There’s a whole tonne of differences between now and then. Your hardee har har level of confidence is genuinely an “lol” from me.
You listed several of the items that have changed specific to Texas that should result in a shift of a few points. Do you also factor in the current overall shift in the national mood as seen in the various special elections? Of note in many of these races GOP PACs have poured in much more money than has been available to the Democratic contender and the shift has averaged about D+17.
Cruz won handily in 2012 by 16 points. You are taking the factors specific to Texas that favor less poor D performance, the factor specific to this race (O’Rourke at least comes off as authentic and as likable, especially in contrast to Cruz), and predict a shift of 8 from 2012. But maybe it will shift more like the average special election has shifted? If so it is a competitive race.
AZ-8 next week may inform some. It is ruby red and Trump won it by 21 points. Would that district shifting by 16 or more (still a 5 point GOP win) alter your prediction of the final spread in Cruz v O’Rourke?
Let’s be clear: No question O’Rourke is an underdog and the race is for Cruz to lose. Texas is not (yet) Blue and won’t be even in the event of a Cruz loss. But in the current environment declaring that O’Rourke has no chance is premature at least.
This has not been my impression for most of the races. I had thought that Dems have heavily outspent Repubs for most of the races. Jones in Alabama, a few special elections elsewhere. I haven’t gone through race by race, but I will after this post.
As I wrote in the Election contest thread, something is causing a lot of GOP incumbents to look fondly upon retirement, and these are not jobs that you usually want to retire from. Upright, anyway. It may be that there is a wave of dissatisfaction sweeping the country as you propose. I think that if there is, it’s a creation of the chattering classes’ profound extant butthurt over Trump winning in 2016, and not a reflection of what is a booming economy and dramatically decreasing unemployment, but reasonable people may differ. I wouldn’t think that Congress critters would retire over a press-created phenomena, but something is causing people like, e.g, Paul Ryan, to do so.
If it exists, it may be that wave that is causing an incumbent Senator to only have a 3.5 pt lead over his main challenger in an initial poll. I would have expected a lead much more in line with the prior initial polls I cited, something in the 10-15 pt range. All I’m saying is that, for at least as far back as I bothered to look, the GOP candidate margin of victory gets undersampled by about 2-2.5x, so if he’s still leading in these initial polls, it’s likely he’ll still win in November.
But perhaps the factors you cite now will prove to be the difference should Beto win? I don’t see it, but we don’t have a lot of data right now.
It has to be either butthurt, or a false impression of the economy? It can’t be anything else? Anything?
Here’s some more data for why I think the Quinnipiac poll may be misleading. We actually have voting data from Texas for this Senate race: the primary vote totals. Now the Democrat primary was a meaningfully contested one; O’Rourke got about 61% of the vote. It’s reasonable to assume—mainly because people I know didn’t bother going to vote in it—that Republicans felt less urgency to vote in this primary, as it was going to be a coronation whether we voted or not. At the very least, I think you can agree that Democrats had more reason to turn out for this primary than Republican voters.
If you guys have already covered this point, well I’m sorry for wasting your time. All that said, the total number of votes cast in the Democrat primary was 1,036,467. That’s all three candidates, all of whom received over 10%. Cruz by himself, again in a primary where I suspect a lot of Republicans didn’t bother showing up, received 1,315,146 votes.
Comparing those totals, and ignoring the ~135,000 or so non-Cruz Republican votes, Cruz would have received 55.9% of the sum, and the Democrat total 44.1%. Or not quite a 12 point win. Give all of the non-Cruz Republican votes to the Dems—call them never-Cruzers if you want—and Cruz still wins that hypothetical contest 51-49.
Given the hard primary vote data that we have, I find the prediction poll with Cruz only leading by 3.5, baffling.
I think what’s tripping you up is that the Democratic candidates themselves have been outraising the GOP ones, but that has been dwarfed by GOP PAC spending in most races.
Current race in AZ - Tipirneni-D has outspent Lesko-R $735K to 564K but outside GOP PACs have spent at total of $1.1 million while Democratic ones a paltry $72K. Lesko will likely win, but a win by less than double digits in that district will … inform. Maybe Lesko will still get that double digit win, but it’s looking less and less likely. Recent polling at least predicts that it might be a late night, and that is simply crazy. Yeah, baffling.
In PA 18 was the same pattern supersized: outside spending of $10.7 million GOP to $1.8 million D.
Alabamawas more the exception case in which many of the big GOP PACs sat it out. Outside PACs spent $4.8 million on the D side to $3.2 million on the R side.
Virginia saw some races with big spending on each side pretty matched but I cannot find exact figures.
The Montana at-large election set spending records with the outside GOP groups spending $5.6 million to the under $1 million by outside D ones. The shift was to +D16 (which came up short).
So there is a shift in terms of people actually coming out to the polls of on average about +D 17 points, despite some massive outside GOP PAC spending. You can read more about it on 538 if you want. For the second, let’s ignore the why of this enthusiasm gap and focus on how that may impact Cruz v O’Rourke. Of course candidates matter. Cruz is, to quote you, creepy. The demographic creep favors O’Rourke. The candidates matter bit does not seem to push this to Cruz.
Those primary numbers? Also appreciate that the GOP numbers were pretty typical ones for midterms (slightly up) while the Democratic side was twice what it was in 2012. Given the baseline that is still advantage R but it speaks to a lesser advantage R than in previous cycles. The gap narrowed.
It is easy to argue that Texas aint Blue - it aint. O’Rourke is without doubt an underdog. Cruz +8 could happen. But I’d bet on a much closer race.
Oh. Didn’t include the Georgia data point, the most expensive House election in history. Individual donations advantage D $8 to 2.1 million. PAC money advantage R $18.2 to 7.6 million. The swing a relatively paltry D +6 which was not enough to win.
The problem is, and I say this whenever Dems get excited by it, is that there’s no reason to think primary numbers are predictive of the actual race. I mean you can make a reasonble argument that they could be but I haven’t seen any number crunchers validate that line of argument.
What’s confusing to me is that you agree with most people here that Beto is a long shot and agree with a 5-10% chance but also think it’s going to be Cruz +8. If I thought the final margin was going to be +8 then I wouldn’t be giving Beto anywhere near 10% chance of a win. But maybe that’s my number instincts being wrong.
How closely do the results for off-year elections match the polls? One issue is that turnout is generally lower. I would expect the ones who turn up are the more motivated voters, and that tends to be more conservative people. So I can totally believe that polls of Texans in general would have Beto close, but I don’t think that voter turnout will be representative of the general population. I would expect a higher percentage of Cruz supporters will actually vote in the election than Beto supporters.
The first question has an actual factual answer and it is “not so great.”
Yes, lower turnout and, as the election gets closer, getting “likely voter” right, confounds good predictions based on polling alone. But also a factual answer - it is not consistently a GOP/conservative leaning error.
That’s why one really needs to factor in special elections as means to gauge partisan voter enthusiasm and why there is reason to believe that the mean (or maybe the median) shift seen by the aggregated special election results informs more assuming that the individual race does not indicate a complete blow-out.
Back in December, per 538, “the average Democrat has outperformed the baseline by 16 percentage points. (The median is 16 points, too, so it’s not just one outlier moving the average.)” Since then that’s moved to more like 17.
But while polls of midterms, even close to election date, have not done so hot, special elections in aggregate have done better.
Notably the error tends to favor GOP. So let’s spot the special elections shift of 17 by more than the average of back to R by 3, and put it smack between average and most ever so far, a move of 5 back to GOP on midterms day. That’s D+12.
If this Texas race hits that average then it is Cruz by 4. Could the other race specific items (shifting demographics, creepy Cruz, a strong campaigner in O’Rourke, etc.) push that over the edge? Well that is the op. There’s a lot of race to go to see.
ETA PredicitIt currently runs 2 to 1 Cruz odds, not 9 to 1.
It’s analogous to the difference between mean and variance. If you had a normal distribution that was centered on Cruz +8, but had a variance of 6.25, you’d get a 10% chance of a negative value for Cruz.
Is that in fact the case or a “for example” situation?
It is in fact the case with the minor nitpick that RTFirefly uses variance where he should use standard deviation.
Here’s an online normal distribution calculator for anyone who feels like verifying this on their own.
Put in mean 8, sd 6.25, and select below 0 to get 10.03%.
Ok, please excuse my ignorance, but where did the SD of 6.25 come from?
I misunderstood your question. I assume he picked 6.25 to get the result of 10%. That said it’s not an unreasonable estimate, but this far out I’d put it a little higher than 6.25.
Ok then it’s all just random numbers. I mean there’s no point saying “this far out” it’s a reasonable estimate because the +8 Cruz isn’t happening atm as far as I know. It’s just Gray Ghost’s guess at the final.
Thanks DSeid for your very informative replies to my financing inquiries.
As to the variance vs SD discussion, that’s a lot more involved than I meant it. I just SWAG’d Q’s poll margin by 2.5-ish and called it a day. Number crunching the primary vote data came later.
For the 2014 Gov race, Abbott ended up getting twice the votes he pulled in the primary, while Davis picked up 3x the total Dem votes cast in the primary. About the same happened in the 2014 Senate race. To assume the same would happen in 2018 would imply that Beto would get 3x ~1million or 3 million votes. Doubling Cruz’s total would give him about 2.6 million. There are 15.25 million registered voters in Texas. 3 million votes for O’Rourke would be an amazing total, and if that paradigm holds, then he would win.
If you think unemployment is dramatically decreasing now, you must been over the moon about Obama’s second term.
Republican control of Congress and 34 state governments was pretty great for the economy, yes.
Kansas might like to have a word with you. And probably Wisconsin. And Michigan …