Saturday, April 16, 2016

Proportional representation : the clue's in the name

With a certain amount of trepidation, I've been having a look at the latest "if you don't vote RISE or Green on the list you're a blithering idiot who doesn't understand the voting system" article on Bella Caledonia, this time written by RISE candidate Liam Stevenson.  A lot of it covers points that have been rebutted a million times before, but there's one general claim that's worth taking up.  It's one that was also made on Stephen Paton's somewhat misleading video about the Additional Member System a few weeks ago.

"I had a conversation with a friend of mine who was particularly passionate about a party that was seeking list votes due to it’s progressive stance on LGBT+ issues – yet, today, opted to give his second vote to the SNP instead. Why? Because he thought that he was voting for Nicola Sturgeon to be the First Minister.  The entire point of the voting system that we use in Scottish parliamentary elections – that is, the Additional Member System (AMS) – is proportional representation: to ensure that single party dominance does not prevail..."

That's a straightforward contradiction.  The point of proportional representation is not to prevent single party dominance, but rather (and the clue is in the name) to provide representation in proportion to how people actually cast their votes.  It's true that PR makes single-party dominance considerably less likely, because it eliminates the distortion of majoritarian systems which often produce landslide majorities for one party on the basis of 40% of the vote or less.  But the bottom line is that if the electorate chooses to give one party 70% of the vote, a PR system will give that party roughly 70% of the seats.  That's exactly what it's meant to do.  The obvious example is South Africa - the PR system there has produced overwhelming landslide majorities for the ANC, far in excess of anything the SNP could ever hope to achieve, in election after election since universal suffrage was introduced in 1994.  If you're a South African opposition politician, and you're disappointed that PR hasn't produced the parliament of minorities that it does in most countries, the rational response is to persuade people that your party has better policies and better leaders than the ANC - not to scream at voters that they're buffoons for failing to use the system in the way you think they're "supposed" to.

In our own system in Scotland, the proportionality is solely based on how people vote on the list ballot.  The d'Hondt formula attempts to make the overall composition of parliament (constituency and list seats combined) roughly proportional to the popular vote on the list.  It doesn't always fully succeed, because the seven list seats in each region are sometimes not enough to correct a really extreme imbalance in the constituency seats.  But to the extent that the system is proportional at all, it's proportional to how we vote on the list ballot.  Constituency votes can't contribute to proportionality - if they have any effect on the overall outcome, it's by detracting from proportionality, ie. by making the composition of parliament less reflective of the popular will.   So it seems very, very odd for Liam to imply that people who vote SNP on the list are somehow not really voting for Nicola Sturgeon as First Minister.  If you're electing a government or a First Minister on either ballot, you're doing it on the list ballot, not the constituency ballot.  If by any chance the result of the election doesn't properly reflect how people vote on the list, that's a failure of the system (caused by the unwise ratio between constituency and list seats), not some kind of glittering success for PR.

But is that failure going to happen?  In other words, will the SNP break the system by winning an outright majority before a single list vote is even taken into account?  Liam seems to think it's a nailed-on certainty.

"Every poll shows that the SNP will clean up in the constituencies..."

Er, nope. What the polls actually show is that IF there is no change in public opinion over the next few weeks (during the most intense period of the campaign when opinions are most likely to change), and IF opinion poll methodology is bang-on accurate (highly questionable after what happened last year), and IF some very dubious "projections" of how constituency votes might translate into constituency seats just happen to be totally right, then the SNP may not need any list votes to win an overall majority. But there again, if it hadn't been for the Foinavon fence and the shockingly unpredictable behaviour of a couple of dozen horses, I'd have probably won a fortune on the Grand National last week.

Don't let me put you off, though.  If you're an SNP supporter and for some reason you don't care at all about whether there is an SNP overall majority, by all means take a punt on your second-choice party on the list. Bear in mind that if the party you switch to doesn't receive at least 5% or 6% of the list vote in your region (an almost impossible threshold for RISE in particular), your so-called "tactical" vote will at best be totally wasted, and at worst will backfire by helping to increase the number of unionist MSPs. But hey, that's the nature of gambling.

Tuesday, April 12, 2016

No, RISE are not "on 4% in Glasgow"

I was assailed on Twitter today by Jonathan Rimmer, who I gather writes for Common Space and various other publications.  His profile doesn't identify him as a RISE member, but it's pretty obvious he must be a dedicated supporter at the very least.  Basically he was incredulous when I made an observation that I've already made umpteen times over the last few months, ie. that RISE are heading for zero list seats, but that it's unclear whether Solidarity will win zero or one, because we know from past elections that Tommy Sheridan has a small concentration of support in Glasgow that national polls may not be able to pick up.  In an attempt to rebut this point, Jonathan made a number of wild assertions - that RISE are on 4% of the vote in Glasgow, that several polls have shown RISE ahead of Solidarity in "every single region", and that "every" poll that has mentioned RISE has had them ahead of Solidarity.  (The latter point, even if true, wouldn't of course be strictly relevant, because it would be perfectly possible for Solidarity to nick a seat in Glasgow while finishing behind RISE in the national popular vote.)  When I asked him for more details, he became strangely evasive, and just vaguely said that he'd seen the polls on Common Space.  I pressed him further to supply some evidence, and at that point he started exhibiting all the classic signs that he'd probably been bluffing.  "Can't do it, buddy, I'm on the phone!  No, can't do it after I get off the phone either, much too busy.  Do you think I carry polls around with me in my pocket or something?  Why, yes, of course the fact that I'm tweeting right now means that I've got internet access, but not of sufficient quality to access opinion polls.  Don't be ridiculous."

I'll just put him out of his misery and explain quickly why his claims are complete rubbish.  There are only five active pollsters in this campaign, so let's go through them one by one...

YouGov : As you may have seen from the update on the previous post, YouGov revised a mistake in today's datasets - it turns out that for the first time, they asked respondents about RISE by name.  On the constituency vote, both RISE and Solidarity are on zero.  On the list vote, RISE are on 1% and Solidarity mysteriously weren't offered as an option.  YouGov don't provide geographical breakdowns, so can't help with the claims about RISE being on 4% in Glasgow, or ahead of Solidarity in "every single region".

Ipsos-Mori : They don't provide a breakdown by electoral region either, but their most recent poll in February more or less disproves the claim that every poll that mentions RISE has put them ahead of Solidarity.  On the constituency vote, Solidarity were on 0.3%, and RISE were on just 0.1%.  Perhaps Jonathan can claim a very technical get-out clause here, because on the list vote the position was reversed (RISE were on 0.3%, Solidarity were on 0.1%).  But that's pushing it a bit - essentially both parties were absolutely nowhere.  This was the poll I referred to a number of times in the debate I had with Tommy Sheridan a few weeks ago.

TNS : One of only two firms that provide a breakdown by region - but they don't mention RISE by name.  Nevertheless, this seems to be the poll that Jonathan is getting the "4% in Glasgow" figure from - he's using the SSP as a proxy, and the number comes from a tiny, unreliable regional subsample of just 39 people.  A mere TWO respondents in Glasgow said they were voting SSP - and that's what's being prayed in aid as evidence of RISE's dominance over Solidarity in the city!  The SSP were ahead of Solidarity in only five of the eight regions - in the other three, both parties were tied on a big fat zero.

Survation : The other firm that provides a breakdown by region - but they don't allow respondents to express a preference for RISE, the SSP or Solidarity.  So nothing to see there.

Panelbase : The most recent Panelbase poll in January did mention RISE by name and gave them just 1%, but didn't mention Solidarity.  No breakdown by region was provided.

As you can see, none of Jonathan's claims stand up to any scrutiny.  It's untrue that every poll mentioning RISE has put them ahead of Solidarity, because Ipsos-Mori failed to do so on one of the two ballots in February.  It's untrue that several polls have put RISE ahead of Solidarity in "every single region", because the only two pollsters that provide breakdowns by electoral regions haven't mentioned RISE by name (and one of the two hasn't even offered the SSP as an option).  And even if you rely on tiny subsamples, it's not strictly true that RISE are on 4% in Glasgow, because that poll didn't mention them by name - but in any case that 4% was made up of a truly laughable TWO people.

Kalamity Kez in crisis after unsparing YouGov poll confirms Scottish Labour's existence is now pointless

Today's new YouGov poll will come as something of a relief for the SNP, because the fieldwork is bang-up-to-date (it concluded yesterday), and there is no sign at all of the slippage on the list vote reported by TNS last week.  Quite the reverse, in fact.

Constituency ballot :

SNP 50% (+1)
Labour 21% (+2)
Conservatives 18% (-1)
Liberal Democrats 5% (-1)
Greens 3% (-1)
UKIP 2% (-1)

Regional list ballot :

SNP 45% (+2)
Labour 19% (+2)
Conservatives 18% (-1)
Greens 8% (n/c)
Liberal Democrats 5% (n/c)
UKIP 3% (-1)
SSP 1% (n/c)

All the same, today's findings are another reminder of the folly of 'tactical voting' advocates who insist there is some sort of polling consensus that the SNP are running light-years ahead of their showing five years ago.  In fact, YouGov are putting them just 5% ahead of their constituency result from 2011, and just 1% ahead of their 2011 result on the list.

Clearly the Greens' 3% support in the constituencies is up for grabs, because they won't have a candidate in most constituency seats.  Their vote seems to split roughly evenly between Yes and No voters, so the SNP certainly won't be getting all of it, but presumably we'd be safe enough to add on another 1% to the SNP's share.

On the list vote, the Greens will probably have mixed feelings - 8% is healthy enough, but we know that polls have overestimated their strength in previous elections, and there's no sign at all that Patrick Harvie's inclusion in the TV debates has led to any boost (on that point YouGov are in agreement with last week's Survation poll).  They're probably quite unlucky that this is the least competitive Holyrood election to date, because there wasn't the usual sense of excitement over the debates.  People are obviously less likely to have their minds changed by a debate if they're not really paying attention.

This poll is very bad news for RISE, and not just because YouGov are still referring to them as the SSP.  They were offered a rare glimmer of hope last week when TNS became the first firm other than YouGov to put them above 1%.  But if they had really been making progress, you'd have expected their most favourable pollster to put them on at least 2% today, and that hasn't happened.  Having said that, we'll wait and see what the next TNS or Ipsos-Mori poll says - it's not totally impossible that RISE are making modest gains on the ground that only 'real world' polls (telephone or face-to-face) will be able to pick up.

Labour's apparent mini-recovery may not be meaningful, as the the last YouGov poll was particularly bad for them, and the changes are comfortably within the margin of error.  Nevertheless, YouGov have consistently been the most favourable pollster for the Scottish Tories over the last few months, so if the Tories find themselves behind Labour in even this poll, it clearly remains unlikely that Ruth Davidson will replace Kezia Dugdale as leader of the opposition in May (although Dugdale may well be replaced by someone from her own party's ranks).

I don't subscribe to the view that the personal ratings of party leaders are a better guide to the outcome of elections than voting intention figures.  Nevertheless, it's fair to say that Alex Salmond's personal lead over Iain "the Snarl" Gray was a necessary prerequisite for the SNP's dramatic late comeback in 2011.  The fact that Nicola Sturgeon's net rating in today's poll is 50 points better than Kezia Dugdale's, and 28 points better than Ruth Davidson's (amusingly, Willie Rennie wasn't even asked about) doesn't leave the opposition parties with much grounds for hope that the SNP's support could prove soft.  In any case, as late as the turnaround was in 2011, it was already well underway by this stage of the campaign.

There's an absurdly long string of supplementary questions asking whether Kezia Dugdale or Ruth Davidson is better on a range of leadership skills and policy areas.  The results are mixed, but what I find more interesting is why the questions were asked at all.  The poll was commissioned by the Times, so it could be that they're trying to frame this election as primarily a battle for second place, and as a personal duel between Davidson and Dugdale rather than one between the Tories and Labour.  (Presumably they're banking on the Ruth Davidson brand being marginally less despised in Scotland than the Tory brand.)

Someone has asked in the comments section whether the YouGov sample still has far too many people who were born outside Scotland.  The answer is yes, but only in the unweighted sample - YouGov introduced weighting by country of birth during the referendum campaign.  They're also having to downweight Yes voters from the referendum quite sharply, and the same was true of Survation last week.  If by any chance the SNP outperform their standing in online polls, it might be a clue that weighting by recalled referendum vote has its limitations.  OK, there's probably not much danger of people falsely recalling that they voted Yes, but it's not hard to see why a few people might be embarrassed to admit they voted No.  The reason why this point is of interest is that downweighting of Yes voters makes a contribution to the slight No lead that we typically see in online polls at the moment.

*  *  *

UPDATE : Since I wrote this post, YouGov have quietly revised today's datasets and replaced references to the SSP with "RISE - Scotland's Left Alliance".  It's therefore unclear whether respondents were actually asked about RISE by name.

UPDATE II : Anthony Wells of YouGov has sent me a tweet clarifying that they did ask respondents about RISE by name, but initially forgot to update the template for the datasets, which previously referred to the SSP.

*  *  *


I didn't have time to update the Poll of Polls last Thursday, so the percentage changes listed below take into account both the YouGov and Survation polls.

Constituency ballot :

SNP 52.2% (-0.2)
Labour 20.4% (+0.6)
Conservatives 16.4% (-0.2)
Liberal Democrats 5.8% (-0.4)

Regional list ballot : 

SNP 46.6% (+0.8)
Labour 19.4% (+0.6)
Conservatives 16.2% (-0.6)
Greens 7.4% (n/c)
Liberal Democrats 6.6% (+0.2)

(The Poll of Polls is based on a rolling average of the most recent poll from each of the firms that have reported Scottish Parliament voting intention numbers over the previous three months, and that adhere to British Polling Council rules. At present, there are five - Panelbase, Survation, YouGov, TNS and Ipsos-Mori. Whenever a new poll is published, it replaces the last poll from the same company in the sample.)

Explanatory note : As a fond tribute to the mainstream media's restrained take on the GERS report, Scot Goes Pop headlines will contain 50% bonus hysteria for an indefinite period.