton’s. He looks (it’s not true because 538 says so but they agree ) like he could win EV at 1% or more down in the national popular, which Romney wouldn’t have (if you move each state by the difference between his actual national margin of -3+ to -1 he still doesn’t pick up enough EV’s to win).
Maybe this effect is underrated in part because Trump and supporters past claims they’d compete in solid blues states without specifically saying that would be part of a titanic national blow out in their favor, as if they’d ‘pick off’ states like NY, even, in an otherwise competitive race, which was ludicrous. But Trump appears to be doing slightly better relatively in EV than popular in this period of a closer race, more subtly.
But for my state, NJ, I’d still say Trump has any real chance here only if you see the national polling average move to definite lead for him (or that’s the election result though polls don’t predict it, nobody knows that won’t happen). But in that case the significant further move in national would toward Trump be Clinton’s huge problem, not getting nosed out in genuine swing states.
*which is interesting because it tends to make a 269-269 tie implausible, not that it could ever be said outright likely so far from the election, but Romney states+OH+FL+NV+IA+NH=269 is fairly plausible, but adds up to 270 w/ ME 2 which Romney lost, and no other 269 combo is nearly as plausible. If Trump won ME statewide plus ME 2 but lost ME 1 and NH, and won all the other states mentioned, 268. Current RCP no toss ups map is all the above but neither NH nor ME statewide, 266.
I’m not sure the Maine polling is shocking compared to last election. Obama won CD1 by 21% but CD2 by only 8%. Sure that’s only a 13% difference compared to the current ~22% but given the difference in all the candidates’ relative popularity - not exactly surprising. It is interesting that this could be the first time Maine actually splits its electoral votes.
You’re looking at the numbers without actually analyzing what moves the numbers. In all honesty, we’re having a debate that neither you nor I (and DSeid) can truly substantiate. You’re looking at trend lines of republicans / conservatives jumping toward and away from Trump over time without actually analyzing what occurred between the two end points. What I do know is that when I’ve seen major breaking news, I’ve predicted what would happen before 538 changed its now cast and before some posters were convinced that the forecasts would change – and I was right. I suspect my hunches are correct in a lot of cases, but sometimes I get it wrong. It happens. My ego’s not tied to the results either way.
I don’t what the bar is for shocking: reasonable sounding explanations can always be given. But it’s a fairly striking relative swing from 2012 (though we’re speaking of one particular poll). The other aspect is that Trump is tied in ME in that poll overall when down 1.1% in the RCP national avg, whereas Romney lost ME by 15.1% points when he only lost the national popular by 3.9%, he did 12.3% points relatively worse than Trump appears in that poll. That’s not as big as Indiana’s massive move between 2008 and 2012 (from Obama won by 9.9 while winning nationally by 7.2 the first time, lost by 10.5 while winning nationally by 3.9 the second time), 17.1% point relative swing, but it’s pretty big if true.
NC appears moderately in the other direction among key states. Trump is up 1.8 in the RCP avg there while down 1.1 nationally, 2.9% points relatively better in the state than nationally, but Romney won NC by 2.2 while losing by 3.9 nationally, 6.1% points relatively better in the state than nationally.
But more EV’s seem to have become relatively more favorable for Trump v his national popular position compared to Romney than the other way around, at least for now.
