As is par for the course this time of year, after another busy election cycle, I’ll be pretty quiet on the website for a few weeks. I’ll likely be posting some data analysis about some of the 2014 races over the next few months, but will kick back into full 2016 mode some time in the spring. Until then, I hope you all have a good holiday season- here’s to the next election cycle!
Well, we end the year where we started. The 2014 Senate Map was favorable to the GOP from the outset. Even without converting a single Blue State from the 2012 election, the GOP could cruise to Senate control if they just converted all Dem Seats within the Red States. And that’s where we end. The GOP seems set to hit 48, likely 49. But, without pushing Toss-Ups, they’re stuck there: 48 or 49. If they sweep all the Toss-Ups tomorrow, they’d hit 54. If they sweep the Toss-Ups and the Leans Dem seats, they’d hit 55. Could 55 happen? Sure, but that would be quite the sweep for them. 53 seems to be a more likely peak. Meanwhile, can Dems still thread the needle and retain control of the Senate? At this point, their best case scenario is likely about 51 seats, and that would include a sweep of every Toss-Up other than Georgia, which has been fading for them. (With GA, they’d hit a miraculous 52.) This “Dem Sweep” option would also include KS-Senate, a state where we still can’t confirm whether Orman will ultimately caucus with Dems.
GOP at 52-53 would be the most likely bet, but 49-54 is all in play. Of course, GOP at 49 or 50 is quite a different effect than GOP at 51 or above. That’s why, though the GOP is certainly on the cusp of taking the Senate, there are enough close races that keep things interesting until the end. Dems should be nervous. Yet, the GOP shouldn’t be cocky. If the data in CO and IA were as off for Dems this year as they were in 2012 (and 2010 in the former as well), we may all wake up with Braley and Udall winning by a few tenths. Of course, that’s a big “If,” but that’s where it stands. Then there’s Alaska. Despite potentially controlling the future of the Senate, we have very little data there. Furthermore, the data that does exist is about as noisy as it gets. Either candidate could win by a decent margin. Or, we could be counting that race for weeks to come.
No matter what happens there tomorrow, the story of the year remains Kansas. While the GOP has run pretty efficiently this year, Kansas remains an exception. Even if Roberts pulls it out, this one shouldn’t have been in play. Even if Orman(I) wins, will he absolutely caucus with Dems? While it seems so if Dems are at 49 or above without him, meaning he’d be able to caucus with the majority, does he caucus with Dems if the GOP hits at least 51 tomorrow?
Then, there are Georgia and Louisiana. While Louisiana is likely to go GOP in a run-off, it’s likely to go to that run-off as no one is destined for 50% tomorrow. Thus, that GOP number will actually likely be between 48-53 tomorrow with Louisiana not solidifying for a few weeks. And what of Georgia? In the closing stretch, the odds of Perdue(R) hitting 50% suddenly seem very possible. Or, will Nunn(D) surprisingly over perform, based on Dems in that state finally becoming more engaged in a close election. Yet, even if she gains a few more points above the current forecast and beats Perdue tomorrow, her hitting 50% looks more remote.
And then there are North Carolina and New Hampshire. While I have North Carolina as a Toss-Up at the bell, it’s not meaningless that I’ve had Hagan up since summer. Though things have been on the decline for her, the hunch is that she hangs on. But a Toss-Up is a Toss-Up, meaning Tillis(R) could certainly take it and it wouldn’t be a surprise. Is New Hampshire really a Toss-Up? It certainly appears it has closed, but I’ve had Shaheen at or above 50% for the better part of the election. Even if Brown closes to 49% by swooping up all the remaining voters out there, still seems hard for him to top 49%. Surprising for Brown to win tomorrow? Yes. Shocking? Not quite.
And that’s our map. When we go to bed tomorrow night (on the East Coast), it seems likely we’ll have a run-off in Louisiana and Alaska may still be counting. Thus, we may yet to know who controls the Senate. Or, the GOP could blow by 51 without much trouble. Or, Dems somehow pull the rabbit out of the hat and sweep up all the races still in play. And thus, we wait just a little bit longer…
Now, some quick notes on the races in play:
This type of election data doesn’t happen in a presidential year. During a presidential year, all close states are pretty much polled to death. In an off-year election? You get a potentially close race with very little data. C’est la vie. Additionally, the data that does exist has about a 15% range in predicted results. Couple those two things together and anything can be expected tomorrow. Polls don’t close here until Wednesday Eastern time. Sullivan could win by 5% and this could be known a few hours into the wee am hours. Or, Begich could win by 5%. I have Begich up 1%, but with likely model lines that reach well into both candidate’s camps, you’d be ill-advised to read too much into the top line. So, sit back and watch Chris Nolan’s Insomnia as you wait for the results. It’ll put you in an Alaskan mood. One final interesting thing to ponder. If Louisiana goes to a run-off and other results are potentially not known or if Dems are over performing earlier in the night by holding NC and NH and picking up IA and CO, voters in Alaska are going to get about a million phone calls and knocks on their door in the final hours as voting will still be open in Alaska. The GOP may have already hit 51 before then, but if it looks like Dems still have a chance to control the Senate come about 9pm EST, those Alaskan ground games are really going to go into frantic mode.
What is going on here (Part 1)? It sure seems like Cory Gardner is ahead by a handful of points. That said, the model is ending at about Gardner +1%. That’s well with the MOE. While some will say just look at the polling averages, such is a lazy (and poor) form of data analysis. The data in CO (and IA) have some funky dynamics. While clean data will have a traditional bell curve indicating a likely final result somewhere around an average, the data in CO has two peaks of frequency. On one hand, there are many reasons to think Gardner is up about 3-5%. On the other hand, there’s a big spike of data showing the race as a Tie or within 1%. The result could go either way. I have the race a Toss-Up not only because I have it at about 1% but also because Udall is ahead in a few lines, though they are tiny. If Udall pulls it out, it’s likely to be very narrow. Additionally promising for Dems is that Udall really isn’t doing much better or worse than Bennett’s numbers in 2010 when he surprised most. Obama also over performed here in 2012. A strong ground game? English only polls missing Latino voters? The answer is a little bit of both. That said, that doesn’t mean that’ll happen again this cycle. Additionally, the early vote results are not devastating for Dems, but they do make it seem less likely Udall is going to over perform. Thus, either peak could end up happening, that being those data points at Gardner at say +4% or those showing it a virtual tie. The model is slightly favoring the notion that the race will end up being in that 1% category, but be prepared for a few options here. A final note? Gardner never really gets much over 48% in the polls. In the last two elections, GOP candidates have not over performed their apparent ceilings. Is that telling? Or just a mirage?
Georgia: What a difference two weeks make. In the middle of October, Nunn was surging. But, as the race closed, it’s shifted back toward a narrow Perdue edge. This is common as states often come home to their ideological bents in the end. Could Nunn still end up with more votes tomorrow? Certainly. Perdue is favored, but I still show some lines that could give her a narrow lead. That said, it’s looking very difficult for her to hit 50%, even if she pulls out more votes. That’d push things to a run-off. Meanwhile, some last minute polls show Perdue hitting 50% a real possibility. This race could thus be over tomorrow. It’s about 50/50 whether that happens. Early vote numbers, coupled with potential under sampling of Latino voters, could give Nunn a tid bit of cushion, so this race may still end up being quite interesting, even if it has seemingly been fading from Nunn as of late.
Iowa: What is going on here? (Part II). See Colorado. Again, there is no clean bell curve here. There are two peaks in the data. One showing Ernst up from 3-7%, the other showing it as a virtual tie. Which peak is correct? Like Colorado, the model thinks the peak showing a race within 1% is very likely. Could Ernst walk away winning by a few points? Sure. Could Braley pull it out? Yep. Like Udall, there are a few likely model lines showing a very narrow Dem victory as being well within the possibilities. Again, we have Dem GOTV efforts that sometimes surprise the data by a few points. That said, as polarizing as Ernst has been, it’s hard to bet against the candidate in a race that has the more passionate voters. Ernst elicits amazingly strong enthusiasm among her base. This type of effect happened in the 2013 VA Gov race. Even though Terry McAuliffe(D) ended up winning, it was by a much narrower margin then most data points showed. While Cuccinelli(R) was toxic to moderates and Democrats, McAuliffe, like Braley, failed to cause much passion with his base. Meanwhile, Cuccinelli had strongly passionate supporters. That seemed to get him a few extra points in a non-presidential cycle. Does a similar effect at least counter Dem GOTV strengths, resulting in Ernst performing as expected? Again, though the early voting data here is okay for Dems, it’s not great this cycle. I think the GOP enthusiasm for Ernst is represented by stronger early voting numbers. Thus, a few interesting dynamics remain at play. I’m not willing to call it.
Kansas: What to make of Kansas? This race should have never been. But when it was revealed that the long-time GOP Senator no longer even lived in his state, all hell broke lose. This race is so close in a state that is heavily GOP. It’d not be surprising if this state broke a few points to the right because of the state’s long standing bent. The other question is: what kind of ground game does Orman really have as an Independent. Then again, in the rare examples of when an Independent is actually viable, they can also have a history of outperforming. It’s really tough to say which dynamic will win out. As a result, with a model around a 1% margin, no one can really say.
North Carolina: Man, this one has closed as of late. While Hagan was riding a 3-5% margin for most of the last two months, GOP voters who have been very ‘meh’ on the deeply unpopular Thom Tillis have been coming home to their party. Hagan’s run about as strong as a campaign as could have been expected in a state where Romney won two years ago. Do I have a hunch that Hagan is still narrowly ahead? Yes. That kind of long but consistent narrow lead is historically very telling. Additionally, early vote totals have been pretty great for Dems. Yet, I have the state a Toss-Up because a few lines still show a Tillis victory as within an easy realm of possibility. Hagan could pull this off by a few points, or Tillis perhaps by a point or so. You could perhaps persuade me to at least bet a Coke Zero or other tasty treat on, but more than that and I wouldn’t wager. That said, if Hagan wins, she may be the only red state Democrat to survive if Udall, Begich, and Braley don’t pull it out.
New Hampshire: There’s not much to say here. Is it possible Brown wins? Sure. Do I really think it’s likely? No. As narrow as things have gotten, Shaheen’s simply been riding that 50% line all cycle. It’d take a mini-Brown wave to really flip this one. Possible, but it’d certainly be an upset.
Kentucky, Arkansas, Louisiana: Beyond the above states, the map gets small quickly. Arkansas has certainly faded. A Pryor victory doesn’t seem in the cards. Same goes with Kentucky. Even if the passion for Grimes leads to Dems over performing, I don’t see that big of an over performance happening. Then there’s Louisiana. The state is set for a run-off when anything can happen. Maybe we’ll revisit this one in a few days once the run-off kicks into gear, but for now, that’s all there is to say.
And there you have it! And with that, the 2014 Senate Models are a wrap!
I didn’t start modeling until the 2012 election. It was a Presidential year. It was a pretty strong sweep for Obama. Avid readers will know, all states that I predicted as “Leans” went for the candidate I had predicted. In two states that I had as “Toss-Ups” until the end, my top-line number showed one candidate ahead but the other candidate ended up winning. Those were ND-Sen and FL-Pres. Yet, in both states, there was a reason I had the races as a Toss-Up. In both cases, I had both the Dem and GOP candidates ahead, depending on the likely model line. Overall, I had Romney a few tenths ahead in FL. Obama won by a few tenths. Polling out of ND-Sen was remote, but I was happy to have that race well within the Toss-Up category, even though many pollsters showed the GOP candidate performing much better. The summary here is this: if a race is a Toss-Up, I really believe it could go either way. Thus, as I always say, the top-line that I report means ‘something,’ but also look at the range lines that I report come my final write-ups. Additionally, look at what number I have the candidate at. If say, I have both candidates at 46%, then they’re both really likely to at least hit that number. The question then becomes, what to make of those extra few points left out there. That’s where the play comes in, really making the races fluid.
With all that said, I wanted to apply my current formula in the Colorado Senate race to the 2010 race where Michael Bennett surprised many folks who relied on polls that mostly ranged from a tie to mid-single digit Buck leads. Alas, Bennett(D) won, 48.1% to 46.5%. I haven’t really received any flak for any of my numbers per se this cycle, but there is a sense out there that some think modelers are over estimating Udall. While many polls have Gardner up with numbers like +7%, my model, the NYT survey/model, and a few other races show the race much closer. So, this is not a clean comparison, but running the Colorado data from 2010 using my current formula at least seems to have some value. The result:
Applying my current formula to the 2010 data, I would have had the race a Toss-Up. It would come out as:
Bennett(D): Likely Model lines range from 45-48%. Overall: 46.6%
Buck(R): Likely Model lines range from 47-48%. Overall: 47.2%
The actual result: Buck 46.6% (-0.6%) from my Model. Bennett 48.1% (+1.5%) from my Model.
All in all, I am happy with those numbers. The model is working well within its range/possibilities with such a result. As a Toss-Up within 1%, it’s clear to expect either candidate could be the victor. I’m sure at the time, I would’ve had some skeptics saying, “I can’t believe you have Bennett only down 0.6% and the race as a Toss-Up. Buck will win by several points!” And yet, even my model would’ve been “Redder” than the actual result. Now, that is not AT ALL to say Udall is going to outperform my current model by about 2% overall. But it’s certainly possible. Gardner could also very easily win by 3-5%. But the point is, I have the race as a Toss-Up for a reason. So, from my perspective, we shouldn’t be counting our chickens yet in many of these races. The GOP could win by tiny margins in every single race. Dems could surprise in several races. Either prediction could be right (though not both). More likely, both are half right and half wrong. There are some factors that make Udall’s predicament worse than Bennett’s. Notably, while Bennett was constantly behind in most polls, the range of numbers was less varied, and few reputable pollsters had data points at +6-7%. Instead, most of those showed a 4-5% race. Yet, with the same amount of error, that’d equate to about a possible one-tenth of a point victory for Udall. At the same time, Udall has a few things in his favor if we limit this to statistics. Notably, while the range for Udall/Gardner is larger than Bennett/Buck, there are still as many, if not more, data points this election showing the race as a tie/1-point margin than the Buck race had. Thus, it’s not surprising that my model is, overall, showing about the same dynamics. That said, don’t discount the noise. A mid-single digit Gardner victory is very possible, but so too is a narrow Udall win. More broadly than Colorado, keep that in mind in these other states that really could go either way. We may very well wake up next Wednesday morning looking like (after some run-offs) the GOP may get to 53 or 54 seats after all. Dems could also hit 50 or 51. That math shows you should feel slightly better if you’re GOP-inclined, but both sides should be holding their breath for the next few days. Cockiness and sureness are asking for trouble.
Cycle after cycle, pollsters get a bad rap. Whichever side is losing claims, “The polls are wrong!” Then, whenever a poll comes out from a pollster showing their candidate ahead, they claim it’s a good poll or pollster after all. Yet, in the end, the polls, overall, are generally right. Sure, there are always bad polls and outliers, but that’s why we call them outliers.
Yet, each cycle, there are exceptions. Sure, when one internal Eric Cantor poll is so far off, it gives fodder to those who don’t understand how polling works. “I’ve never been called, so I’m not in that poll,” you’ll often here. Or, “They only called 600 people, how on earth can it speak for the voices of two-million voters!?” Well, the simple answer is, if the poll is done properly, it can, plus or minus a margin of error, pretty well speak to how voters are going to vote overall. That’s a sampling error. In other words, if the questions on a poll are properly worded, the public is properly randomly sampled, etc…by calling only 600 people, a poll is accurate 95% of the time, + or – a set percent, say 3.5%. So, if a poll shows Gardner(R) at 47% and Udall(D) at 44%, +/- 3%, with a 95% confidence level, Gardner is going to pull anywhere from 44-50% and Udall anywhere from 41-47%. So, 19 times out of 20, the final result is going to be within that range. Now, that sounds good and bad, doesn’t it? The poll is accurate, but Udall at 41% is very different than 47%. This is why I always warn you that races that are close really can go either way. The top-lines that I report mean something, but if Udall is at 46% in the model and Gardner is at 47%, be prepared for a small victory for either candidate.
Having said that, a margin of error is again relating to the sample size. So, if the poll has other underlying problems, look out! Now, what could those problems be? Poor question wording. (This happens quite often with polls asking about ballot initiatives where the wording is key.) More commonly, you get into debates about cell phones versus landlines versus the internet. You also get into questions of language. Overall, professional pollsters do a pretty good job at all of these things. Yet, small errors can certainly mean a 1-3% difference from reality, something quite meaningful in a year when we have so many close races.
Today was the first day of in-person early voting in North Carolina. I live in Durham County, the state’s sixth largest in population. It is also the state’s Bluest county as Obama won over 75% of the vote here. If Kay Hagan is to win statewide, she’s going to not only need large margins here but also high turnout. Durham and neighboring Orange County are generally the two Bluest counties in the state, but they are quite different. Durham County is anchored by the city of Durham, a majority minority city. Whites and African Americans each make up about 40% of the city while the Latino population has grown significantly in recent years, now likely making up over 15% of the city. Durham is anchored by a strong African American middle class. Called the City of Medicine, the County is home to Duke University and its portion of the Research Triangle. The ‘funkiest’ of the Triangle cities, it’s a mix of diverse neighborhoods, good food, and young professionals. Orange County is anchored by Chapel Hill, a pristine college town home to the University of North Carolina. Often considered the most liberal city in the South, Chapel Hill is highly educated, wealthy, and mostly white. As different as the counties are, both are crucial for Democrats statewide. In addition to winning decent margins out of the much more moderate and suburban Wake County (home of Raleigh), Democrats need large margins out of Charlotte and to win the smaller cities of Greensboro, Winston-Salem, Asheville and Wilmington to pull it out statewide. Democratic candidates that can do okay in rural parts of the state do extra well. This includes not only white rural counties but also Native American heavy counties in the Southeast (the Lumbee Tribe) and Western (Cherokee) parts of the state, new Latino growth in pockets all over the state, and rural African-American-heavy counties in the Central and Southern part of the state.
Curiosity killed the cat, so I decided to show up and vote as soon as polls opened at 9am. As much as it was nice to get voting out of the way, I also wanted to see what early voting turnout was like in Durham County. Though Durham is a huge county, only the Board of Elections downtown was open for early voting today. Three other locations open up for early voting in the County starting tomorrow, including one at North Carolina Central University, a Historically Black College in the city. Thus, voting locations are still rather limited in a city of this size.
I arrived a few minutes until nine and ended up being voter number 38. It took about 40 minutes to make it through the line and cast a ballot. Those in line were rather representative of the county at-large. A group of young professionals across from me all in line to vote before going to work. A retired gay male couple behind me. About 60% of the line was African American, a mix of ages. A couple men and women in suits. A priest. A minister. And an eclectic group of people everywhere in between. Upon exiting the voting booth, the line was still wrapped all the way down the hall where I started, indicating it was still about 40 people deep as a new batch of voters kept streaming in.
Almost every voter carried a blue pamphlet in their hand, provided to them by the Democrat volunteers outside. Other than these volunteers on behalf of the Democratic Party, only a local judicial candidate and her volunteers seemed to be present. While that may seem bad for the Tillis campaign, considering that the county is a Democratic stronghold means the lack of his campaign’s presence isn’t all that surprising.
Democrats in North Carolina commonly get out to vote early. In my neighborhood in Durham, we’ve actually received more contact from Democratic outlets than in 2012. Mailings. Phone calls. Volunteers at our door. We’ve had it all. I must say, it makes a difference as it stands out much more than the absolute barrage of campaign ads that have dominated our airwaves since last winter. If I didn’t examine communication and politics for a living, I would think the ads all blur together. Their effectiveness is perhaps long gone. (Though, I will say, the positive ads where a candidate is speaking to a camera are actually able to grab your attention among all the scary music and dark font ads that dominate the airwaves.)
If Kay Hagan is to win this election, she’ll need that ground game. Thom Tillis is highly unpopular, even more so than Hagan. Ever since the GOP took control of the state in the 2010 elections, they have enacted sweeping reforms. Though popular with the base, voters in a purple state rarely like one-party control. The result? Two years now of Moral Monday protests that started in Raleigh and have since traveled around the state. The NAACP, the gay-rights group Equality NC, unions, and many other activist groups have been attempting to keep their voters engaged in an off-year election. If Kay Hagan wins, it will be because of the local dynamics more than the national ones. It’s what’s causing problems for Democrats in Colorado as large chunks of that state have become fed up with one party Democratic control. Moderate Kansans have become tired of the GOP-dominated state government. The media keeps talking about national waves, but this year still doesn’t appear to be one. While the GOP still has good odds to win the Senate, potentially even by a few seats, they may do it entirely by winning Red States and may not even carry North Carolina, a state Romney won in 2012. That doesn’t happen in a wave year.
Though I’ve spoken about Durham and Orange Counties, the secret to Hagan’s success may actually be the 1-40 Corridor in general. Crossing the state all the way from Wilmington on the coast to the Tennessee border, it is home to important voting counties. Hagan needs high turnout and big margins in the Triangle, especially Durham and Chapel Hill. She also needs to turn out liberal Asheville in the mountains and pick up a decent margin in Raleigh and its suburbs. She then needs to hit about 50% in New Hanover County, a county just a percent or two redder than the state overall and home to the city of Wilmington. And don’t count that out. It’s the small things that matter in a close election. A small city like Wilmington, home to much of the state’s film and television industry, has lost a lot of jobs as of late after the state legislature, where Thom Tillis has ruled as the House leader, cut the state’s entertainment incentives. And off the productions, and their jobs, went to other states…
But then, Hagan also needs to pull a decent net out of Forsyth and Guilford Counties, home of Winston-Salem and Greensboro, respectively. These counties ever so slightly lean Blue with strong turnout (Obama won 53-57% here), and Hagan needs to pull at least in the upper-50s to win statewide. The good news for her? She’s from this part of the state. While the Triangle is filled with Northerners, Westerners, and other folks who don’t have that “authentic” North Carolina accent and palette for BBQ and Bojangles, Hagan is not from this part of the state. Instead, she’s from a moderate swing-part of the state, and that served her quite well in 2008 when she far out-performed Barack Obama statewide, even when he carried the state.
Tillis is from the Charlotte area, and that’s good and bad news for him. If he can keep Democratic numbers down there, it’s a plus. But he’s far removed from the 1-40 Corridor and has not run a great campaign here. Couple that with I-40 media outlets across these cities hammering the actions of the state legislature over the last few years, and it’s a tough area for him. (Tillis is the Speaker of the House, making him something of a state version of a Mitch McConnell or Eric Cantor/John Boehner type. No matter the party, being a leader associated with an unpopular governing body is never good. Though Harry Reid made it through and Mitch McConnell might do the same, it’s not an easy climb. Ask Eric Cantor. Ask Tom Daschle.) Furthermore, Tillis doesn’t excite the base as much as say a Joni Ernst-type could in the small NC counties like Yadkin (a rural county home to most of my in-laws), a county that voted as much for Romney as Durham did for Obama.
And so, North Carolina may be a canary in the coal mine come election night. If Hagan wins here be 3-5%, it’s unlikely to be a wave year. Michelle Nunn could then win in Georgia, yet Mark Udall could lose in Colorado. With those types of dynamics, the GOP is still likely to take the Senate, based on the landscape of this year’s races. If Hagan wins by 5% or more, look out. Maybe the Democrats will somehow hold on. If she loses? It’s going to be a very red election cycle. But then, maybe North Carolina is not a canary at all. Maybe it’s just one individual race in a pile of individual races where there is no canary this cycle. We’ll know very soon. In the meantime, I’ll keep you up to date on the early vote numbers here in North Carolina, birthplace of Pepsi, Bojangles, Krispy Kreme, and all things wonderful.
The media buzz out there is that Dems are doomed! The GOP is pulling away! Yet, that’s an exaggeration of what the data actually shows. Yes, the GOP is still favored to take over the Senate, but that’s far from a done deal. There has been some small movement toward the GOP in Colorado and Kansas as of late, but there’s also been positive movement recently for Democrats in Iowa, Kentucky, and Georgia. Now, until and unless Democrats actually seem to pull ahead in Kentucky or Georgia, the loss of Colorado would be pretty lethal. But so far, while Udall’s data still looks ugly, I’m far from considering that race anything but a Toss-Up.
And so, with less than two weeks to go, we’re still sitting at a really familiar picture. The GOP is likely at 48 Seats. Democrats are likely at 47 Seats, if you include the Independents that caucus with them in Maine and Vermont. That leaves five races that will likely determine the Senate: IA, CO, KS, KY, and GA. Then, on the outside looking in, there are some potential upsets that lean toward a candidate but are still in the realm of possibilities of potential surprises: AK, AR, LA, and NC. NH is also hanging out there, but that would frankly be a huge statistical surprise if Shaheen actually were to lose, no matter how close Scott Brown may or may not be. (All signs still point toward a mid-single digit victory for Shaheen.)
Thus, to maintain control, Dems need to hang onto their leads in North Carolina and New Hampshire and then pick up three of the remaining eight states in play. Republicans are favored in three of those (AK, AR, LA). Meanwhile, IA and CO are historically the best chances for them. IA may remain their best chance as Braley is holding his own as of late. CO is historically very poorly polled, especially as pollsters often underestimate Latino voters. But, the data for Udall in Colorado continues to slowly worsen. Then there’s GA and KY. Either would be a coup for Democrats, but they may need both if either CO or IA slips away and they can’t gain traction in any of the states they’re currently behind in (AK, LA, AR). And then there’s the Wild Card: KS. It’s hard to even talk about that race because no one ‘officially’ knows what Orman will do from a caucusing perspective. While there’s a good chance he’d go with Democrats, he may be very unlikely to do so if they’re not in control of the Senate. Of course, then his seat wouldn’t determine Senate control. So, if his seat were to truly matter, it would only matter if Dems were at 49 and the GOP was at 50. This could be a very real possibility if Dems pick up IA and one other state, be it CO, KY, GA, or another surprise. Then, Orman is the kingmaker, and I’m not in the business of predicting him from there.
The point is: Either party still has a good chance to control the Senate, though neither may control it until there are run-offs in Louisiana and Georgia. There is simply no wave this year, unless I suppose if the GOP picked off all these states including NH. The GOP is currently doing well in states where they should do well: red states. They’re also doing well in purple states like IA and CO, but neither leans their way as of now. And yet, they also haven’t completely put away red states like AR, AK, or LA were upsets are still possible. Meanwhile, while NH looks close, it’s a purple state that is still trending against them. So is NC, a state they should pick up in a real wave. In a wave year, a state like MN or MI should be more competitive. They’re not. Thus, there’s no wave this year, just more of the status quo reverting to standard dynamics. Thus, even without a wave, I could still see the GOP hitting 53 seats when all is said and done. They could even sneak to 54 if Tillis were to pull it out in NC. 55 is ‘possible’ but highly doubtful. Meanwhile, Dems still have a good chance to get to 49. If they hold NC, IA is basically 50/50 odds for them. I’m still watching CO as, though Udall’s in rough shape, Gardner is nowhere near 50% yet in the model. Then, 50 comes down to: KS? KY? GA? A Red State? Many possibilities remain.
So, stay tuned. A quick note on how to read the models with about ten days to go. If you’re a daily reader, you’ll likely see slow climbs in the last ten days in each race. The models will accelerate toward final predictions, pushing candidate model numbers up across the board. You’ve already seen this happen in some states. If a candidate is hitting a number in the model, they’re very likely to hit that number come election day. But, what that means is, in a state like Colorado, there’s a lot of room for error, late breaking undecideds, and state fundamentals to kick in that still somehow are not showing up in the model. While Udall(D) being stuck at 45% is bad news, Gardner(R) is still only at 46% in the model. So, where do those other 9% of voters go? The model is essentially predicting that they could break a couple of ways, if they’re still unassigned. Could Udall win by a few points? Yep. Could Gardner pull out a mid-single digit victory? Yep. While this may drive some readers crazy, I always say it’s a better illustration of reality then anyone that tries to convince you that a candidate will actually win by +1.4 to +1.6%. No data is that consistent in all races. The more data you have, the more you should look at the actual top-line numbers I report. The less data, be very careful. If neither candidate is hitting 50% come election day, those races are very much worth watching. So, sit back and either pull your hair out or enjoy the show, whichever your political-style may be.
So, let’s run through the states:
AK: As a Dem Incumbent in a red state, things don’t look great for Begich who remains stuck at about 43%. That said, Alaska data remains quite limited and on-the-ground efforts in this less conventional state may improve the room for error. Sullivan is in the driver’s seat but is also below 50% himself. Race still worth watching.
AR: Pryor finds himself in a similar situation as Begich. He’s stuck below 43% in the model with only a few days to go. Ouch. That said, Tom Cotton has hardly put this race away. I’d like some more data here, but again, it is GOP favored but much is left on the table.
CO: Udall’s numbers continue to fall. Yet, Gardner’s numbers have still been peaking right around the 46% mark. Are all of his voters already showing up in the polls? If so, the question remains: are the remaining voters going to stay home, or are they going to less than enthusiastically vote for Udall, a candidate who likely matches their ideology better than Gardner? This race is very much up in the air, despite some noisy poll numbers.
GA: Nunn-mentum? More and more, it looks like Nunn may very well win a plurality of votes come election day. But, the question is, can she get a majority since Georgia law requires a candidate to get 50% to avoid a run-off? That seems doubtful, but not impossible. Dems typically fall off hard in run-off elections because of low turnout, so Nunn has to do everything she can to hit 50% in less than two weeks time.
IA: While the bottom seemed to drop out for Braley a few weeks ago, I’ve never had him down by more than 1-2%. As of late, you’ve seen many pollsters come back to that line. This is essentially a 50/50 race. Like in CO, the question is whether the GOP candidate is close to their peak already in a purple state that leans slightly blue. This is common in off-year elections: the party out of power polls well all along as their base is fired up. The question then becomes: do Democrats show up? Do moderates stay home?
KS: This one may be a mystery ’til the end. Does Orman(I) have any kind of GOTV effort? He’s being pummeled by big money as GOP voters come home to Pat Roberts. Yet, never discount a fired up Democratic base in Kansas who may actually have a chance to win this year, or at least theoretically win if the candidate caucuses with them. Couple that with potentially high Independent turnout and there’s plenty of room for error in some of these poll numbers. Roberts has been shoring up his base more than Orman has been losing voters. All eyes may be on Kansas very soon…
KY: Don’t count out Grimes. There’s no doubt that recent data has shown positive trends for her. What’s against her is that she’s stuck at about 46% in a very red state. But, in part like NC, southern Democrats in Kentucky are not like Democrats elsewhere. Obama got clobbered in Kentucky in 2008 yet McConnell far underperformed John McCain’s numbers. That means many voters who had no appetite for Barack Obama were willing to split their ballots and vote for a Democrat for Senate. So, don’t assume this one is going to McConnell with data points like these. Is this 2008? No, but don’t assume state dynamics are the same as national ones.
LA: See AK and AR. GOP is favored in a heavily red state, but Landrieu is keeping it close. If that holds, keep an eye open for her as the Landrieu family machine is quite strong in this state. But, then there’s the run-off effect, just like GA. This race isn’t likely to be decided in November, and that doesn’t bode well for Landrieu in the long run.
NC: Hagan continues to be hanging in there. That said, while the Tillis numbers are woeful, Hagan being stuck at 46% remains rather scary for her. This has continually leaned her way, but a lot of voters are left out there. As I live in a Hagan friendly county, I can tell you her ground game has been really strong as of late attempting to get her base out. Like Grimes, Hagan is no Obama. She’s from the 1-40 Corridor part of the state that is not part of the liberal Research Triangle. This gives her a strong base in a part of the state that Democrats need to carry to win statewide, and she plays better with rural and moderate voters than the President does in the state. The question remains: will she turn out the base in this quickly changing state? Will she win big margins on the 1-40 corridor?
NH: I’d still be shocked about this one being anything but Shaheen. Has it closed? Yea, but a Brown win still seems to be an entirely different story…
Make no bones about it. Tom Cotton(R) remains the favorite in the Arkansas race. But, a Pryor victory is certainly within the realm of possibilities. Both candidates have bad news. Neither has much good news. Tom Cotton keeps falling from his peaks a few weeks ago. He’s now stuck below 46% in the model. Pryor’s (D) numbers look even worse as he’s stuck below 43%, not good for an incumbent in a GOP-leaning state, at least at the federal level. But, Pryor’s proven far more energetic than Cotton in the closing weeks. Couple Bill Clinton with Pryor’s domination in debating a rather bland Cotton and the Democrat still has a fighting chance.
Now, many fundamentals remain against Pryor. In many ways, the remaining group of voters should break against him, based on historic dynamics and current figures of who remains out there. But, Pryor is well-liked in the state, other than his affiliation with the President’s party. Additionally, he has plenty of money to close this race and appears far more comfortable on the campaign trail.
For now, Cotton still has the advantage. He’s closing by tying Pryor to Obama’s decisions around ISIS and even Ebola. Will that work? Maybe, but it’s not the strongest close with undecided voters. Meanwhile, Pryor keeps trying to sell himself as a moderate Arkansas guy. If this type of data hold, this race may take awhile to count on election night. In the last update, Cotton was up 4.0%. Now, it’s dropped to 3.0%. If that trend continues, this race will certainly be worth watching until the end, though it remains Cotton’s to lose.
Dan Sullivan(R) still looks like the slight favorite in the Alaska Senate race. No non-partisan poll has shown him trailing over the last month, and that’s normally pretty good news for a GOP candidate in such a GOP friendly state. Mark Begich(D) was always going to have an uphill climb here. The state is dominated by GOP voters and Begich(D) barely won in 2008, despite his opponent being investigated by the FBI at the time. Jump ahead six years and Begich is doing about as well as can be expected, but it still may not be enough in a state where the fundamentals have always been strongly against him.
Dan Sullivan has shown to be a competent candidate, and that’s perhaps all the GOP needed in Alaska this cycle. That said, things remain close and Begich is by no means out of it. In fact, I have Begich(D) has close as 2% in one model line. With the lack of data here and the unpredictability of the state, keep an eye out. After all, Alaska could end up deciding control of the Senate. If that’s the case, you have to remember that the state is still voting when all the other competitive Senate elections are already reporting their results. This effect has been one of the factors that has led to varied results compared to expectations as many voters may simply stay home if the election seems to be done. But, if Senate control is looking competitive at 10pm EST, expect some last minute ground game action here. That would probably generally favor Sullivan, but that’s tough to say. As neither candidate is near 50%, this one is likely to remain competitive ’til the end, but Sullivan remains a decent bet. So, for now, the state shifts back to “Leans GOP.”
About two weeks ago, media pundits cried, “Iowa is slipping away for Democrats.” Yet, I was never seeing movement for Ernst much above 1% overall. But, when a singular DMR poll showed Ernst suddenly up 6%, the sky seemed to be falling for Braley. Now, DMR is out with another poll showing both candidates up, but Braley quickly closing: Ernst 47-Braley 46. Meanwhile many pollsters show the race to be virtually tied. Others show Ernst up 5-9 points. So, what gives?
First, some of these numbers are from partisan pollsters. Second, while the margins of these polls greatly vary, the medians for the candidates across all of these polls generally do not. Even when Ernst is shown leading, her numbers are still generally stuck in the mid 40s, about 44%, to be exact. Sure, a few pollsters such as Quinnipiac have shown her hitting 50%, a number that should not be entirely ignored. Still, that number remains an outlier for her. While many polls have Ernst holding a decent margin, it is often not her strength but Braley’s weakness, a la the first DMR poll showing him way down at 38%. Yet, in that poll, Ernst was at her magical 44% mark. I’m over simplfying, but if you look at the medians, Braley comes in at about 43% overall. And thus, you have a pretty good illustration of this race, once you take out much of the noise. The candidates are quite close, but many voters are still quite unhappy with their choices.
While this is arguably not good news for either candidate, and while Ernst certainly has some occasionally very positive numbers, I still think it’s troublesome for her that I have her stuck at 44.2% in the overall model and not yet hitting 45% in any major model line. When it’s an off year election, your base is fired up, and the President of the opposing party remains unpopular, those numbers should already be showing up for her. Generally, the indication would be that many of the remaining voters (which still number over 10% in the model) are either a) not going to vote or b) not wanting to vote for either candidate but are more in line with Braley ideologically than Ernst. The question remains: will those voters not show up or will they hold their nose and vote for Braley by a decent net margin? At the moment, it seems like Braley may still be able to net a few votes out of that which remains. Though, I’ve been saying this for months. So far, it hasn’t happened. At the same time, neither has Ernst put it away.
Two obvious things can happen between now and election day. 1) This race breaks one way or the other. Or, 2) The model numbers for both tick up and this remains a ground game. In that case, it’s really tough to say who comes out ahead. Democrats in Iowa have an amazing ground game that turns out their base early through the use of absentee ballots. The GOP, in Western Iowa especially, are sure to come out to vote against the President and to vote FOR Ernst. Don’t discount either effort. But, again, the GOP has just as much of a problem on their hands as Democrats here, based on numbers like this.
The Current Model: Ernst(R) 44.2% – Braley(D) 43.9%
Do I believe Alison Lundergan Grimes has likely pulled narrowly ahead in the Kentucky Senate race, a la the recent SurveyUSA poll in the state? Doubtful. Does the model think the race has significantly tightened to the point where there’s a statistically decent chance that Grimes could pull out a victory? Yes.
While Mitch McConnell(R) had pulled into a high single digit lead in September, Grimes seems to have been clawing back as of late. An Ipsos poll had her down 4%, a margin many pollsters have had her at earlier in the cycle. Additionally, that now famous Survey USA poll certainly caused a lot of chatter this week. Keep in mind another thing: despite Jack Conway’s disastrous end to his campaign post-aqua Buddha commercial against Rand Paul in 2010, he still outperformed most polls, even in Kentucky, even in an off-year election. I say ‘even-if’ a lot because most folks would tell you that is not supposed to happen. But that’s why you shouldn’t listen to most folks.
The model still shows McConnell probably has a small lead. In none of the model lines is Grimes ahead, but the lines now range from a Tie to a 4.8% McConnell lead. Overall, the number comes in at McConnell +2.3%. With those dynamics, and neither candidate hitting 47% at the moment, I’m moving the model back to a Toss-Up today. I would love to see more data, but Grimes certainly appears to have a fighting chance. How has she done it? I think Bill Clinton and guns are a good place to start. McConnell’s horrendous unpopularity and being tied to an even more unpopular governing body in Congress never helps either.
Grimes has run one of the best campaigns in the country. The dynamics of the state remain against her. Unlike other Democrats, she has not focused on the Koch Brother attack strategy like Braley has an Iowa, a tactic that is a waste of energy. Voters are not deciding based on the Koch Brothers, despite whatever research convinced Harry Reid that it was a good 2014 strategy. If Grimes is somehow able to win, it will be because she has convinced enough Kentucky voters that, though they may not like Barack Obama, he’s only going to be President for two more years. Instead, she could be their Senator for the future instead of the McConnell past. It’s a strong strategy that she has been using, but it’s still got an uphill climb.