Evidence on the Tory Strategy of Targeting Marginals

Can be found here. In short, it looks to be successful. Longer, the BPIX poll that was released yesterday is being touted as having a large enough sample size to say something about the marginals that the Tories are targeting, and claims the swing in said seats is significantly higher than the national swing.

Note, this poll shows a Tory plurality of only 2% over Labour, whereas over the past several weeks YouGov has settled into a 5% to 6% range, with that one "blip" of 2% last weekend. A uniform swing would produce a C 252 / L 309 / LD 56 distribution -- in all likelihood a Labour minority government. In order to gain a plurality share of the seats, the Tories would have to take 29 additional seats off of Labour (not the Lib Dems, but Labour); to obtain an outright albeit narrow governing majority, they would need 74 seats in addition to what a uniform national swing would predict.

Unfortunately, nothing on this poll appears to be available aside from the superficial information I've discussed above. I'm especially keen to know how, and to what degree, the marginals were oversampled. While an N of 5655 is impressive for a poll of this nature, a purely random sampling would equate into an N of 8.7 for each of the 650 constituencies. Of course, they didn't sample in purely random fashion as some form of stratified sampling was certainly employed, but still, how large can the N be for the 75 or so odd marginals that the Tories are targeting?

Read more...

More on British Polling and Margins of Error

I promise that this will not be a daily habit of mine, but the Tory +2% lead poll did generate the predictable breathless excitement on these islands.

YouGov's daily tracker indeed moved back towards +6, as anticipated, at Conservative +7. Additionally, a ComRes poll was released yesterday for The Independent which places the Tory lead at +5 (37/32/19). I want to emphasize that the +2 poll released on Sunday, while at the fringes assuming a true value of +6, is still within the margin of error. It was not an outlier in a pedantic understanding of the word, which is a case three standard deviations removed from the mean.

In other words, the findings reported in the Sunday poll were not "wrong". 95% of samples will yield the true value within the error band of +/- 3%, as this was. This is why we have margins of error in the first place.

What we should focus on is not the point estimate itself, but the trend -- and the consistent trend is clearly away from the Tories and towards Labour.

Speaking of which, the +5% Tory lead reflected in The Independent's poll still yields a distribution of seats as Labour 287, Conservative 272, Liberal Democrat 59 -- assuming a uniform national swing.

UPDATE ( 3/3/10): Again as expected, Wednesday's YouGov tracker is rather consistent, at Conservative +5 (C 38 L 33 LD 16). Given the last seven to ten days of polling, this suggests a true value of support around + 6%, and if not precisely 6 (and it isn't) it's slightly below 6%.

Read more...

We May Have a Ballgame Here

Until this is held, due by June 3 (and the smart money is still on May 6 in order to coincide with the annual local elections) I will be paying more of my scant attention to the forthcoming British general election.

The latest YouGov poll, released on Sunday, has the Tory lead at a mere 2%. While Labour supporters should not get too excited, this is consistent with the trend over the past month to six weeks. Since YouGov essentially became a tracker poll on February 17, the Tory lead has been +9, +7, +8, +6, +6, +6, +6, +6, and now +2. If those +6 results reflect the true value of support at C 39, L 33, then this true value rests comfortably within the margin of error of the most recent poll. Hence, this Times column on the volatility of polling is sound advice, aside from his contention that "all polls have a margin of error of 2 points or so plus or minus"; +/- 3% is the norm due to the inefficiencies of diminishing marginal returns. This poll has an N slightly over 1400, which would place the margin of error slightly below 3%; in order to hit 2% the N would have to be around 2500.

I expect the next YouGov poll to move towards the earlier 6% lead, but the trend is clear: Labour are closing the gap for a variety of reasons (save for Gordon Brown, who still trails Cameron in head-to-head approvals). So what does either scenario mean?

If it's Tories +6, at 39 to 33, the Tories would have 293 seats, Labour 280, the Lib Dems 46.
If at +2, at 37 to 35, then it's Labour 316, Tories 256, Lib Dems 48.

Both scenarios assume a uniform national swing, which while not a completely safe assumption, is necessary in order to calculate the distribution of seats. The Tories 'ground game' strategy is to (intelligently) target the marginal constituencies at the expense of running a purely national campaign, and this may yield dividends that would warp the results expected from a uniform swing. However, even here, the Tories would come up well short if the gap is only 2%. In an analysis by Peter Kellner, president of YouGov, the Tories might realize 270 seats to Labour's 300.

What is clear in any of these calculations, be it Tories +2 or +6, due to the vagaries of the British electoral system, neither party would hit the magic 326 necessary for an outright majority. This would result in a minority government and a new election within a year.

Read more...

Reading the Tea Leaves

The Observer published the new Ipsos-MORI poll on Sunday on voting intentions for the forthcoming British election, and the media are all aflutter about its implications. Specifically, the Tory lead has shrunk to six points down from over 20 this past summer: 37% Conservative, 31% Labour, 17% Liberal Democrat.

This matters not only because of the electoral system writ large, but the way the constituencies are drawn, weighted (Scotland and Wales still have a built in advantage in population : seats ration, even post-devolution), and how partisan support is distributed. Here at the University of Plymouth we are considered experts in the field of British electoral politics with our Local Government Chronicle Elections Center. Two of my colleagues in the Elections Center have produced a handy media guide that breaks down the redrawn constituency boundaries for the 2010 election, with a matrix that predicts the distribution of seats in the new parliament assuming a uniform national swing. When 37% Conservative is compared to 31% Labour, we end up with a distribution of C 283, L 273, LD 62: a hung parliament.

However, let's not get carried away, yet. I do have a few critical comments about how the poll is being interpreted. Ipsos MORI are a highly respected polling firm, but nowhere in their releases, hence nowhere in the media, do we find any explicit information regarding the margin of error. We do, however, have the N: 1,006. This basically equates to an MoE of 3% assuming a 95% confidence interval. In other words, the "true" value of support for the Tories is between 40% and 34%, Labour 34% and 28%, etc., with 95% certainty. The best case scenario for the Tories with these numbers equates to: C 329, L 227, LD 63. A comfortable majority.

But wait, there's more!

The overall N and the estimates reported by Ipsos MORI do not match. The support estimates are based on a rough likely voter model / filter which the firm terms "certain to vote". This reduces the N to 513, and roughly increases the MoE to 4.5%. Meaning, the true value is somewhere between 41.5% and 32.5% for the Tories, and 35.5% and 26.5% for Labour. When matched against UK Polling Report's poll tracker, the 6 point Tory lead is an outlier -- not an egregious outlier as it is at least consistent with the trend from the past month, but an outlier nonetheless. (Anthony Wells at UKPR also has an informative take in his blog on this poll hitting different issues than I have here.)

Interestingly, the total size of the sample offering a voting intention of any likelihood is 799, and those numbers are 34-34-16. This suggests that Labour's best strategy is to mobilize their base, or those that are unlikely voters but if they were to vote would vote Labour.

Considering the above, I'm not going to comment on Nick Clegg's tactics regarding the Lib Dems role in a potential hung parliament, or his own grasp of what democracy is all about, nor am I going to consider all the possible ramifications and political gymnastics leading to a hung parliament, but then I am also not going to boldly come out and proclaim that a hung parliament ain't gonna happen, cowboy.

I recognize that the media have a news hole to fill, and in terms of electoral politics here in the UK, this is the most interesting story in a while. However, let's wait for a few more polls to see if this one is indicative, or merely an outlier, before we get all excited about the prospects of a hung parliament.

Read more...

Griffin, the BNP, and the BBC Redux

I, and several others, suggested that in the wake of Nick Griffin's appearance on the BBC, support for his non-racist British National Party would not appreciably increase. Indeed, Andrew Rawnsley writes a scathing, at times hilarious, piece in "Comment is Free" in today's Observer, in which he admits surprise that Griffin turned out to be "a big, blubbery wuss . . . a nervous, sweaty, shifty, amateurish and confused man, manically grinning when confronted with his back catalogue of repulsive quotes and occasionally venting bursts of incoherent nastiness."

In fact, his own party is critical of his appearance on Question Time for not going far enough, for attempting to portray the Party as modern and moderate, and, worst, for not being prepared: "Maybe some coaching could of been done so that Mr Griffin could of answered any questions articulately."

However, while Rawnsley points out that:

The more people see of the BNP, the more poisonous they will see them to be. I take that view even though they claim – not a boast to take at face value anyway – that they got 3,000 new recruits from a programme watched by an audience of 8 million. So the BNP's "breakthrough moment" won over, on his own figures, less than half of a thousandth of those exposed to its leader.

A YouGov poll conducted for The Times in the aftermath of Question Time has some surprising findings -- surprising if considered without context. It's common for any hyped appearance by political actors to receive a polling bump following said appearance, the most obvious of which is the "convention bounce" that the two nominees for the U.S. Presidential election receive following their convention appearances. The same largely occurs in the UK with the party conferences. I interpret this as no different. Indeed, those who would vote for the BNP if an election were held tomorrow increased from 2% prior to QT to 3% in this poll: neither a large boost, and well within the margin of error for a sample size of around 1300. In other words, meaningless.

More interesting, and likely to scare more people, is that 22% of respondents would "seriously consider" voting for the BNP in an upcoming election. Again, I'm not terribly worried that the UK will suddenly become a fascist state (at least not more of a fascist state, at any rate). We don't know the motivations for these responses, but my strong suspicion is that it has far more to do with the general mood in the UK regarding Parliament and the major parties considering the ongoing MP expenses scandal and a general if perhaps unarticulated disquiet with the electoral system. On the latter, the third Labour government was elected with only a little more than 35% of the vote in 2005, meaning 65% of those who voted voted against Labour and Tony Blair / Gordon Brown. Hence I suspect that this vague support for the BNP is a classic manifestation of protest voting, which will (and has been -- the EUP election as an example) manifest itself to greater degrees in second-order elections: those that don't matter as much if at all.

Again, Rawnsley:

The BNP has two main sources of support. At the core are extreme racists. The greater and softer section comes from disaffected voters who feel ignored and disenfranchised by the conventional parties and to whom the BNP presents itself as a stick with which to beat the political establishment.

Hence, I'm not terribly worried about Griffin's QT appearance dramatically inflating support for the BNP. There were only 3,000 more membership applications filed out of the 8 million viewers, and his own party isn't terribly happy with his performance.

The only people happy with the performance are likely the BBC.

Read more...