Utah Will Have New Senator

Republican incumbent senator Bob Bennett was dropped from consideration on the second ballot at Utah's nominating convention, finishing a somewhat distant third place to candidates Mike Lee and Tim Bridgewater. Although this contingency had appeared to be quite likely based on pre-convention polls of delegates, it was obviously very disappointing for Bennett, the three-term incumbent, who was reported to be in tears afterward. Bennett pledged to support the eventual nominee.

Although a dramatic development in a state which rarely makes political headlines, my guess is that people are going to read a bit too much into the national implications of this. The 3,500 delegates who select Utah's Republican candidates -- chosen at local precinct meetings -- are highly informed and extremely conservative activists who are not representative of Utah Republicans as a whole nor the Republican primary electorates in other states. Some polling has suggested that Bennett would have been favored to win a conventional primary, although there were no guarantees.

The two prospective Democratic nominees, Sam Granato and Christopher Stout, are inexperienced. In such a ruby-red state, they are unlikely to pose much threat to Lee or Bridgewater in the general election.

There's More...

Contract Post

Hyundai's Very Undude Car Ads

Have you seen those new Hyundai Sonata ads? In one of them we are watching an obviously frustrated, presumably senile septuagenarian fumbling with a medicine bottle, when Oscar winning-actor Jeff Bridges’ unmistakably charcoal voice pipes in with this foreboding enthymeme: “Millions of elderly adults continue to drive on our roads, year after year, with nothing more than a periodic eye exam required to renew their licenses...better get yourself a safe car. The all-new Sonata, from Hyundai.”

If this ad doesn't sound familiar, that’s because there is no such ad. You can bet that Hyundai, or its Madison Avenue consultants, wouldn't dare run an ad like that. The political blowback would be so swift and foreceful the company's CEO would soon be on all the networks apologizing to seniors and probably announcing a huge corporate donation to some senior-related charity.

So why is Hyundai in fact running a pair of television commercials--including the visually entertaining "Slingshot" ad, below--that encourage consumers to buy their safe Sonatas because of the threat posed by newly-licensed young drivers? And why is there not a peep of complaint from anyone about this ad campaign?



Now, it’s true that teen drivers in America have the highest fatality accident rate per miles driven. In 2001-02, American drivers (in passenger vehicles) under age 20 were involved in about 8 fatality accidents per 100 million miles driven, as the figure below taken from this report by the Insurance Institute for Highway Safety shows. That rate begins to tail off for drivers by the time they reach 25, and then remains more or less constant at around 2 fatality accidents per 100 million miles until drivers reach 70--at which point the rates begin to rise again, spiking up dramatically for those over age 85.


Of course, when we talk about young drivers a major contributing factor is alcohol and drug use, which are involved in a much higher share of accidents for younger than older drivers. But we have all sorts of public campaigns--not to mention serious penalties--to discourage all drivers, and especially young drivers, from operating a vehicle while under the influence, not to mention related safety campaigns discouraging if not ticketing people for driving or texting while driving. There is no such thing as driving while really, really old. We expect seniors to police themselves or family members to intervene to take them off the road when their abilities or reflexes decline. We are not so casual when it comes to young folks--and that's fine.

My point here is not to justify teen driving behavior. It's a serious problem, but one that receives ample attention, publicity and action by state legislatures. But few dare mention the fatality accidents on the other end of the age spectrum because seniors are politically powerful in ways teenagers are not. Worse, Hyundai apparently thinks it's clever or cool to sell cars by making people fear young drivers. They would never do this if political clout were similar for younger drivers and older drivers, who happen to be--albeit for different reasons, of course--our most dangerous drivers.

Well Hyundai, you ought to be embarrassed. You should take down these ads and issue an apology. And damn, Jeff Bridges: As one of my favorite actors, I wish you hadn’t lent your legendary voice to these commercials. Very undude.

There's More...

Contract Post

Telephone Polls Closer to Mark in UK

Here are the results for the eleven organizations that released polls of Great Britain in advance of yesterday's general election, along with their sampling technique.



An error score is calculated simply by adding the difference between the projected and actual results for each of the three major parties. (It's OK to do this in the UK because all pollsters allocate out undecideds rather than leave them be; in the United States, we'd tend to want to look more at margins.)

Nobody nailed this election, because everyone had the Liberal Democrats with at least 26 percent of the vote when they in fact finished with just 23.6 percent. That may not be the pollsters' fault if voters changed or made up their mind while casting their ballots, as sometimes happens for third parties whose viability is questionable.

The closest was an Indian-based firm called RNB Research which had results of 37/28/26, but in general the telephone pollsters were closely bunched together. All but one of the pollsters had the LibDems trailing Labour, although always by small margins.

The performance of the Internet-based pollsters, on the other hand, was decidedly mixed. The more highly-regarded firms, Harris, Opinium, and YouGov, did OK, although Opinium and YouGov were marginally worse than any of the telephone pollsters. Angus Reid, on the other hand, just did badly, whereas a shop called OnePoll with a decidedly do-it-yourself feel produced a poll that was wildly inaccurate.

One firm, TNS BMRB, did in-person sampling; its performance was also subpar.

It is probably wrong to lump all the Internet polls together -- I certainly root for the guys at YouGov, who are thoughtful and transparent. But at this point, the challenges facing Internet pollsters are relatively formidable. Moreover, the bias of the Internet polls -- too many votes assigned to LibDems, who have younger and more-wired voters, and too few assigned to Labour, for whom the opposite is true -- ran in exactly the direction that you might expect.

UPDATE: Opinium had originally been characterized as a telephone pollster; they are now properly classified as an Internet pollster instead.

Also, here is a simple visualization of the data, with the telephone pollsters indicated in black, the Internet pollsters in orange, and the in-person pollster in indigo.

There's More...

Contract Post

UK Election Returns: Overnight Blog

9:25 AM [Nate]. And what do the Liberal Democrats do? I'm afraid that I don't really see a good option for them. I don't know that they want a new election to be called quickly, because what has usually happens in these cases in the UK -- when a party comes just short of a majority and calls a new election to try and secure an actual one -- is that the votes for third (and fourth, etc.) parties get squeezed, and the LibDems definitely remained the third party in this election having emerged with no real gain in seats or votes and having been a massive disappointment. But if Cameron knows this (as he surely will), he'll know that they have no real bargaining power and therefore won't give very much up for their support. LibDems could still formulate a coalition, but it would only seem to marginalize them as Labour, having done slightly better than expected in this election and presumably under new leadership, would return to their traditional role as the opposition party.

9:10 AM [Nate]. If Conservatives absolutely ran the table the rest of the way, they could finish with perhaps 315 seats and could get to 323 if they formed an alliance with the Democratic Unionists, who have 8 seats (having lost one today). 323 might technically be enough to form a coalition, but it would be manifestly tenuous and it relies on a scenario in which Conservatives have their coin come up heads 10 times in a row in the remaining, uncalled constituencies.

The bottom line is that Cameron is very likely to be the next prime minister, either in a coalition with the Liberal Democrats or (more likely IMO) as the head of a minority government.

8:47 AM [Nate]. What I don't see is how Gordon Brown expects to remain Prime Minister. If Labour and LibDems absolutely ran the table the rest of the way and Conservatives lost all the remaining competitive seats, they'd collectively finish with about 320 seats. I suppose if they solicited the help of some minor parties they might be able to form a multi-way coalition, but it would presumably be extremely unstable, and it's more likely that Labour and LibDems will collectively finish with about 310, or perhaps 315, seats between them.

8:45 AM [Nate] Signing off shortly. The final vote share looks as though it should be very close to Conservatives 37.0, Labour 29.6, LibDems 23.5. That works out to a seat count of Conservatives 326, Labour 232, LibDems 63 under our model or Conservatives 293, Labour 262, LibDems 62 under uniform swing.

At the end of the day, this whole debate looks to be a bit of an anti-climax. Some combination of uniform and proportional swing -- or some version of uniform swing that accounts differential voting patterns in marginal constituencies and perhaps regional effects, looks to be the most prudent approach. Because LibDems did not wind up improving their vote share, we didn't really strain the models as much as we otherwise might have.

8:18 AM [Nate]. Note that the initial exit poll should actually do quite well! (Although we don't know for sure if it actually did well or if it had errors which cancelled one another out.)

8:16 AM [Nate]. Counting up the remaining seats: Labour should finish between 250-260, Conservatives between 300-315, LibDems between 54-62.

7:16 AM [Nate]. One place where a proportional swing model clearly seems to be faulty is in assuming that a party will lose an especially large amount of votes in VERY safe seats. That happened in some safe Labour seats but not most where there was no real alternative party.

6:58 AM [Nate]. Conservaties take back a seat from LibDems in Camborne & Redruth. Interesting seat -- Conservatives were in third place last time and leapfrogged both parties.

6:48 AM [Nate]. Conservatives failed to pick up Halifax, a set that looked like a good opportunity for them.

6:46 AM [Nate]. LibDems take Wells in Southeast England from Conservatives -- a bit of an upset given how things have gone for them tonight, as they had lost by 6 points in 2005.

6:36 AM [Nate]. At least Nick Clegg won his own seat!

6:24 AM [Nate]. To be frank, while the results are a long way from uniform, our model also doesn't seem to have demonstrated any particular skill!

6:20 AM [Nate]. The national vote is now tracking to Conservatives 37.1, Labour 29.3, Liberal Democrats 23.6, little changed from before. On those numbers, our model would have projected Conservatives 332, Labour 226, LibDems 63; uniform swing would project to Conservatives 295, Labour 260, LibDems 63.

6:11 AM [Nate]. The BBC forecasts that Conservatives will finish with 308 seats. That's an awkward number -- perhaps not quite enough for them to form a coalition with some of the smaller parties, but likely also not enough for Labour and LibDems to form a coalition. These last few seats could matter quite a lot.

6:10 AM [Daniel]. Well Nick Griffin lost in Barking, fairly comfortably as well. The BNP actually lost 2% of the vote from last time when Nick Griffin was not their candidate.

6:07 AM. From the BBC: "0547: There seems to have been a sharp difference in the swing from Labour to Conservative depending on the ethnic minority population of the constituency, says our analyst Prof John Curtice. On average, where the ethnic population is less than 2%, the swing from Labour to Conservative has on average been 5.1%. In contrast, where more than 25% belong to an ethnic minority, the average swing is just 1.7%."

5:55 AM [Nate]. Labour are generally holding their own in working-class cities like Liverpool and Manchester.

5:54 AM [Nate]. Conservatives are performing quite well in some of these LibDem margnials -- just days ago, LibDems would have expected to make some inroads against Conservatives.

5:51 AM [Nate]. Greens win their first ever seat in Parliament, flipping a Labour seat in Brighton Pavilion.

5:44 AM [Nate]. So, here's where we stand right now. The national vote is tracking to Conservatives 37.1, Labour 29.1, Liberal Democrats 23.7.

On those figures, a uniform swing model would predict Conservatives 296, Labour 258, Liberal Democrats 64; our model would predict Conservatives 333, Labour 224, LibDems 64. It appears that the actual results will be somewhere in between the two figures, with Conservatives short of a majority but north of 300 seats, perhaps somewhere in the neighborhood of 310-315. Liberal Democrats, which took a number of tough defeats, also appear although as though they will underperform both numbers.

There's More...

Contract Post

Liveblog: UK Election Returns

5:20 [Renard] Best the Lib Dems could ask for is legitimacy at this point -- something only the Tories would be able to do. Embarrassing losses in Edinburgh South, Oxford East, Devon Central, etc have shown they are not ready yet to be a national force.

5:18 [Thomas D] The Lib Dems sure blew it, but can they win by losing? A Labour/Lib Dem coalition looks out of the question now--not only will it not be a majority, it probably won't even equal the Tories' total. How about a Tory/Lib Dem government in exchange for electoral reform--more likely instant runoff voting than proportional rep.? The Tories get a real govt., and the Lib Dems get their Holy Grail/single-payer system/White Castle burger/female Viagra.

5:16 [Renard] Also, good pickup by the Tories in Cardiff North. Strong showing altogether in Wales for the Conservatives.

5:13 [Renard] Nice hold from Labour in Southampton Itchen. Survived a 10.3 point swing from the Tories.

5:08 [Renard] Another Lib Dem target seat doesn't fall -- Islington South & Finsbury (London seat) required a 0.8 percent swing, but Labour held with a 3 point swing.

5:03 [Daniel B] Karen Buck in Westminster north made it out for a Lab hold; Lab to Con swing of just 0.6 percent.

4:58 [Renard] Edinburgh South is a slim hold by Labour by 0.7 percent or 316 votes; Lib Dems are really missing their chances.

4:55 [Renard] Mad Cap'n Tom? Wasn't he the House majority leader from Texas a while back? I always wondered what his next move would be.

4:54 [Thomas D] Conservative Mark Field won the Cities of London and Westminster with 52%. (A majority!) Not a surprise in a safe Tory seat. Still, want to commend the 84 voters who cast their ballots for Mad Cap'n Tom, an Independent candidate.

4:51 [Daniel B] Balls survives by 1100 votes; enduring a 9.3 Lab to Con swing in the seat.

4:48 [Renard] Big blow to the LDs in Oxford West -- lost a seat to the tories that they won by 14 points in 2005. Tories inched it out 42.3 to 42.0

4:45 [Thomas D.] That 9-point swing in context--Jacqui Smith had been made the postergirl of the parliamentary expenses scandal after she made her sister's house her official domicile and billed porn rentals as an official expense. So she was hardly expected to be a uniform swinger.

4:39 [Daniel B.] Jaqui Smith, bye, bye. Con gain on 43.5 to 30.3 -- 9 point swing.

4:36 [Renard] Hold for the Lib Dems in Carshalton & Wallington. Pulled a 4.3 point swing from the Tories.

4:33 [Nate] Good hold for Labour in Luton South -- that's the sort of seat Conservatives would want to claim a majority.

4:32 [Thomas D.] Still waiting for returns from Greater London and what Conservative pickups there are there. The BBC's map is showing a real North-South divide, with Labour's support clustering in Scotland, Wales and the north of England. And personally, I'm waiting for Penrith and the Border, to see if Rory Stewart goes to Parliament.

4:22. While Scotland has been pretty dull (of 54 reporting seats, no changes), Wales has been flipping all around. Tories are have picked up 4, Labour have lost 4 and gained one , Lib Dems are just down one, and Plaid Cymru gained Arfon as expected.

4:15 [Renard]: Devon Central (Con hold, LDs needed 1.85 percent swing to flip it) shows the Conservative strength. This seat is in the heart of the South West, with LD seats all about in the region surrounding.

4:11 [Daniel B.]: Ed Balls (Normanton) is in a recount; this is one the Tories would really relish.

4:08 [Renard]: Hampstead and Kilburn will be an important seat to watch as well. Strong pick-up change for the Tories based on recent performance in local elections, but Lib Dem surge was seen there and the Labour incumbent not bad.

4:02 AM [Nate]: Stockton South was another good pickup by Conservatives, as was Wyre Forest. Their chance of claiming a majority, whilst not great, is better than generally believed.

4:00 AM [Renard]: For all you Lib-Dem watchers out there, the next seats to pay attention to will be Edinburgh South and Solilhul.

3:56 AM [Daniel B.]: The Liberal Democrats are actually beating Labour in England in terms of votes, but they are down 54-15 on seats. So even without Scotland and Wales
they still have a lopsided advantage.

3:47 AM. If the trouble in polling previously was the "shy tory," it seems this election around it is the "loud but flaky Lib Dem."

3:43 AM. Looking at Oxford East now. Notional 2005 results on new boundariesput LDs on 36, Lab 35.6, Con16.9; however, Labour have won it with 42.5 percent to the Lib Dems 33.6. That's very bad news for Clegg and co.

3:36 AM. In Dunbartonshire West, it was the SNP who lost in second place (by far), but again a swing towards Labour of a whopping 5.5 percent, with the SNP and Lid Dems both losing ground.

3:34 AM [Renard]: Looking at Scotland, there was actually a LD to LAB swing of2.1 percent in Dunbartonshire East (LD hold), which suggests little LD progress will be seen in the Edinburgh seats either.

3:32 AM. Assuming a national vote of 37.1/28.5/24.0, there are 59 seats that our model would project differently from a simple uniform swing. So far, 15 of them have been called: our model is right in 7 cases, uniform swing in 8.

3:14 AM. [Daniel B.]. Wales is equally lopsided for Labour, 22 out of 29 seats on 37% of the vote. A particular note must be reserved for the Scottish Tories who are now leading the Liberal Democrats but have failed as of yet to win a seat. Their odds of taking more than 2 at this point are quite limited.

3:10 AM. If you interpolate from the votes counted so far in Scotland, Wales and England, the national vote share (excluding Northern Ireland) would be Conservatives 37.1, Labour 28.5, LibDems 24.0.

3:04 AM. Cameron's body language suggested a touch of disappointment.

3:03 AM. [Daniel B.] A quick look at the BBC list of announced results reveals the Labour bias in the electoral system.The Conservatives, despite a 3% lead in votes cast, actually trail Labour, 76 to 73 on seats. This has a lot to do with the districts, but also the lopsided margin by which Labour is winning Scotland, with 29 out of 35 seats currently in its hands to 5 for the SNP and 1 for the Liberal Democrats.

2:57 AM. [Thomas D.] British campaigns may be fast, but the counting sure is slow! The polls have been closed for five hours now, and only a quarter of the constituencies have been reported. Can someone bring in Amy Winehouse in a hologram for her analysis of the situation?

2:50 AM. Very few results in from London so far -- only 3 constituencies have reported.

2:45 AM. Two impressive Tory gains in Erewash (from Labour) and Harrogate (from LDs). The Conservative-LibDem marginals in particular appear to be behaving very unpredictably.

2:43 AM. Finally a LibDem gain -- from Conservatives in Eastbourne.

2:39 AM [Thomas D.] As the Lib Dems look like they're having a flop of Howard Dean proportions, the pundits will surely be asking what went wrong. Could be a case of too much tact. As a Tory government looked more likely in the past week, and Clegg flirted with joining a Tory coalition, leaning Lib Dem voters could have reverted back to Labour.

2:37 AM. LibDems hold off Conservatives in Somerton & Frome.

2:27 AM. Now Conservatives are starting to get some quite good results in marginal constituencies.

2:23 AM. Big and somewhat surprising Conservative gain over LibDems in Montgomeryshire, Wales.

2:17 AM. Hard to find many patterns here, to be honest, other than that LibDems are having a very disappointing night.

2:11 AM. Intrade chances of Conservative majority down to 37.3. Clearly Labour are getting some better results here than they were early in the night -- the hold in Gelding, for instance.

2:06 AM. Conservative gain from Labour in Loughborough. However, as they only needed to make up 4 points there, that was widely expected.

2:04 AM (BST). Trying to process everything, but in general think Labour's chance to prevent a Conservative majority is increasing based on results seen in last 1/2 hour or so.

2:03 AM (BST): Labour gained a seat from other in Blaneau Gwent.

1:52 AM (BST) [Daniel Berman]: Its also worth noting the Liberal Democrats just failed to take their number one conservative target, and on a massive 6.5% swing from the Liberal Democrats to the Conservatives. It does seem quite possible that the exit poll may be more accurate than thought.

1:50 AM (BST). Labour starting to get some better results now, such as a good hold in Tooting that wasn't guaranteed. Both we and uniform swing had the Conservatives winning that seat, though we made a bigger error as we had a 6.5-point Conservative win there.

1:46 AM (BST) [Daniel Berman]: It's worth noting that the country as a whole is recording a swing against the Liberal Democrats, something especially pronounced in Scotland where their former leader, Menzies Campbell, saw 5% swing against him. The fact that they failed to take City of Durham, one of their top targets, and in fact only gained a .6% swing, may argue that the Clegg surge, by diffusing and confusing their targeting strategy may have actually cost them seats they otherwise would have taken on their past record of maxmizing their votes in target seats.

1:45 AM (BST): Conservative gain in Battersea on results of CON 47.3, LAB 35.1, LD 14.7. Our model had CON 49.4, LAB 32.2, LD 13.5 there, uniform swing had CON 46.6, LAB 32.5, LD 14.8.

1:37 AM (BST). Becoming difficult to track everything but at a macro level. Labour has lost 3 seats so far, 2 to Tories and 1 to Plaid Cymru.

1:27 AM (BST). Putney is the first result in from London and was a safe Conservative seat, which they held. The results were CON 52.0, LAB 27.4, LD 16.9. We had CON 51.7, LAB 29.9, LD 15.4. Uniform swing had CON 47.1, LAB 29.4, LD 16.6.

1:26 AM (BST). Intrade now gives Conservatives about a 65 percent chance of claiming a majority.

1:25 AM (BST). Arfon in Wales is a Labour loss to Plaid Cymru. The result was PC 36.0, LAB 30.4, CON 16.9, LD 14.1. Here, we had LAB 28.5, PC 27.9, CON 23.3, LD 15.2. Uniform swing had PC 32.1, LAB 25.7, CON 21.2, LD 16.1. Uniform swing did better there and correctly forecast the pickup. We appear to be having some trouble with races involving fourth and regional parties, something I should probably have spent a bit more time working on.

1:18 AM (BST) [Thomas D.] As Dan Berman pointed out today, Scotland remains a Labour-SNP battleground, with the Conservatives effectively shut out. And even after a big win tonight, the Tories will still likely have no more than a couple of Scottish seats. What will it mean for Scotland if they get a Westminster government that they didn't vote for, and that owes them nothing?

1:16 AM (BST). Torbay: LD 47.0, CON 38.7, LAB 6.6. We had: LD 43.1, CON 36.4, LAB 9.9. Uniform swing was LD 42.1, CON 40.5, LAB 6.3 -- kind of a wash.

1:07 AM (BST). [Thomas D.] Pretty big swing in Kingswood, a bellwether constituency. Also, the First Minister of Northern Ireland, Peter Robinson, was just ousted. He's having a sour grapes press conference.

1:06 AM (BST). Kingswood is the first Conservative gain from Labour. The results were Conservatives 40.4, Labour 35.3, LibDems 16.8. We had Conservatives 37.4, Labour 34.1, LibDem 23.2. Uniform swing had Labour keeping that seat: Labour 37.8, Cons 36.9, LibDem 18.4.

1:03 AM (BST): In general, we're doing quite a bit better at predicting the share for the winning candidate. Not necessarily better on predicting totals for 2nd/3rd place, in part because the "other" vote seems to be pretty high.

1:02 AM (BST): Durham North: Labour 50.5, Con 21.0, LibDem 21.0. We had: Labour 52.9, Con 24.2, LibDem 23.0. Uniform swing had Labour 56.0, Con 21.5, LibDem 19.5.

12:59 AM (BST). Thornbury: LibDem 51.9, Con 37.2, Labour 7.0. We had: LibDem 52.9, Con 33.1, Labour 7.7. Uniform swing had LibDem 54.6, Con 35.7, Labour 2.7.

12:54 AM (BST). Darlington: Labour 39.4, Con 31.5, LibDems 23.4. We had 40.5/33.3/20.5. Uniform swing had 44.2/30.9/18.8

12:40 AM (BST). Obviously, a significant lull here. I haven't spent much time thinking about the Northern Ireland results so I'm not in a position to comment on them intelligently.

12:28 AM (BST) [Thomas D.] We've been over the shoddiness of exit polls before, but The Guardian shows another reason for disbelief. Even with the Lib Dems running out of gas in the last week, it's hard to believe that they'd lose seats from 2005. http://bit.ly/cs85M9

11:51 PM (BST) In the three districts that have reported so far, all in Metro Sunderland, Labour is -10.5% from last time, Conservatives +6.0%, LibDems +0.2%, "Others" +4.3%.

11:47 PM (BST). All of those Sunderland district have had boundary changes, so there's some additional uncertainties there.

11:43 PM (BST). Sunderland Central: Labour 46%, Cons 30%, LD 17%. Uniform swing better there -- only about a 5% swing. Good news for Labour.

11:39 PM (BST). Sunderland Central next. We are at Labour 38%, Con 30%, LD 18% here. Uniform swing is at Labour 42%, Con 29%, LD 17%.

11:37 PM (BST). Bottom line: the first two results imply a greater chance of a Conservative majority than implied by the exit polls.

11:29 PM (BST). Labour got 69% of the vote in that constituency in 2005 (on a notional basis). So a huge drop.

11:28 PM (BST). Uniform swing botched that one badly on Labour's total. We were a bit closer, but still too low high.

11:25 PM (BST). Washington and Sunderland West results:

Labour 19,615 (53%)
Conservatives 8,157 (22%)
LibDems 6,382 (17%)
BNP 1,913 (3%)
UKIP 1,267 (5%)

11:23 PM (BST). Washington and Sunderland West next to report. We would forecast Labour 58, Conservatives 22, LibDems 20 in that district, assuming national results of 38/28/23. Uniform swing would project Labour 61, Conservatives 20, LibDems 16.

11:13 PM (BST). Caveat: the Sunderland South results may be an especially dodgy data point because that constituency had a significant boundary change.

11:06 PM (BST). It's just one constituency, but as the only hard data point so far, Sunderland South results imply that EITHER uniform swing is wrong-ish (and our proportional model is right-ish) OR that Conservatives did even better (and Labour even worse) than the exit polls suggested. Either conclusion would be bad news for Labour.

11:02 PM (BST). Without making the regional adjustment, our model would have forecast Labour 49%, Conservatives 22%, LibDems 16% in Sunderland South.

Uniform swing would have predicted Labour 54%, Conservatives 21%, LibDems 15%.

10:56 PM (BST). Assuming that raw results nationwide were Conservatives 38, Labour 28, LibDems 23 (see below), our model would have forecast Sunderland South results of Labour 49%, Conservatives 22%, LibDems 17%. Actual: Labour 50%, Conservatives 21%, LibDems 14%. So far, so good.

10:53 PM (BST). Sunderland South results:

Labour 19,137
Cons 8,147
LD 5,292
Independent 2,462
BNP 1,961
UKIP 1,022

10:49 PM (BST). Daniel Berman(whom you'll hear from later) says one of the networks (Channel 4) suggested that raw exit poll results were about Cons 38, Labour 28, LiDems 23.

10:47 PM (BST). Not sure if those BBC numbers are in marginal seats or countrywide. If countrywide, would be pretty close to the numbers we assumed below (Conservatives 38.5, Labour 29.3, LibDems 23.3) and would probably imply a Conservative majority if our non-uniform model is the right approach.

10:45 PM (BST). BBC says there was about a 7 point swing from Labour --> Conservatives in England. 1 point swing from Conservatives --> Labour in Scotland. Also, slight swing from LibDems --> Conservatives.

10:35 PM (BST). It's quite frustrating that we don't know what assumptions, and what raw numbers, went into the exit poll projections.

10:32 PM (BST). If the exit pollsters are using something more advanced like the PoliticsHome model to project their results -- which accounts for Scotland results and gives Conservatives some extra credit in marginal districts -- we've generally been about 20 seats higher on the Conservative totals than their model on the same numbers. So that would imply Conservatives at ~327 seats, almost exactly the 326 they need for the majority.

10:22 PM (BST). Oldie but goodie: why you shouldn't trust exit polls.

10:21 PM (BST). The one thing that's fairly unambiguously implied from the exit poll is that LibDems dropped significantly from their polling averages. Our model does not really differ at all from uniform swing on LibDem totals if they were only in the range of say 22-24% of the vote.

10:20 PM (BST). Be VERY careful with the numbers below. There's a LOT we don't know (or they're not telling us), including precisely how the seats projection was extracted from the exit poll.

10:13 PM (BST). Here is what our model would project on a popular vote of Conservatives 38.5, Labour 29.3, LibDems 23.3:



10:11 PM (BST). If the exit polls are right but the seat projections are based on uniform swing, we would show a Conservative majority on those numbers.

10:07 PM (BST). We would obviously project a more favorable result than just 307 seats for Conservatives on those numbers. Calculating now.

10:06 PM (BST). IF they're using straight uniform swing to project the numbers, that would imply a popular vote result of about Conservatives 38.5, Labour 29.3, LibDems 23.3.

10:02 PM (BST). BBC nerd says: "The exit polls are based on uniform behavior", a.k.a. uniform swing. So we haven't really learned anything about whether uniform swing is the right approach; it's baked into the projection.

10:01 PM (BST). That would actually be a DROP for LibDems from the last election.

10:00 PM (BST). BBC exit poll predicts Conservatives 307, Labour 255, LibDems 59.

There's More...

Contract Post

UK Quick and Dirty

For readers who have not been tuned in to the UK election much (despite our dense coverage of the last month or so), here is a brief take of mine on the whole thing at Foreign Policy.

Also, in addition to Tom Dollar's smart article from yesterday, here are a couple other cases where minority governments governments worked pretty well.

From 5pm US time/10pm UK time (when the polls close) this evening, we will be doing an election return liveblog with Dan Berman, Tom Dollar, myself and Nate.
---
Renard Sexton is FiveThirtyEight's international affairs columnist and is based in Geneva, Switzerland. He can be contacted at sexton538@gmail.com

There's More...

Contract Post

Scotland: An Electoral Bastion for Labour

From FiveThirtyEight's Dan Berman

While most media coverage and electoral analysis of the UK election has focused on marginal seats in the English heartland, the election campaign has carried on quite differently in the other regions. In Northern Ireland, for example, no major party candidates(with the exception of UUP candidates standing as Tories this time) are standing for election, while the nationalist parties in Scotland and Wales are in position to possibly win additional seats.

Even in places where candidates from all of the major parties are standing, like Scotland, there is a very different race. For instance in Scotland, the Conservatives hold just one seat out of 59, and if YouGov’s breakdowns for the last week --Lab 37 LD 22 SNP 20 Con 17 -- are anything to go on, they are unlikely to pick up more than one or two additional seats.

The first thing to note about Scotland is that many seats have four major parties, not three, competing for seats. This difference is not particularly new. Scotland, at least outside of the cities, remained loyal to the Liberals long after they had faded in the rest of the UK, and the Unionist party, which stood candidates in place of the Conservatives, was a legacy of the Liberal Unionists who split from the Liberals over Irish Home Rule in 1886.

In a sense there were two electorates in Scotland. The first resided in the cities, pitting a heavily Labour working class against a Unionist middle class. Outside the cities, a small-town Presbyterian tradition, closely linked culturally and politically with the Northern Irish Unionists, battled residual Liberals and Labour candidates.

Change came in the 1960s, however. The Unionists were assimilated into the greater Conservative party, reinforcing the impression of them as a London party, which lost much of its base when the Conservatives abandoned “Unionism” in Ireland. With that change, “Unionism” increasingly came to be regarded as simply Scottish subordination to London, rather than unity of the global British Empire.

The Liberals, in their 1960 revival, embraced a suburban strategy in the south of England rather their traditional Scottish roots. As a consequence, there was greater demand for a legitimately ‘Scottish’ party to challenge Labour.

The Scottish National Party (SNP) had existed for some time, but it won its first seat from Labour in a 1967 by-election. In the 1970s the SNP were increasingly successful, winning 31 percent of the vote and 11 Westminster seats in the September 1974 elections. The SNP then went into decline for several election cycles, falling to 2 seats in 1979, and recovering only in 2005 election, when they won 6 seats.

This new revival took place at the same time as the Scottish Tories ceased to exist, falling from 25 percent of the vote and 11 seats in 1992, to 17 percent and not a single seat in 1997. They won back one seat in 2005, but made little progress in terms of votes. Their rural voters increasingly went to the SNP, whether tactically or for conviction, while the Unionist ones went to Labour.

Since devolution, the SNP has had substantial success on the local level, winning control of the Scottish executive a s a minority government in 2007. However, the SNP under First Minister Alex Samond is currently unpopular, kept in office, in a touch of irony, by the Scottish Conservatives and independence has minority support. Their success owes a lot to the proportional elements of the electoral system used for Scottish Elections if the same results as in 2007 were repeated at the Westminster level on May 6th, the SNP would at best pick win 14 of the 59 seats.

Indeed, the Scottish electoral system is even more favorable to Labour than the UK system as a whole. In 2005 the Conservatives won 16 percent to the SNP’s 18 percent and the Liberal Democrats 23 percent. Yet the Conservatives won only 1 seat, the SNP 6, and the Liberal Democrats 11. Labour, winning 39 percent, won 40 seats.

This bias is not entirely new. When the SNP reached its high point in 1974, winning 31 percent, it won just 11 of Scotland’s then 73 seats. When the Tories repeated that percentage in 1979, they won a more respectable 22 seats, but were still dwarfed by the 49 won by Labour. In the bias’ most spectacular performance, Labour actually won 37 of the 73 FPTP seats in the 2007 Scottish Parliamentary elections, despite winning only 32 percent of the vote and coming second to the SNP, which on 33 percent only won 21 seats.

As a consequence, Scotland has, at least since the 1950s, been a massive reservoir of seats for Labour even if it wasn’t always in terms of votes. And all of the polling indicates that with the exception of perhaps 2-3 seats, this is likely to remain the same after Thursday. At worst, Labour may well be reduced to 35 seats, which would still leave them with a majority of the seats from the region. It is for this reason, as much as for any other, that little attention has been focused on the general election in Scotland.

---
This article was authored by research assistant Dan Berman. Please send comments or suggestions to sexton538@gmail.com

There's More...

Contract Post

UK Election Day Scenarios

Although we have our "final" projection out for today's election in the United Kingdom, the polls on the other side of the pond have not always been so great, so it is necessary to build in some tolerance for error.

In fact, the polls have often been pretty bad. They were fully 6.4 points too low on projecting the Conservatives' vote in 1992, and 4.7 points too high on Labour. They were also too high on Labour in 1997, although the error was somewhat masked because uniform swing calculations, which underestimated the number of seats they'd gain given the vote they actually had, pushed in the other direction. Things have been better in the last two elections, but not perfect. Missing the Labour-Conservative margin by 4.2 points, as the polls did in 2001, or even 2.5 points, as they did in 2005, would be considered a middling performance for a high-turnout general election here in the United States.



What if the average bias observed in the polls over the past four elections -- 2.9 points too high on Labour, 2.8 points too low on Conservatives, and also slightly low on Liberal Democrats (but too high on 'other') were replicated today? In that case, Conservatives would finish with 38.3 percent of the vote, LibDems 28.2, and Labour 24.5, which according to our model would produce a fairly impressive Conservative majority as follows:



But, of course, the error could run in a lot of different directions. Maybe the pollsters -- aware that they've overestimated Labour's vote in the past -- will go too far in the other direction and underestimate it instead. Maybe they're shortchanging the number of new voters who will turn out for Liberal Democrats, even if their momentum has flagged some in recent days.

Let's quickly run through what would happen if the pollsters were 4 points too high, or 4 points too low, for each of the three major parties. As a reminder, here is our benchmark: a forecast of Conservatives 312, Labour 204, LibDems 103, assuming that the polls (and our modelling assumptions!) are perfect.



If Labour were to do 4 points better than the baseline forecast -- with their gains coming about equally from the other two parties -- they could still finish with a plurality of seats (a majority looks difficult to achieve even under somewhat best-case assumptions):



If the polls are 4 points too high on Labour, on the other hand, they could still have a devastating election:



If Conservatives outperform their polls by 4 points, they could wind up with a very large majority that is simultaneously embarrassing to Labour and disappointing to LibDems:



If they're 4 points high on Conservatives, though, we could have this messy scenario in which no one (save perhaps the LibDems) emerges with momentum or mandate:



The LibDems, having lost about 2 points from their peak in the polling, are probably now in a position where even a 4-point error in their favor might still leave them as but the third largest party:



With a 4-point swing against the LibDems, we actually have them losing two seats against the 62 they have now:



My intuition is that most of the risks are still to the downside on Labour's vote, given the history of overestimating their vote in the past and that some pollsters are distributing undecideds based on how people claimed to have voted in the last election -- something which tends to help Labour because their voters seem the most indifferent about this election. On the other hand, our model is more aggressive than most others in taking away seats from Labour for a given vote share.

As of this writing, betting markets now give Conservatives about a 38 percent chance of claiming a majority. I don't know that there's much arbitrage either way there, but I'd be slightly inclined to take the Tories on those odds.

There's More...

Contract Post

Final UK Projection: Conservatives 312, Labour 204, LibDems 103

Polling during the past 48 hours has tended to show very slight gains for the Conservatives and Labour at the expense of the Liberal Democrats. Our projection model now forecasts that Conservatives will have 312 seats in the House of Commons (up from 308 in our previous forecast), Labour 204 (up from 198) and Liberal Democrats, 103 (down from 113).



Whereas in Monday's update, our polling model had been pegged to an "average of averages" (those calculated by PoliticsHome and UK Polling Report, respectively), those sites have not updated their figures so I'm taking a simple average of the final polls released by all the UK pollsters instead. That would project a popular vote outcome of Conservatives 35.5, Liberal Democrats 27.5, Labour 27.4, excluding any votes cast in Northern Ireland.

The only other change is that last week's regional polling from YouGov has been added in averaged with their regional polling from the week previous.

Further discussions and disclaimers about our model can be found here. Our model continues to show substantially different results from uniform swing, which would project the election at Conservatives 283, Labour 253, LibDems 81 on these totals.

EDIT: If, as before, we take an "average of averages" from UK Polling Report and PoliticsHome -- both of whom have their numbers up now -- our forecast is only trivially different, with a prediction of Conservatives 315, Labour 205, LibDems 99.

There's More...

Contract Post

Dino Rossi: #51?

Our Senate forecast update from two weeks earlier concluded that the Republicans were unlikely to take over the Senate, partly because they aren't running credible candidates in enough races, and partly because they'd have to completely run the table even where they are.

Both these challenges remain in place. There are several important contests, like Ohio, California, and perhaps Colorado and Florida, where it's not clear that Republicans are favored at all, let alone by a comfortable margin. And with the decisions of Tommy Thompson and George Pataki not to run, the Republicans will have no margin for error.

Subsequent to the publication of our analysis, however, SurveyUSA came out with a new poll in Washington State, showing the incumbent Patty Murray 10 points behind prospective GOP challenger Dino Rossi. Rossi has yet to declare for the race and very well may not -- as we learned with Thompson and Pataki, it is dubious to assume that someone will run before they decide to do so, especially this late into the cycle. Nevertheless, the poll suggests that picking up a 51st seat is still a plausible scenario for Republicans.

On the other hand, the poll was quite out of step with other recent surveys of the same race, including one today from Rasmussen Reports:



It is not surprising that Research 2000 shows the Democrat in a significantly better position than Rasmussen. Rather, that is fairly typical, since both pollsters have had strong house effects. But what about that SurveyUSA poll?

Unfortunately, I just don't know that there's a terribly satisfying answer to this one. SurveyUSA is generally a quite strong pollster, although they've done disappointingly little polling this year that would allow us to place this result into context. Their poll does use a likely voter screen, but then again, so do both of the others.

The poll does contain a few oddities. For instance, 52 percent of the respondents are gun owners, whereas the actual percentage of gun owners in Washington is about 33 percent. And the poll also has Murray trailing Rossi 49-46 in Seattle and its suburbs, whereas Rossi lost to Chris Gregoire 59-40 in Metro Seattle in 2008's governor's race.

Our regression model, which is formulated by looking at the results of polling in other states, thinks that Murray should be about an 8-point favorite over Rossi based on her approval ratings and the political tilt of Washington state. It just doesn't think that the national environment has gotten quite bad enough for Democrats that a relatively noncontroversial incumbent like Murray would be favored to lose in a relatively Blue state, although it may underrate Rossi slightly because his name recognition is strong relative to his electoral experience (if Rossi were credited with being the equivalent of a U.S. Representative instead, it would give Murray a 2-3 point advantage).

So perhaps the poll will prove to be a modest outlier. Nevertheless, Rossi does not to defeat Murray by 10 points to take her seat; as Rossi himself knows, elections are often much closer than that. And the fact that SurveyUSA -- a good pollster which has not historically had a much of a house effect -- shows this kind of a result for a presumed-to-be-safe incumbent ought to be a bit scary for Democrats.

EDIT: Adding to the confusion, an Elway poll of registered voters has Murray with a 51-34 lead over Rossi. Not even going to try to unpack this one -- this has been a weird year for polling and it's just hard to know who you can trust; hopefully things will converge some before we get to the stretch run in August/September.

There's More...

Contract Post

A Hung Parliament Could Work Well For Cameron

From FiveThirtyEight's Tom Dollar

After thirteen years in power, the center-left party finds itself in opposition. The party's flagship prime minister had retired several years earlier, and was replaced by his deputy--a technocratic, uncharismatic finance minister. A series of expense and kickback scandals tarnished the government's reputation. More than anything, the people wanted something new.

The opposition conservatives stood to gain from the government's stumbles, but when all the votes were counted, they could only eke out a plurality in parliament. A strong finish by a forward-looking, center-left, third party, and lingering support for regional parties resulted in a hung parliament. As the conservative prime minister formed a new, minority government, expectations were low. Some even predicted a snap election would be called within a matter of months.

The United Kingdom in 2010? Try Canada in 2006. But if--as 538 projects--the Conservatives come up short of a majority tomorrow, David Cameron might do well to have a chat with Canadian Prime Minister Stephen Harper. For the past three years, Harper has managed to maintain his government without the benefit of a parliamentary majority or standing coalition. From 1993 until the January 2006 election, Canada was governed by the center-left Liberal Party. During this time, the conservative opposition was split between the eastern-based Progressive Conservatives and the western-based Canadian Alliance. Like Tony Blair, PM Jean Chrétien was able to take advantage of the opposition's disarray to present the Liberals as the natural party of government. Chrétien stepped down in 2003, and was replaced as PM by his finance chief (and erstwhile rival) Paul Martin. The Liberals lost their majority in the 2004 election, and formed a minority government against the (now united) Conservatives and sovereigntist Bloc Québécois. Fallout from an MP finances scandal (sound familiar?) led to a broad and deep erosion of Liberal support.

A Motion of No Confidence passed the House of Commons in November 2005, and elections were called for January 2006. The Conservatives gained 51 seats for a total of 124--31 short of a majority. Conventional wisdom was that this government would be short-lived, the Conservatives would have to form a coalition with the Bloc Québécois (hardly a natural partner), and Harper would have to govern as a centrist, Liberal-lite prime minister.

Wrong. Harper governed as a conservative, pushing an agenda of tax cuts, increased oil drilling, international cooperation with the Bush Administration, weakening of gun regulations, increased defense spending, tough anti-terror laws, and re-opening of the same-sex marriage debate. He didn't get his whole agenda through, but neither did his government implode. Harper's 2006 budget passed unanimously because the opposition parties forgot to stand in opposition during debate. The Liberals suffered greater damage from their fall from power than had been anticipated. When Stéphane Dion beat out Michael Ignatieff for the leadership spot, their power struggle betrayed a party in disarray. Without a cogent agenda, the Liberals offered little incentive for voters to pick a new government. Harper knew that any threat of a No Confidence Motion was a bluff, as the Liberals would come out of a new election worse off than they were going into it.

Despite having approved a law requiring fixed election dates, Harper called for the dissolution of parliament in September 2008. October elections yielded a larger Conservative plurality, but still eight seats short of a majority. The New Democratic Party --a social democratic, environmentalist party--picked up eight seats for a total of 37. The Liberals fell to 77 seats, ending Dion's career with the ignominy of being only the second Liberal leader never to become PM.

Ignatieff replaced him, and gave Harper a more aggressive opposition. Harper's 2008 budget would have cut Federal funding of parliamentary elections--to the benefit of the Conservatives. All opposition parties formed an ABC coalition--Anyone but Conservative--to defeat this. The defeat of a budget resolution is considered a Motion of No Confidence, which would have triggered new elections only six weeks after the previous ones. Harper resolved this crisis by having Governor General Michaëlle Jean prorogue the parliament -- effectively locking it out until late January 2009. In the meantime, Harper was able to secure Liberal backing for procedural votes in exchange for scrapping the budget resolution.

Since that shakeup, Harper's government has been forced to tread a finer line to stay in power. Harper also had parliament prorogued again in early 2010 -- ostensibly so that MPs could attend the Vancouver Olympics without having to fret over politics. The fact that this prorogation coincided with increased scrutiny over Canada's role in Afghanistan led to international protests.

So what can David Cameron learn from his friends in Ottawa? The principal lesson is Go Big or Go Home: voters would rather see the government boldly pursue its agenda than dither over coalition-building. A hung parliament does not have to mean inertia, and a plurality of 308 seats should give the Tories enough leeway. Opposition threats of No Confidence Motions and rejection of the Throne Speech could just be a lot of hot air.

Labour could (quite rightfully) conclude that snap elections would simply add more Tory MPs--especially if the perception is that a Conservative government is effective. (Of course, if the Tories--and more specifically George Osborne--are seen as incompetent managers, these threats become more serious.) And prorogation exists as a "nuclear option" for the British PM too. Whether a Harperesque approach could work for the British Conservatives also depends on a number of exogenous factors: how badly Labour handles its defeat, whether Labour and the Lib Dems can form ad hoc coalitions to block certain measures, whether a Conservative-Lib Dem coalition is available as a fall-back option, etc. So while Cameron campaigned as Britain's Obama, the realities of governing may force him to be less of a hoper and more of a Harper.

Note: This post has been corrected to note that Michael Ignatieff replaced Dion as Liberal leader after the prorogation crisis had already begun.

---
This article was authored by research assistant Thomas Dollar. Please send comments or suggestions to sexton538@gmail.com

There's More...

Contract Post

NC Runoff, Coats and Fisher Win Amid Light Turnout

A quick roundup of primary results tonight from the Senate primaries in Ohio, North Carolina and Indiana:

Starting in North Carolina, the Democratic Senate primary to determine who will face Republican Richard Burr this November is not settled yet because no candidate got 40 percent of the vote. Elaine Marshall finished first, with 36 percent, and will face Democratic establishment favorite Cal Cunningham (27 percent) in the runoff on June 22. As the Raleigh News & Observer reported, both "had high praise for third place finisher Ken Lewis, whose endorsement would be a key for either campaign."

In Ohio, Lt. Governor Lee Fisher has beaten Jennifer Bruner (with 86 percent of precincts reporting, as I file this) in the Democratic primary for the US Senate seat in the Buckeye State. Fisher will face former Republican congressman and Bush43 OMB director Rob Portman in the contest to replace the retiring George Voinovich. Polling in this expected matchup has thus far been tight.

Finally, in his bid to recapture the Indiana seat he inherited back in 1989 when Dan Quayle vacated it to become vice president, former Sen. Dan Coats won, but not exactly convincingly: 39 percent to Marlin Stutzman's 30 percent, with the third-place finisher John Hostettler garnering another 22 percent. Coats will face Democratic Rep. Brad Ellsworth (who beat Hostettler in 2006 for the 8th District House seat) in the general election.

But the most telling result of the night may be this, as reported by the Associated Press: "Turnout was exceptionally light in Ohio and North Carolina, a possible indication that the anger fueling voters across the country over economic woes, persistently high unemployment and Congress itself wasn't translating into votes — and, perhaps, the limited influence of the conservatives and libertarians who make up the fledgling tea party coalition."

This was supposed to be the year of Nationalized Scott Brown Phenomenon, of sound and fury signifying something. If in fact it turns out that tea party activism is more bark than bite--as I wondered aloud in a post last night--then expected Republican gains this November may not be as great as some are projecting. Of course, the "perhaps" in the AP quote is an important qualifier, given that the main contest in two of the three states was a Democratic primary fight. Perhaps there just wasn't much for the anti-incumbent crowd to get excited about today.

In any case, although Democrats are facing a tough Senate cycle, two of the seats they have at least some chance of flipping are in Ohio and North Carolina. After today they know the identity of their nominee in the former but not the second. The DSCC will have to decide how much money it funnels into each of these races, but right now Fisher in Ohio looks like the more promising bet.

There's More...

Contract Post

Guest Article: UK Elections Modelling, The Third Way

We present this article from Charles Barry as a potential third method for allocating votes to Parliamentary seats in the upcoming United Kingdom elections. An extensive discussion of the two more common ways -- uniform swing and proportional swing (upon which the 538 model is based), can be found here.


by Charles Barry

When pollsters and journalists want to project the outcome of the UK election from the latest polls, they prefer to use the Uniform National Swing (UNS) model. In this article, I’d like to present an entirely different way of doing things.

I don’t blame them for using UNS. Up until about 1987 UNS had projected virtually every election with astounding accuracy. Furthermore, it was simple to set up and use and could provide objective results quickly. Best of all, on election night, the broadcasters could look at the first results to come in, examine the swing from Labour to Conservative or vice versa and then provide an accurate forecast of the whole election, allowing those who didn’t fancy staying up till 3 in the morning an opportunity to see who was going to be their Prime Minister when they woke up. But over time, and with the rise of third parties, the UNS model has become less and less accurate.

So how does my model work? Simply put, it looks at past general elections and determines – mathematically – the ‘character’ of each constituency, for instance whether it is marginal, safe, volatile, stable and so on. It looks at the relationship in each constituency between national vote share and the local constituency vote shares.

To phrase this a bit more mathematically, we assume that there is a function that relates the independent variable x (national vote share) to the dependent variable y (constituency vote share). What is this function? After some consideration, I decided to use the Cumulative Kumaraswamy distribution which looks like this:



...where y is constituency vote share, x is national vote share and A and B are mathematically determined constants. (NB while this is actually a cumulative probability distribution, I am not using it to model probabilities)

So why did I choose this formula? Firstly, it is incredibly flexible and not particularly difficult to use, allowing it to simulate a variety of electoral scenarios, which I shall demonstrate below. Secondly, it doesn’t come up with stupid results such as “Party A gets 13% of the votes in the constituency despite receiving 0% nationally”. Thirdly, it doesn’t assume that swing behaves in a constant fashion – in most cases, it can model swing in a non-linear fashion. More on this below.

To determine what values A and B take, we use an optimisation routine called the BFGS method; effectively we let the computer hone in on what values produce the curve most likely to model the election results, as implied by the results of the previous four general elections.

My model simulates the seats according to which region in the UK the seat lies – in England, I produce three equations of the type above to model the Conservatives, Labour and the Lib Dems. For Scotland, I also have an equation for the Scottish National Party and in Wales I have an equation for Plaid Cymru. I use national polling data for English seats, Scottish polling data for Scotland and Welsh polling data for Wales. In Northern Ireland, I simply don’t bother trying to perform a simulation, because opinion polling is non-existent. This is a shame, because at this election Northern Ireland may prove crucial as to whether the Conservatives get an overall majority. (The more seats Sinn Fein or the UUP get, the less well off the Conservatives.)

Finally, we can aggregate the results from these 3000 or so unique equations into a general election result. I also introduce errors and then repeatedly simulate the result so I can get a probabilistic forecast as well.

So what about those curves I was mentioning before? I have produced two sample charts which show how the model develops different scenarios for different seats. In all the following charts, the red curve models Labour, blue models Conservative and yellow models the Lib Dems. The horizontal axis is the national vote share for that party, and the vertical axis is the constituency vote share. (The large vertical lines at 100 on the graph are meaningless and simply reflect EXCEL tripping over itself.)



The diagram above shows a very safe Labour seat: Liverpool Walton. You won’t be surprised to learn that in 2005 the Labour party won this seat with just over 72% of the votes in the constituency – a remarkable result considering that they got roughly half that number of votes nationally. For the Conservative party to take this seat they would need to have a lead in the national polls of over 50%.

Going back to what I was saying about non-linear swing, from 0% to about 30%, we can see that Labour get a greater swing than would be expected on a linear swing, while beyond that point Labour get less than would be expected on a linear swing. The reverse is roughly true for the Lib Dems and Conservatives.



Above is Bedford, and it is most definitely not a safe seat! In fact, Bedford is one of the most interesting seats in the upcoming election because it is one of the “M1 Truels” (a truel is a three way duel, as fans of game theory and The Good, The Bad and The Ugly will note). Any of the three main parties could win this seat.

You will note the curves in this seat are much more like straight lines, and the swing in the centre (20% to 50% national votes) is linear. This is why UNS earned its spurs –- quite a lot of seats have linear swings. But you’ll also note that the Lib Dem curve is steeper – which is why they can move from third place to first place in the constituency result, while still staying in third place in share of national votes.

So enough of the fancy charts – let’s get down to business. Let’s look at the predictions. Although, as I said earlier, my forecast model is probabilistic, those distributions take a long time to calculate so I’m going to remove the randomness and look at the central predictions instead.

The weekend polls are just out, so we can compare what would happen under different pollster’s results.

YouGov, ICM and BPIX have all published similar results, with an average of 34.6% Conservative, 27.6% Labour and 28.2% Lib Dem. This is the prediction under those numbers, as compared with the same under UNS:



AngusReid have published a poll which shows drastically lower levels of support for Labour, at 35% Conservative, 23% Labour and 29% Liberal Democrat. If this were accurate, this is the result:



The Tories are only 6 seats short of an overall majority, and they could be just over the line, depending on the results in Northern Ireland. Labour under my model is facing destruction as a party in England. Note also the much higher support for others in the New Model.

Finally, ComRes have published a poll showing the Tories with a runaway lead, and a lower level of support for the Lib Dems, at 38% Conservative, 28% Labour, 25% Lib Dems. Here is the result under their numbers:



If this situation were true, under my model you would already hear the champagne bottles popping at Conservative HQ. Regardless of the situation in Northern Ireland, they would have a comfortable majority sufficient to govern as a single party.

I want to end with a quick discussion of the drawbacks of this model. There are many potential problems with this model, and I highlight many of them in the explanation on my website. For now, I just want to deal with the most important.

The main problem with this simulator is that it only uses data from the four previous general elections. Statistically speaking, this is a very small sample size and we cannot be sure in any way that in the coming election the seats will behave anything like I have predicted. The upcoming election could be a significant deviation from past trends and so there is this inherent uncertainty in the predictions of the model.

However, in my defence, I estimated the average mean error of all 3000 equations by comparing the predictions of the model to past general election results – and it came to only 2.15%. 60% of all the estimations were able to produce a general election result to within 1% of their actual value, and an additional 27% were accurate to within 2.5% of the actual result. Only 2% of all equations were unable to produce a reliable value. And when I use a probabilistic forecast, these errors can be accounted for.

The other main problem with this model is that it requires a “track record” of the seat to be able to “learn” what its characteristics are. The 2010 boundary changes have abolished 9 seats and created 13 new ones and I simply do not have enough data to predict these new ones properly – so I have to rely on simple old UNS for these seats. Additionally, there were substantial changes to the boundaries in Scotland in 2005, and some of the seats I forecast did not exist in 1992, meaning that their track record is also not as accurate as other seats.

However on the whole, I feel that my model could provide a refreshing –- and perhaps more accurate -- alternative to UNS.

You can download the model for yourself here (Excel 2007 format), and I provide detailed probabilistic forecasts on my website.

Charles Barry is a UK student with an interest in economics, politics and finance. He blogs intermittently at http://charlesbarry.wordpress.com and tweets more frequently at http://twitter.com/charlesbarry

There's More...

Contract Post

Today's Democratic Senate Primaries

Democratic voters, though not a whole lot of them, will troop to the polls today in Ohio and North Carolina to nominate candidates for U.S. Senate seats currently held by the GOP. There's not a lot of drama in either primary, as compared to the Republican contest in Indiana. But Ohio will nominate a candidate who is immediately competitive in the general election, while the big question in North Carolina is whether the campaign will go to a runoff in seven weeks.

Lt. Gov. Lee Fisher, a fixture in Ohio politics for a generation (his first race for office was in 1980), is heavily favored to win the Ohio primary over Secretary of State Jennifer Brunner, a candidate who attracted considerable progressive netroots support but not enough money to overtake the front-runner. Two recent polls (Quinnipiac and Suffolk) have shown Fisher opening up a big lead over Brunner (43/23 in the Q-poll, and 55/27 in Suffolk), and he's outspent her nearly four-to-one.

The really good news for Ohio Democrats is that both Fisher and Brunner have been running even or slightly ahead of the unopposed Republican nominee, former U.S. Rep. Rob Portman, best known for being budget director and U.S. trade representative in the Bush 43 administration. A late-April Quinnipiac survey had Fisher leading Portman 40/37, and Brunner up 40/36. A win in OH could help Democrats offset expected losses elsewhere.

In North Carolina, longtime Secretary of State Elaine Marshall (perhaps best known outside NC for beating NASCAR legend Richard Petty in her first race for SoS in 1996) has led the contest from the beginning. But poor general-election trial heat showings against incumbent Sen. Richard Burr spurred the DSCC to recruit and help finance another candidate, former state senator (and Iraq War vet) Cal Cunningham.

Marshall must have a sense of deja vu, since she lost her first Senate primary in 2002 to Erskine Bowles, another DC recruit.

With a third candidate, African-American attorney Ken Lewis, receiving significant but not top-tier support, it's unclear whether one of the front-runners can cross the 40% threshold necessary to avoid a June 22 runoff. The most recent PPP poll of NC shows Marshall leading Cunningham 28/21, with more than 30% still undecided. PPP's Tom Jensen (a North Carolinian) believes the undecided vote will break heavily to the two better-known candidates, probably producing a nominee tonight. As for November, Burr is not a political ball of fire, and is theoretically vulnerable, but has led his two most likely opponents consistently. A mid-April PPP survey showed Burr's approval/disapproval ratio at a less-than-impressive 32/41, though he also led Marshall by 43/37 and Cunningham by 43/35.

There's More...

Contract Post

The Incumbency Effect in the UK

New article on Guardian CIF's election blog up today. (series here)

Initially, our prediction model included a small but seemingly robust incumbency factor, a decision that was certainly influenced by our US election experience. We have since taken it out, determining that the incumbents effect is quite difficult to generalize across the 2010 electoral playing field, and probably will be quite different than what was observed in the 2005 data from which the factor was drawn.

---
Renard Sexton is FiveThirtyEight's international affairs columnist and is based in Geneva, Switzerland. He can be contacted at sexton538@gmail.com

There's More...

Contract Post

Rising Optimism in the White House?

Every White House touts its supposed achievements and good news, while ignoring failures and minimizing bad news. But things are starting to turn around in a way that you get the sense that the Obama Administration is feeling a bit more cheery than they were, say, six months ago.

First, there was the 162K new jobs in March, and we will soon enough know whether that was an aberration or part of a broader turnaround once the Bureau of Labor Statistics announces its April figures this Friday. If the new number is six figures and positive, watch for some serious chirping from Robert Gibbs, David Axelrod and company.

Second, objective improvements in the economic forecast matter electorally only insofar as people perceive the economy to be improving. So you can imagine the cautious optimism in the West Wing as the Administration circulated this New York Times piece and poll, out today, showing that 41 percent of Americans now say the economy is getting better—up 8 points from a month ago. (Only 15 percent say it is worsening.) This news should be taken with a grain of salt, however, for optimism seems to skew heavily in the direction of demographic groups that tend to support the Administration anyway. “Younger and better educated Americans are more likely to describe the economy as on the mend,” writes Marjorie Connelly. “Sixty-one percent of Democrats said the economy is getting better, but only 16 percent of Republicans and 38 percent of independents agreed.” (Do note, however, that independents are more or less right where the country is overall.)

Third: Health care. Now, keep in mind that many of the provisions of the new legislation do not take effect until years from now, after both the upcoming midterms and even President Obama's re-election bid have passed. Still, WH communications director Dan Pfeiffer--or, more likely, some staff surrogate writing under Pfeiffer's name--was blogging quite effusively today about another New York Times piece (this one an editorial) on the subject of how benefits from the healthcare reform bill are already being felt by some Americans.

There is another, fourth reason the White House may be feeling a bit more optimistic about its political situation and the Democrats' situation this autumn: Over the past few months, Organizing for America has been ramping up its organizing efforts. But that is a subject for another, separate post.

Anyway, with a lot of talk about the Democrats potentially losing their majority in the House of Representatives and, though less likely, control of the Senate too, I will go out on a limb tonight--with election day six months from tomorrow--and predict that Democratic losses this fall, though significant, will not be as bad as some are speculating. A lot can happen between now and November, of course, including developments or news that worsen matters for the Obama Administration, Speaker Nancy Pelosi and Majority Leader Harry Reid.

But with six months to go, the political landscape for Democrats is looking a little less uphill than it did with nine months or a year to go until the 2010 midterms. Or I certainly get a sense that the White House thinks so.

There's More...

Contract Post

Conservatives Climb To 308 in UK Seats Projection; Labour 198, LibDems 113

David Cameron's Conservatives now project to win 308 seats in the House of Commons on Thursday, according to our latest projection model; this reflects a gain of 9 seats from the total of 299 that we had projected them at four days ago. Labour's projected total is barely changed -- 198 seats rather than 199 -- while Liberal Democrats now project to win 113 seats, down from 120 last week.



Our model is now calibrated such that the overall popular vote projection reflects an "average of averages": specifically, the polling average as calculated on UK Elections Report, averaged with the one calculated by Robert Ford's team at PoliticsHome. We have also added in several recent polls from Scotland and Wales to refine our regional model. Finally, we have removed the incumbent retirement penalty as we are no longer satisfied that it is a robust effect.

If seats in the House of Commons are distributed exactly as our model suggests, it would create an unusual situation in which Conservatives alone could not form a majority, but Labour and Liberal Democrats could also not combine to form a majority. My personal view is that the most likely outcome of such a scenario would be a Conservative minority government, possibly followed by new elections in relatively short order as Conservatives sought to win an outright majority.

Bear in mind that our model methodology is controversial and unproven. On the other hand, traditional methodologies like uniform swing -- which would project more modest gains by Conservatives and Liberal Democrats -- are in my view also controversial and unproven. Betting markets, on the other hand, actually forecast more gains than this by Conservatives, but fewer by Liberal Democrats.

There's More...

Contract Post

How Far Right, Hoosier Republicans?

There are three primaries being held on May 4, in Indiana, Ohio and North Carolina. I'll do a preview of the Democratic Senate battles in Ohio and North Carolina tomorrow. But perhaps the most interesting contest is the Indiana Republican Senate primary, which offers Hoosier GOPers the choice of three different flavors of conservatism in their effort to win an open Democratic Senate seat against Rep. Brad Ellsworth (who will become the Democratic nominee by action of the state Democratic Party soon after the primary because no candidate filed following Sen. Evan Bayh’s last-minute retirement announcement).

The front-runner all along has been former Sen. Dan Coats, whose career has nicely encapsulated modern Indiana politics (at least before he retired from the Senate and moved to Virginia).

He was a House staffer for Dan Quayle who won his boss’ seat when Quayle defeated long-time Democratic senator Birch Bayh in 1980. He then was appointed to the Senate when Quayle became vice president in 1989, and was eventually succeeded in 1998 by none other than Birch Bayh’s son Evan.

Coats was considered a fairly conventional conservative in the Senate, mainly distinguishing himself by close ties to the Christian Right (James Dobson has endorsed his current bid for office) and for sponsorship of faith-based social initiatives of the sort that eventually were identified with George W. Bush’s so-called “compassionate conservative” agenda. He was an early member of the shadowy Washington-based conservative evangelical prayer-and-power cabal known variously as The Fellowship and The Family, made famous recently by the antics of John Ensign and Mark Sanford.

But Coats also made a habit in the Senate of voting for mild gun control legislation, which has come back to haunt him among Second Amendment absolutists during the current campaign. More problematic than that has been his post-Senate career as a DC-based lobbyist who has represented clients like the Bank of America, Goldman Sachs, and Chrysler, and who also promoted cap-and-trade legislation. (It also probably doesn’t help that the firm, if not Coats personally, lobbied for foreign governments, including Yemen.)

The big question for tomorrow is whether Coats' comeback effort will be snuffed by one of two strangely similar-but-different hard-core conservative challenges.

Coats’ most prominent early rival for the nomination was another blast-from-the-past, former paleoconservative Rep. John Hostettler, who’s main national proponent is Ron Paul. Hostettler managed to hold on to a competitive House seat from 1994 until 2006, becoming a hero of “movement conservatives” for votes against No Child Left Behind, Medicare Part D, and the deal whereby Newt Gingrich ended the “government shutdown” crisis of 1995. For dessert, he also voted against Katrina relief funds. He also continued to support elimination of the Departments of Education and Energy long after other Republicans gave up on such positions, and probably earned his Paul endorsement by being just one of six House Republicans to vote against the Iraq War Resolution.

After his involuntary separation from the House, Hostettler endorsed the far-right Constitution Party’s presidential candidate in 2008. Aside from that apostasy, his main problems are that he has a career-long difficulty in raising money (as of the last quarterly campaign finance report, he had only raised a bit over 50 grand, and obviously couldn’t afford paid media), and that he was trounced by Ellsworth in 2006, losing by the largest margin of any House incumbent. He is leading Ellsworth handily in current general election trial heats, but remains dogged by the impression that he’s determined never to win without winning ugly.

Given Coats’ and Hostettler’s various vulnerabilities, the hot ticket among conservatives going into the primary has been former state senator Marlin Stutzman, who is advertising himself as part of a “new generation of conservative leadership.” While little known statewide compared to his old-school rivals, Stutzman has separated himself from several other obscure right-wing candidates by attracting national attention. He’s been endorsed by Jim DeMint’s Senate Conservative Fund, by RedState’s Erick Erickson, and by Mike Huckabee, and has raised enough money to run some TV ads.

The most recent poll of this contest by a national pollster, SurveyUSA, released on April 29, showed Coats with 36%, Hostettler with 24%, and Stutzman with 18% (but this survey involved a small GOP subsample and carried a 5% MoE). DeMint’s Senate Conservative Fund released an quickie one-night poll a day after SUSA’s, showing Coats at 28%; Hostettler and Stutzman tied at 18%; and 30% undecided, accompanied by some spin suggesting that it’s becoming a two-man race between Coats and a “surging” Stutzman.

Whether or not DeMint’s folks did an accurate survey, they are accurately conveying the anxiety in “movement conservative”/Tea Party circles that Coats will get the nomination via name ID and an even split of the serious right-wing vote. That is, indeed, the CW expectation for tomorrow night, and it’s a sign of the times that a Coats victory would probably be interpreted as a small triumph for “moderate" Republicanism. We’ll soon see if that happens, or if, instead, Hostettler shows he was simply ahead of his time in opposing much of the bipartisan policy consensus of the late twentieth- and early-twenty-first centuries, or Stutzman is the wave of the Red State Future.

FWIW, the SUSA poll shows all three major Republican candidates beating Ellsworth by double-digit margins. But an extended general election campaign is another matter entirely, and Democrats will draw some hope from a contest against a GOP nominee with Coats’ residence and lobbying record; Hostettler’s fundraising struggles and prior failure against Ellsworth; or Stutzman’s relative obscurity and lack of any record of success beyond winning the endorsement of out-of-state conservative agitators.

There's More...

Contract Post