Bank error in your Favor!

I made a data entry error when I ran my national model using WAPO state probabilities. I dropped a decimal point and tabulated West VA as a near certainty for Democrats, when in fact WAPO (and pretty much everybody else on planet Earth) makes it a near certain pickup for Republicans. So, according to the WAPO tabulation, Colorado is "a speedbump too far."

Running my national model (zero local variance)with the correct WAPO probabilities(checked, double checked and checked again)gives Republicans a 74% chance of controlling the senate. That's in much better agreement with the WAPO assessment of 87%, but I can't improve the fit by adding local variance. WAPO isn't very forthcoming with the details of their prediction model, so I can't account for assumptions that would explain this unresolvable lack of fit between our models. The fact that just one state can make such a big difference shows how knife edge the outcome seems to be.

As I noted earlier, most other prognosticators show more states in play than WAPO does, but that doesn't mean their viewpoint isn't credible, especially this early in the season.

As I said in comment 7, Republican pick ups in the range of 4-8 seem most likely to me, given the limited data available. The WAPO numbers give results within in that range, their isn't a lot of scope given so many state probabilities near one and zero.

Exactly..

Considering all the catastrophes and scandals plaguing the Obama Administration now, there is absolutely no reason to think that an October Surprise will favor Dems..

Alaska - Begich is doing a lot better than expected, although the state is notorious for being very hard to poll accurately.

Considering all the bone head moves made by Federal Agencies in Alaska, the prevailing mood is "Throw The Bums Out!!" Begich is not going to win..

Arkansas - Pryor also doing very well in the polls. The GOP considered this a walk-off, but it is not turning out to be the case.

Another state feeling the heat from Obama's Southern Border fiasco...

Colorado - easy shot for Dem.

I'll give you that one. Not as easy as, say, Montana for the GOP.. But still...

Georgia - Sam Nunn's daughter could indeed steal this state from GOP. Now that GOP primary is over, let's see what the polling looks like in a few weeks.

GA is the GOP's version of CO for the Democrats..

North Carolina - Another one Dems could easily win.

"Could"... But unlikely.. Another state reeling from the Obama Admin's bonehead moves..

Of the 8, Dems have ONE semi-lock (CO) and one true toss-up (IA)..

The Dems themselves have already conceded that they will likely lose the Senate...

Turnout's the key..

With the Right spoiling for a fight and the Left completely disillusioned by Obama's incompetence, scandals and lack of transparency, it's entirely possible, even probable, that the Right will sweep...

Just keep in mind how you felt about Romney's chances, that's all I ask this year. Sometimes surprises happen...

Trooo... I could be wrong again this year..

But, so could ya'all.. :D

Conventional wisdom would indicate the latter.. :D

Michale

Of the 8, I could see as many as six or seven going Dem. Remember, it's only July and who knows what October will bring?

Alaska - Begich is doing a lot better than expected, although the state is notorious for being very hard to poll accurately.

Arkansas - Pryor also doing very well in the polls. The GOP considered this a walk-off, but it is not turning out to be the case.

Colorado - easy shot for Dem.

Georgia - Sam Nunn's daughter could indeed steal this state from GOP. Now that GOP primary is over, let's see what the polling looks like in a few weeks.

Iowa - Braley is fighting to recover from one bad anti-farmer gaffe. But before the gaffe, the state was considered in the bag for Dems, so we'll see...

Kentucky - OK, I think Grimes is a longshot, admittedly. But McConnell's approval rating in KY is so incredibly low, she just might pull it off.

Louisiana - You're right, this one is probably a loser for Dems.

North Carolina - Another one Dems could easily win.

We see it from different angles, but there are a whole lot of different outcomes I could see from these 8. Granted, I think Dems have already lost in MT, SD, and WV, so I'm already seeing them currently at only 52 to GOP 48.

Just keep in mind how you felt about Romney's chances, that's all I ask this year. Sometimes surprises happen...

:-)

-CW

All the others are practically shoe-ins for a GOP win..

I mean, Landrieu in Louisiana??? Com' on!!!

Begich in AK???

KY?? GA?? "toss ups"!???

On what planet???

Michale

I like realclearpolitics too, but it only gives poll results and margins of error. RCP poll only maps course grained qualitative odds: likely, leans, tossup.

I wish more pollsters would give numerical odds, and be transparent about how they compute them.

We'll know for sure in 101 days 18 hours and 36 minutes... :D

Michale

Here's a site to keep track (it's a slightly-leaning-right, but still-fairly-reality-based site, so you shouldn't have any "librul media" complaints):

http://www.realclearpolitics.com/epolls/2014/senate/2014_elections_senate_map.html

Right now, the split is 46/46 with 8 races up for grabs. This gives a split of a possible 54/46 split, either way. That's about the outside of all possibilities, at this point.

You're welcome.

If that's not enough, check in the left-hand column's links for "Senate: no toss-ups" to see how the field shifts as new polls come in.

-CW

"... am I missing something???"

Yes.

First, the forecast of a net 7 gain is just the highest probability net gain or loss, which isn't all that interesting by itself. What you really want to know is the probability of winning control of the senate, which you get by summing up probabilities of the different winning outcomes (Dems treat a tie as a win, Reps don't). You left out the take home until comment 21.

Second, you left out who made the prediction, until comment 21, which is important. WAPO is a bit of an outlier. Their assigned probabilities give the fewest competitive states I've seen.

CW admirably covered some other stuff for me (hat tip to CW.

As I pointed out in comments to Boehner's Laughable Lawsuit a few days ago, assumptions about the independence of individual state behaviors make a big difference in how you predict overall likelihood of victory from state by state odds. In 2012, I found the most successful forecasts were consistent with the assumption that states move in tight herds, with nearly all the model variance expressed in one national roll of the dice.

If I apply the latest WAPO state probability table to a national variance model, I get a very different prediction:

The Dems have a 54% chance of retaining control, 26% by outright majority, and 28% by a tie. That's down just a little bit from what happened when I ran WAPO's numbers earlier in the week.

With a strong state herd effect, Republicans have a 95% chance of winning 19 races, one of which (Montana) is a takeover. They have a 74% chance of winning all the above and 4 more takeovers (Arkansas, Louisiana, Iowa and Alaska. Then they hit Colorado, the only genuinely competitive race according to WAPO tables and the key getting a two seat majority. The odds of that are 46% The odds of adding another seat and getting a 4 seat majority are just 9%. My strong herd model is in pretty good agreement, with the 538 and NYT predictions, but since both of those probably use a very similar model, treat them as just one case.

There is no guarantee the strong herd effect I saw in 2012 will apply in this year, but I certainly don't discount the possibility.

In general, I'm skeptical about which states are genuinely competitive at this early date. Polling data is fairly sparse, and a lot of it is probably unreliable.

Be interesting to see how ya'all react when DEMOCRATS are on the receiving end of a total prick Majority Leader... :D

Michale

It STILL fires Harry -The-Iraq-War-Is-Lost- Reid and gives uS Majority Leader McConnell does it not???

Michale

What would be the numbers if every single Senate race went the GOP's way???

Democrat's way???

Michale

52 is more than 50, if my math is correct..

It STILL fires Harry -The-Iraq-War-Is-Lost- Reid and gives up Majority Leader McConnell does it not???

Michale

What may have thrown you off is that there are two "I" senators, but these Independents reliably vote with the Dems. So you may see (technically accurate) stories which state "Democrats currently have 53 senators"... which would mean 47 GOPs, by subtraction (and make your math correct, incidentally)... these stories almost always have a "and two independents who caucus with the Dems" qualifying statement.

-CW

Um, no.

Not to quibble with the projections (which is, indeed, what I wanted to write for this column, in a way), if the GOP is +7 after November, it'll be at 52. They're at 45 right now.

Just to keep you honest... (heh)...

-CW

Troo... The Administration could actually do something that ALL Americans would be proud of...

Stranger things have happened.. :D

Michale

Michale

Well, that's simply not true!

The chances that I won't be reading comments and responding here after early November are relatively excellent.

(note: comment #9 needs to be deleted/ignored; that shouldn't be a problem)

WOOT!!!!! :D

Michale

Well, that's simply not true!

The chances that I'll be reading comments and responding here after early November are relatively excellent.

Uh... no...

If GOP gains a net +7 that means they will have 54 seats which is a clear majority and gives the GOP effective control of the Senate..

Or am I missing something???

Michale

That's consistent with the NYT estimate, which I consider credible, but that basically means the race for senate control is pretty close to a tossup. All remotely likely forecasts boil down to just 9 or so competitive, or slightly competitive, states. You can almost ignore the rest.

At this early date, net Republican gains in the range of 4 to 8 seem to cover the fat part of the frequency distribution. The chances of anything else seem pretty remote.

Yep... He's toast...

Michale

"REALLY!!??? That's great!!!! And... You can PROVE that, right!?? Oh yea, that's right. I forgot. You were absent the day they taught Law at Law School."
-Tom Cruise, A FEW GOOD MEN

:D

Michale

{{cough}} {cough} bullshit {{cough}}

:D

Forecast models show the GOP with a net gain of 7 seats in the Senate, thereby insuring an effective majority.. :D

Liz,

That any American politician would make a public statement such as that betrays an extremely sad state of affairs and does not bode well for the future of American leadership, at home or abroad.

Something we can totally and utterly agree on...

Michale

That any American politician would make a public statement such as that betrays an extremely sad state of affairs and does not bode well for the future of American leadership, at home or abroad.

"That is not my job. It is the primary responsibility of the state Commerce Cabinet" - Mitch McConnell

In the old days, MM would blanket the airwaves with ads about the pork he brought home. That was more effective than saying "not my job". Now he says that he had nothing to do with funding that dam. Somebody else did it. Ditch Mitch.