USA Mid-term Elections--Off to the Races

[onejayhawk]

UVa has a new article saying a lot of the same things. The interesting part of this one discusses the effect of a possible third independent. There is a fair (say 10%) chance that Greg Orman will be the tipping vote in the leadership votes.

http://www.centerforpolitics.org/cr...can-chances-of-senate-takeover-are-improving/

J

 

[Antilogic]

Trend as opposed to history. FDR is an exception to many rules. Since then parties changed in 1952, 1960, 1968, 1976, 1980 (early), 1992 (late), 2000, 2008. That's 5 of 7. Both exceptions are a single term off. 8, 8, 8, 4, 12, 8, 8 is a trend.

I give you "strong" may be too strong. Perhaps we are due for a close election.

J

It's not just FDR, he's actually in the #3 trend since his party only won the presidency for 5 straight elections.

But my big point is, though, the recent "trend" is a result of coincidental outcomes of a handful of close elections and not indicative of some underlying structural factor. Take 2000, for example--say Gore managed to eek out an extra couple thousand votes in Florida or New Hampshire out of over 100 million cast. Then your pattern has back-to-back 12s, and we might be discussing whether the new "trend" is back-to-back 3 terms. Take 1976, a shift of 46,000 votes (not much out of 80+ million cast) in Ohio and Wisconsin would have given Ford a term of his own, turning your pattern into 8, 8, 24!, 8, 8 assuming no other changes. Same with 1960: about 19,000 votes in Hawaii, Illinois, and Missouri would have blocked JFK, an extra 22,000 in New Jersey would give Nixon an outright victory. And this is all after we arbitrarily pick our starting point after the last big trend of 20 years. The campaign trope of the 8-year itch may be nice fodder for the new candidate running, but experience shows us the person seeking the third term isn't doomed.

In a fit of hilarity and questionable analogy-making, modeling the presidency as a series of coin flips would yield a similar pattern. We'd have some longer HHHHHH or TTTTT runs, some rapid flips like HTHT, some trends may appear like HHTTHH, but those short-term occurrences have no meaning beyond us reading patterns into a void of randomness.

UVa has a new article saying a lot of the same things. The interesting part of this one discusses the effect of a possible third independent. There is a fair (say 10%) chance that Greg Orman will be the tipping vote in the leadership votes.

http://www.centerforpolitics.org/cr...can-chances-of-senate-takeover-are-improving/

J

Yeah, I largely agree with his characterization so far. I'm surprised he didn't update his estimate for Senate pickups--I know he's hedging, but he could have done a 5-9 range given he updated 4 races in the GOP's favor and only 1 in the Dems'. And I still think Louisiana should be at least lean GOP.

 

[JollyRoger]

Perhaps we are due for a close election.

J

The 2008 Democratic primary was the last close election with the Presidency on the line, so I agree - the 2016 Democratic primary, with the Presidency on the line, could be close.

 

[onejayhawk]

Yeah, I largely agree with his characterization so far. I'm surprised he didn't update his estimate for Senate pickups--I know he's hedging, but he could have done a 5-9 range given he updated 4 races in the GOP's favor and only 1 in the Dems'. And I still think Louisiana should be at least lean GOP.

Point taken and I would add Alaska. Given the list below, 5-9 is solid, even if you take away Kansas.

Someone I know pointed out that Nate Silver slammed PPP in one of the articles above. Here's the quote:

It may be that some of the mediocre polls will converge toward the stronger polls in states like Iowa and Colorado; a Public Policy Polling survey of Iowa to be released later this week is also likely to show Ernst ahead, for instance.​

That is a reference to an earlier article: http://fivethirtyeight.com/features/are-bad-pollsters-copying-good-pollsters/

The 2008 Democratic primary was the last close election with the Presidency on the line, so I agree - the 2016 Democratic primary, with the Presidency on the line, could be close.

That would be one possibility. The dog show that both parties are pushing forward could make for some interesting moments.

J

 

[onejayhawk]

Another week with little change. Iowa and Alaska slip away from the Democrats, but North Carolina looks solid and Kentucky tightens. Kansas looks increasingly like a lost cause for the Republicans. NYT hgad the following:

Probability that...
Republicans hold every currently Republican seat 55%
At least one “noncompetitive” race is won by the underdog 38%
Greg Orman wins and caucuses with Democrats* 37%
The Senate is split 50-50 20%
Control is decided by Greg Orman 18%
Republicans control 54 seats or more 9%
Control is decided by Louisiana runoff in December 8%
Every race is won by the current favorite 7%
Democrats control 53 seats or more 3%
* Mr. Orman has said he will caucus with the majority party.

http://www.nytimes.com/newsgraphics/2014/senate-model/index.html

 

[Antilogic]

Iowa is still pretty competitive, it's been a 1-2 point race since the primaries ended (on RCP, the race was around +2 Braley just before the primaries, spiked for Ernst after she won, then back to +1 Braley and now currently +1.3 for Ernst).

Alaska's RCP average is finally all on LV models now instead of RV models, they were the last holdout because of infrequent polling. I don't think that race has changed too much but rather the polling has finally worked out that RV kink we discussed earlier.

The shakiness in Colorado has really thrown me over the last couple weeks. Udall has recovered a bit in last week's poll, we'll see what happens when Quinnipiac releases a new one--I am 99% sure that Gardner +8 was an outlier. Otherwise, I don't think anything has significantly changed in the environment since the last update.

 

[onejayhawk]

Iowa is still pretty competitive, it's been a 1-2 point race since the primaries ended (on RCP, the race was around +2 Braley just before the primaries, spiked for Ernst after she won, then back to +1 Braley and now currently +1.3 for Ernst).

Alaska's RCP average is finally all on LV models now instead of RV models, they were the last holdout because of infrequent polling. I don't think that race has changed too much but rather the polling has finally worked out that RV kink we discussed earlier.

The shakiness in Colorado has really thrown me over the last couple weeks. Udall has recovered a bit in last week's poll, we'll see what happens when Quinnipiac releases a new one--I am 99% sure that Gardner +8 was an outlier. Otherwise, I don't think anything has significantly changed in the environment since the last update.

I agree on Colorado. It is the only true tossup left. Anyone that makes a prediction is merely expressing an opinion. Polls are sparse and within margin. The head scratcher is the Quinnipiac poll from three weeks ago--respected pollster, sizable sample, largest margin by 2-1. We may have confirmation.

Two polls out since I started typing the last post contradict what I said in my last post. North Carolina is tighter. A CBS/NYT poll from late September looked like an outlier. The polls show Hagen about where she has been, but Tillis is pulling support away from third candidate Haugh.

Sean Hannity is calling Orman of Kansas a Democrat, which is laughable. I suppose calling him a RiNO is out, because he is not a Republican in name. Regardless, Kansas is one of maybe five states that Orman is not right of center. Note that the NYT gives Orman a 18% chance of having the controlling vote.

J

 

[Antilogic]

Before that likely outlier, though, the Colorado race was in Udall's favor by the same margin that McConnell and Perdue were putting up in KY and GA, sometimes greater in the case of GA. I don't think it was unreasonable to call that race a lean D and focus on states like Iowa as tossups.

I would say the "truest" tossups, in the sense that the polling margins are within the margin of error and turnout operations could dominate the outcome, are Iowa and Colorado (until we get an updated poll). They are followed by the cluster of Georgia, North Carolina, and Arkansas (all have around 3-4 point polling leads right now).

Kentucky may fall into that range if there is confirmation for that favorable Grimes poll, right now it is the only poll she is leading in out of several that show McConnell +4 on average. Kansas is also right on the 4-point lead margin for Orman.

 

[onejayhawk]

Before that likely outlier, though, the Colorado race was in Udall's favor by the same margin that McConnell and Perdue were putting up in KY and GA, sometimes greater in the case of GA. I don't think it was unreasonable to call that race a lean D and focus on states like Iowa as tossups.

I would say the "truest" tossups, in the sense that the polling margins are within the margin of error and turnout operations could dominate the outcome, are Iowa and Colorado (until we get an updated poll). They are followed by the cluster of Georgia, North Carolina, and Arkansas (all have around 3-4 point polling leads right now).

Kentucky may fall into that range if there is confirmation for that favorable Grimes poll, right now it is the only poll she is leading in out of several that show McConnell +4 on average. Kansas is also right on the 4-point lead margin for Orman.

I might agree about Colorado if not for the well documented media blitz. Once ad spending evened out, so did the polls. Kentucky and Georgia do look like the Democrat's only chances for flips, but RCP has 10 races under a 5% rolling average, including New Hampshire, with Michigan one more notch out. 538 puts only two races under a 2-1 lean, Iowa and Colorado. Kansas is right on the line, but there is no Democrat running.

One interesting poll is this one by the Washington Post--no one cares.

http://www.washingtonpost.com/blogs/the-fix/wp/2014/10/06/the-election-is-in-29-days-no-one-cares/

J

 

[onejayhawk]

The day after a nothing going on post, excrement flies.

UVa has a new map that, finally, includes the independent. Noteworthy is South Dakota, downgraded to leaning. This was long considered the safest of the Republican flips. Rounds, the Republican, has been embroiled in a scandals, or several small scandals. One of them led a former aid to suicide. It is now competitive, largely because of the third candidate holding the Democrat back, and vice versa. Much like Kansas, the Democrat could be the third choice.

The NYT is echoing in an article this morning.
http://www.nytimes.com/2014/10/10/u...ef=2014-midterm-elections&_r=0&abt=0002&abg=0

Even giving Alaska, as most do, this makes the Republican margin shaky.

J

 

[KMRblue1027]

If Orman somehow looses but the Dems win all the tossups than that's a 50-50 split, unlikely but interesting scenario with Biden as the tie breaker.

 

[onejayhawk]

If Orman somehow looses but the Dems win all the tossups than that's a 50-50 split, unlikely but interesting scenario with Biden as the tie breaker.

Having both those happen really is unlikely, mostly because Roberts needs a Republican shift to beat Orman. At this stage, I think about four states go Republican before Kansas.

If Orman does get the deciding vote, Angus King of Maine will likely go with him. While the specific scenario has not been proposed, King has often said he would caucus with the majority. Orman has said he will with other independents before deciding. One party or the other is almost assured a 52-48 majority.

J

 

[onejayhawk]

More on the impact of Kansas in the punditry.

Nate Silver's seat graph has an unusual shape. The compare Orman to a red dog Democrat and give a 75% chance he will caucus with the Democrats if possible. Here is the explanation: http://fivethirtyeight.com/datalab/upheaval-in-the-kansas-senate-race-is-making-our-chart-kinky/
Here is the graph: http://fivethirtyeight.com/interactives/senate-forecast/

With regard to the outcomes, this is very interesting. The two closest races, Iowa and Colorado, are also the most polarized.
Method: http://fivethirtyeight.com/datalab/the-senate-races-with-the-biggest-ideological-stakes/

Finally, Nate Silver on why the Democrats still have a 43% chance:
http://fivethirtyeight.com/datalab/the-maroon-6-a-better-way-to-look-at-the-2014-senate-races/

J

 

[Cutlass]

Having just spent the week in Maine, my impression is that the independent candidate for governor is running out of steam. And so it will effectively be a 2 party race after all.

 

[onejayhawk]

The election is now three weeks away. While the numbers have not changed much, the time factor is becoming significant.

NYT summary page is now including Princeton Election Consortium, often identified by the lead number cruncher, Dr Sam Wang. This led me to notice a tiff between Wang and Nate Silver. It occurred last week while I was traveling. Academic politics can be as much fun as Washington politics. Woodrow Wilson claimed he learned all he needed to know at faculty wives luncheons.

http://politicalwire.com/archives/2014/10/02/nate_silver_rebuts_sam_wang.html
http://politicalwire.com/archives/2014/10/06/sam_wang_factchecks_nate_silver.html

J

 

[Antilogic]

I've been putting off a longer post and more stuff keeps happening. The short:

I wasn't initially sold on the new rage of the South Dakota race, but there was just a second poll that showed it as a 4 point race, so it might be close enough to warrant all the money being dumped in.

Orman's race has tightened up a bit. Don't have much else to say, I don't think it was expected he would sail to a 10-point victory.

National committees are starting to pull funding on races they don't think they can win. It looks like the GOP is ditching Land in Michigan, and the Dems are dropping Grimes in Kentucky possibly due to polling plus the recent gaffe on whether she voted for Obama. I understand she did well in the debate v. McConnell so it looks like the timing was just off for her.

I'm still betting on GA as a sleeper race to watch--the RCP averages show a bunch of +4s for the GOP candidate awhile back, but the most recent polls are +3 R, +1 R, and tie, and apparently the local GOP is starting to freak out due to the voter registration drive.

The RCP averages (and most of the other sites) currently show narrow GOP leads, including Colorado and Iowa. It's obviously not decided yet but if that holds, they will definitely take the Senate and possibly with a comfortable independent-proof majority. For the Dems, it sucks to lose several races by 1-2 points instead of winning most and losing one in a blowout, but it looks like the former is playing out at the moment.

There is also a lot going on in the gubernatorial races, looks like the Dems have some solid pickup chances and endangered candidates like Quinn are recovering.



EDIT: The discussion between Wang and Silver was pretty interesting if you care about stats, but otherwise it's probably not worth a thread.

 

[onejayhawk]

I think Orman was a flash. He had no money of his own. Getting money from Democrats has hurt, but mostly it was Roberts getting off his duff and campaigning.

South Dakota is an embarrassment for the Republicans. They were willing to tolerate it, because they were going to get the seat. When another of their own made a run at it, they may have taken their eye off the ball. There is no runoff. If Pressler and Rounds split the Republican vote, Weiland could back in. It's a long shot, but stranger things have happened. I would not be surprised if Pressler won this.

I know someone in the head office of a MLB club. His opinion of Silver vis a vis baseball is low. Just sayin'. Personally, I am not impressed with Wang at this level.

I tend to trust the Iowa numbers more than Colorado. That said, the best poll of them all has the Iowa race at 1% +/- 3%.

The new poll reveals three potential reasons why this race has tightened in the final sprint:

• The Democrats' aggressive early voting push is aiding Braley, an eight-year congressman from Waterloo. They're rounding up ballots from Iowans who would not otherwise have voted.

• A majority of Iowa likely voters appreciate having a U.S. senator from each political party. Retiring Democratic U.S. Sen. Tom Harkin, 74, has been an old-school liberal street fighter for Iowa for 30 years. And Republican U.S. Sen. Chuck Grassley, 81, who intends to keep adding to his 33 years in the Senate, is a conservative powerhouse.

• Likely voters find more of Braley's policy positions closer to their own views than Ernst's positions among 10 issues tested. A majority of likely voters favor six of Braley's stances to four of Ernst's.​

http://www.desmoinesregister.com/st...owa-poll-ernst-braley-race-tightens/17114281/

Ideologically, these two races are huge. Ernst is a nightmare for the Democrats, opposing them on all the "women's issues", eg gun control, military, choice, education, etc. Braley is a rank and file Democrat. Nothing red about his dog. Colorado has a similar spread, without the sex appeal.

The LA Times did an article on projected turnout. With both sides treating this as a boring exercise, the Republicans have a distinct advantage.
http://www.latimes.com/opinion/op-ed/la-oe-mcmanus-who-votes-20141015-column.html

J

 

[Antilogic]

Nate Silver's record on political elections is quite good, and on the two Senate races he missed calls on, Sam Wang got them both right. The fundamental difference between their approaches is that Sam Wang only uses polling data whereas Nate Silver has polling data plus state fundamentals, which basically means PVI, candidate money, etc. Why you are unimpressed with real data grinding that gives these two men far better predictive records than the TV hacks is baffling to me.

I can't tell which of the anti-Rounds candidates are stronger in South Dakota. Weiland (D) has been solid in high 20s and rarely in the low 30s, but Pressler (I) has been at 12%, lagging far behind Weiland, and at 32%, ahead of Weiland's 28% in that poll.

This next week of polls will factor in the impact of all the debates being held now, that will probably be the last big shock to the numbers.

Also, I gotta say I think Sabato is making some weird calls. Cook has played it far safer with his organization's ratings.

 

[onejayhawk]

Nate Silver's record on political elections is quite good, and on the two Senate races he missed calls on, Sam Wang got them both right. The fundamental difference between their approaches is that Sam Wang only uses polling data whereas Nate Silver has polling data plus state fundamentals, which basically means PVI, candidate money, etc. Why you are unimpressed with real data grinding that gives these two men far better predictive records than the TV hacks is baffling to me.

I can't tell which of the anti-Rounds candidates are stronger in South Dakota. Weiland (D) has been solid in high 20s and rarely in the low 30s, but Pressler (I) has been at 12%, lagging far behind Weiland, and at 32%, ahead of Weiland's 28% in that poll.

This next week of polls will factor in the impact of all the debates being held now, that will probably be the last big shock to the numbers.

Also, I gotta say I think Sabato is making some weird calls. Cook has played it far safer with his organization's ratings.

Where does this come from? I said I liked Silver's approach (use all the information) more than Wang's (polls only). End of story. I do not know which TV hacks you refer to, but they were never part of this conversation. Chances are I do not know where they stand.

The lack of polling in SD is an issue. The poll where he was ahead of Weiland is the reference here. He was also within 2% of Weiland in September. There was a sharp upward trend til the most recent poll. As you say, Weiland's numbers have been steady, until that poll, which also has a 7% undecided.

I am not sure debates are having much shock value. There is still a large undecided block in some polls, but not all.

J

 

[Antilogic]

Where does this come from? I said I liked Silver's approach (use all the information) more than Wang's (polls only). End of story. I do not know which TV hacks you refer to, but they were never part of this conversation. Chances are I do not know where they stand.

The lack of polling in SD is an issue. The poll where he was ahead of Weiland is the reference here. He was also within 2% of Weiland in September. There was a sharp upward trend til the most recent poll. As you say, Weiland's numbers have been steady, until that poll, which also has a 7% undecided.

I am not sure debates are having much shock value. There is still a large undecided block in some polls, but not all.

J

I know someone in the head office of a MLB club. His opinion of Silver vis a vis baseball is low. Just sayin'. Personally, I am not impressed with Wang at this level.

From this, and general inference from discussion in this thread.



And I swear this is just a coincidence, but czech this out:

Link to video.

I put in the youtube because words cannot describe what I just saw. The Vermont governor's debate was also a joy, we'll have to find a highlights reel for it.