Last night I blogged about the downsides of the Euro electoral system and the tactical voting it encourages. I argued that there were two seats in contention, and they could be between any of the Tories, the Greens, the SNP, UKIP and the Lib Dems. Just after I posted it, though, YouGov’s final Scottish poll was published. It’s got a 533 sample size, so the margin of error is bigger than with a normal 1000 person poll (up to 4.2%, rather than around 3%), but even so..
LAB – 28%
SNP – 26%
CON – 15%
UKIP – 13%
GRN – 11%
LDEM – 6%
The way this would work under the Euro electoral system is as follows (it shows the SNP on double UKIP’s vote, let’s assume they’re a notch above that, which only affects the order in which the last two seats are won):
| Parties | Votes 1 | Seat 1 | Votes 2 | Seat 2 | Votes 3 | Seat 3 | Vote 4 | Seat 4 | Vote 5 | Seat 5 | Vote 6 | Seat 6 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| SNP | 26% | 26% | SNP | 13% | 13% | 13% | SNP | 8.7% | ||||
| Lab | 28% | Lab | 14% | 14% | 14% | Lab | 9.3% | 9.3% | ||||
| Con | 15% | 15% | 15% | Con | 7.5% | 7.5% | 7.5% | |||||
| Lib Dems | 6% | 6% | 6% | 6% | 6% | 6% | ||||||
| Greens | 11% | 11% | 11% | 11% | 11% | 11% | ||||||
| UKIP | 13% | 13% | 13% | 13% | 13% | 13% | UKIP |
The last two vote columns are key. If this were the result, the last two seats would go SNP/UKIP, with both parties’ votes at 13% at that stage (the SNP having been halved because they won the second seat). But look at the Votes 6 column. The Greens would, on this poll, fall just 2% short of claiming that last seat from UKIP. No-one else is even close at that stage – Labour would need 39% instead of 28% to take it, the SNP likewise, the Tories would need 26% instead of 15%, and the Lib Dems would need 13% instead of 6%. With a 4.2% margin of error, even assuming it all went in favour of Labour, or the SNP, or the Tories, or the Lib Dems, they couldn’t stop UKIP.
Only a boost to the Green vote, on this showing, could realistically stop Scotland electing a UKIP MEP.
If we took that two percent from the SNP, that’d leave the last two seats as Green/UKIP rather than SNP/UKIP, which cannot be the objective. So the tactical message to all non-SNP voters who don’t want to see UKIP in is lend the Greens your vote. This applies most of all to Lib Dem voters: your seat is lost, as all the polls agree. Your final message is about in or out of Europe – a Green MEP for Scotland rather than a UKIP MEP makes that case better than a wasted vote for the Lib Dems.
#1 by Derek on May 22, 2014 - 11:22 am
Bear in mind, though, that this poll has a smaller than usual sample, and is out of kilter with every other poll in putting the SNP way below 30%, and UKIP (and indeed Greens) higher than anywhere else (for Scotland). I can’t imagine that was simply down to David Coburn’s charming and coherent performances at the stump this last week.
#2 by James on May 22, 2014 - 11:24 am
I did point out the sample size. And yeah, my guess would be that the SNP would beat Labour (but going on guesses isn’t as good as going on actual professional polling).
#3 by Daniel on May 22, 2014 - 11:36 am
Since the start of the year UKIP and the Greens have generally been rising while the SNP has been steadily falling..
#4 by Mike on May 22, 2014 - 12:25 pm
YouGov polls give a much more accurate result because they actually prompt for everyone. Most polling companies say “will you vote Labour, Lib Dem, Conservative, UKIP or other”, and you have to choose other to get a Green option.
I think your analysis is correct.
#5 by Gaz on May 22, 2014 - 12:36 pm
It’s not the sample size that is the issue here, it is that the demographic/voter id weightings are done for the poll overall and not for the sub-samples.
This is why you get so much variation in UK poll sub-samples – they are not at all representative other than at the top level.
You’d almost be as well drawing numbers out of a hat.
#6 by James on May 22, 2014 - 12:45 pm
UK poll subsamples are c100. This is 533. Your two paragraphs are arguing with each other.
#7 by Gaz on May 22, 2014 - 12:53 pm
Eh, no they’re not.
#8 by Scott on May 22, 2014 - 1:11 pm
Very interesting, but 2 problems in the analysis as I see it.
1. The headline figures in the poll add up to 99% (presumably 100% but 99 due to rounding). Thus any votes for ‘others’ are discounted. Granted, these are fewer than last time, when those not included in the list of parties above took c.10% of the vote. But last time round even the bnp & NO2EU – who are standing again – took 3.5% between them.
For details see https://en.wikipedia.org/wiki/Scotland_%28European_Parliament_constituency%29
2. You only calculate the 4.2% margin of error upwards, but logically it’s just as likely in the other direction – i.e. ukip and the Greens could be substantially lower and SNP or Labour substantially higher meaning that a third seat for one of the big parties is indeed a possibility (SNP or Labour on 30% and ukip or Greens on 9% is all within margin of error).
In sum, not a very useful poll at all actually! But it is fun to consider how low the Lib Dem vote might actually go within the margin of error…
Lastly, since I normally never comment – Thanks for a great blog.
#9 by James on May 22, 2014 - 2:25 pm
Thanks for that kind comment – and yes, the MoE could affect Greens downwards, of course, although I considered it at the strongest position for the SNP’s argument (i.e. them up by that much) and they were still nowhere near. It’d be quite odd for it to be wrong in that very specific way – entirely overestimating Greens and entirely underestimating the SNP, but we’ll see. I reckon almost all the BNP vote has already gone to UKIP, and I’d be surprised if No2EU got anything about 0.5%, though.
#10 by Kristofer Keane on May 22, 2014 - 2:13 pm
Your overall analysis is correct, but a minor correction to make. Your figures for Lab3 and SNP3 are wrong. With each candidate elected, the party’s vote has one added to the divisor, in other words they should now have half of the original vote where as you’ve halved them again, leaving them with a quarter of the vote.
The final round figures should be:
SNP 8.7%
Lab 9.3%
#11 by James on May 22, 2014 - 2:22 pm
Thank you – this was indeed my error.
#12 by Edwin Moore on May 23, 2014 - 8:38 am
We were going to vote Green anyway (my normal vote is Labour) pleasing to see our instinct reflected in, as you say, professional polling. It was Better Nation of course that first raised the notion that UKip might nick a Scottish seat and everyone laughed, Well as Bob Monkhouse used to say, they are not laughing now.
As Scott says, great blog.