If by odds you mean the likely outcome ranges then they are based on the combination of data points mentioned in that post such as presidential approval, the parties' standing in the Generic Congressional Ballot, state polling, candidate quality, fundraising, and the partisan leans of each district/state. You can add all of those things (and more) together to get a decent idea of the state of each race, which helps develop a plausible outcome range if the elections were held with current fundamentals.



But... it's not perfectly accurate. The people who understand elections recognized that McFadden was the likeliest nominee, and he led in polling too. He just didn't win a straw poll months earlier, and that's fine because those are practically meaningless. You seem to imply that because McFadden became the nominee, that does marginal damage to Franken's reelection prospects because it's true that McFadden is preferable to the GOP than the alternatives. However, because everyone predicted that McFadden would be the nominee anyway, him making it to this point doesn't impact the calculus in any way unless of course his underperformance boosted Franken even more.

I don't really know how many different ways the same factual evidence can be presented: no one expected Ortman to be the nominee. Everyone expected McFadden: forecasters, analysts, essentially everyone who was not affiliated with the Ortman campaign or who at the very least had some semblance of an understanding of all the different determining factors that were pointing to this outcome for months.

This post was edited by Pollster on Jun 2 2014 02:25pm

Ortman led in Suffolk polling in both February and April, plus she won the straw polls. I am implying that since McFadden doesn't have to devote some of his war chest to a primary first against a party-nominated competitor, that money can now be put to use combating Franken.

The Washington Post's "Elections Lab" has been criticized (rightfully) for producing some truly silly race predictions, but it has set out to start "fixing" the problems that plagued its earlier predictions. You can see their "new" predictions here, post-update: http://www.washingtonpost.com/wp-tran/politics/election-lab-2014?hpid=z1

That said, they still have many predictions that are utterly laughable (such as a 60% chance of a Republican victory in the Michigan Senate race, a 97% chance of Republican victory in Kentucky, and a 99% chance of Republican victory in Georgia despite the fact that the Democrats are currently leading all of those races and despite the fact that Michigan is a reliably-Democratic state in federal elections). But if anyone actually likes looking at these types of models, similar to the popular 538 model, then feel free to browse around.



...But that was always going to be the case. That, too, was widely expected months ago. There's no part of that that wasn't already previously understood. Ortman won meaningless straw polls (including one from 8 months ago) and that 200-person likely-voter sample (LOL), while McFadden led consistently in private polling. It turns out, to the surprise of no one, that all those people who understand elections were right about their prediction.

Didn't look like the case when the "anybody but McFadden" camp pushed Dahlberg up to 56-44 on the 8th ballot at 1 AM. 1 more late ballot probably would have resulted in a different nomination (albeit of an even easier target to beat). Saved by recess. Meh. At this point it doesn't matter, MN is going to contribute a piece of shit to the Senate regardless. Maybe we can get another letter to the IRS out of Senator Smalley. *joy*

Super Tuesday started off strong for the GOP "establishment." They were hoping that former Randolph Mayor and wealthy businessman Tom MacArthur would win the primary in NJ-03, the state's only swing district this year that was vacated by Rep. Runyan's retirement. His competition was carpetbagging former Bogota Mayor Steve Lonegan, who famously took on Senator Cory Booker last year. The Democrats hoped to be going up against Lonegan this fall but they will have to battle MacArthur instead, who is winning in a blowout.

Mike Rounds, the popular former Governor of South Dakota, easily won the GOP primary to become the party's candidate for U.S. Senate. He's hovering around 60% of the vote in a four-way race and he's cruising as predicted. He'll be challenged by Democratic nominee Rick Weiland, in what is the GOP's #1 pickup opportunity in the Senate.

It might be time to hit the pause button on the victory party for the establishment though. The night's marquee matchup, the Mississippi Senate GOP primary between incumbent Senator Thad Cochran and Tea Partier Chris McDonald, is neck-and-neck with 39% of the vote in. Cochran leads by less than a point, 49.7%-48.6%, and needs to get above 50% to avoid being pushed to a runoff. The Democrats didn't get their man in NJ-03 but they are salivating at the prospect of taking on McDonald in a race for an open seat.

Did you do all that yourself or did you find a source for those calculations?

It's somewhat of an original product that combines a lot of other sources. You can find similar models, like Larry Sabato's "Crystal Ball" or Nate Silver's "538" model, but each of them really has at least one major deficiency to them to where their models are not very accurate (at least this far away from Election Day) enough to make me rely on a custom model. Most of the models do hover around somewhat of a consensus though. For example, here's Sabato's current predictions compared to mine in parenthesis:

House: R+5 - R+8 (D+2 - R+7)
Senate: R+4 - R+8 (R+1 - R+7)
Governors: D+3 - D+1 (D+5 - R+1)

The only differences come from some disagreements on individual race ratings (for example he has the GOV races for HI/MA/CO at only "Leans Democratic," which is absurd) or some disagreements on how much one metric matters (he has KY's Senate race at "Likely Republican" because he thinks McConnell is likely to win it, instead of rating the race based on how competitive it actually is). Overall all of them are very similar.

So what you are telling me is you shifted some predictions to favor democrats.
Who saw that one coming?

Well, no, thanks for the laugh by embarrassing yourself and all, but I make predictions that best reflect what's actually occurring in the races. Anyone who can read at a 3rd grade level (which naturally would exclude you) could also note that my Governors ratings include possible outcomes, D+0 and R+1, that are more-favorable to Republicans than Sabato's ratings. The main difference between his ratings and mine is that my ranges are larger.

People like Larry Sabato and Charlie Cook are well-known media personalities who put great care into how their products are received by the public. They are interested in generating as much interest (and revenue) for their products as possible, so they make a conscious choice to offer predictions that come with a noticeable GOP lean. It's a shrewd practice. I am not a pundit and have no interest in building readership, so my ratings exist only to account for all likely outcomes.

To use Sabato's ratings: he has a net gain of R+4 at the low-end of his Senate scale, meaning that he sees an R+3 result (and everything that's more friendly to the Democrats) as an unlikely outcome. When you evaluate the map and the races it's easy to understand why his prediction is a poor one. Not only is R+3 possible but it's incredibly likely: the GOP could only pick up seats in SD/WV/MT, or those 3 + AR but lose KY. R+3 is a very likely outcome but Sabato's scale excludes it, and everything that's less GOP-friendly. It's poor electoral analysis.

You can see the same flaw in his House ratings also where he predicts a likely outcome of between a net gain of 5-8 seats for the GOP. He alleges a +5 net gain but there's nothing to back that up. The GOP only has two likely gains (UT-04 and NC-12) but they have two losses that are as guaranteed (CA-31 and NY-11). They have other pickup opportunies (NY-21, AZ-02, WV-03) but also their own seats that are as equally vulnerable (CO-06, NE-02, FL-02). There's absolutely nothing to suggest that the GOP will have at least a net gain of 5 seats in the House but that's the low end of Sabato's prediction. My predictions don't "favor Democrats," they're just more attuned to what's actually happening. The wider scales allow for possible outcomes that Sabato inexplicably ignores.

This post was edited by Pollster on Jun 3 2014 08:42pm

Thats exactly what you did whether you think you had good reason to or not. Don't lie and hide behind your usual "embarrassing yourself' crap.