First, a warning: a lot of what follows has involved some quick and dirty detective work, and a bit of guessing. But second, a claim we hope will outweigh all that: all our workings will be clear, transparent, consistent and (we hope) compelling.
Back in early November, we took a historically-informed look at the data to sketch out UK Labour's likely score at the next General Election, now due just over four years away in May 2020. Our conclusion? Well, given where its ratings were then, we decided that Labour was dicing with a really heavy, heavy defeat.
That was two months ago: and we thought we'd return to the numbers now we're in the New Year, and take another look at how Labour are doing. The answer is: about the same really, and to be honest a bit worse. Gradually, Labour's hopes of getting anywhere in this Parliament are draining away. Take a look at the quotation above if you don't want to take our word for it. It's taken, with all copyright and rights reserved, from a new article by Professor Tim Bale, of Queen Mary, University of London. The article from which we've taken this just been published in The Political Quarterly. It predicts 'total and utter disaster' for Labour. And we concur, completely.
In this blog, we're going to show exactly why, as we to try to refine and to reapply our early-November techniques, and to take a closer look at what the numbers we have now - and we have quite a lot - tell us about the likely scale of Labour's vote in 2020. In the end, the conclusion is exactly the same as Prof. Bale's, and here we're going to give you lots of numbers as to exactly why.
Let's start with where Labour are in the polls now. During December (the last month for which we have a full round of data), Labour stood at 31.2% in the public opinion polls. Their position has since deteriorated a little - again - but we'll leave that to one side. Since 1979, from this point in each Parliament, there has been an average 3.55% fall between the main Opposition party's score right now and its vote share at the subsequent General Election. That implies a Labour score next time of 27.6%. Labour has, historically, always under-performed its mid-term scores much more seriously than the Conservatives have. If we applied that party's fall from this point in each Parliament (7.7%), we'd get to a prediction of just 23.5% next time. Grim stuff, we're sure you'll agree.
Ah, we can hear you thinking. But it's not as simple as that, is it? The polls usually overstate Labour when we actually get to elections, and given that the apparent 'fall' from this point in each Parliament might be a statistical artefact - a trick of the polling light. If pollsters have now fixed that propensity to overstate Labour's position, then we need to adjust for the fact that their poll ratings will not 'fall' by quite so much - since they will never have been quite so strongly ahead in the first place. Now that's totally fair enough. A more-than-justified objection. So what we propose is this. We can use the average 'error' at each General Election to look at the size of that polling gap, and then subtract that from Labour's supposed decline from this point in each successive Parliament. Then we might get a truer picture of the 'real' fall away in voters' inclination to put their trust in Labour between the end of Parliaments' first year and their end. It's a very gritty and quite nasty way of looking at the problem, but it is one way of trying to anchor projections in reality - a benchmark for polling error, if you will. The adjustment probably flatters Labour, as telephone polls are probably still overrating the party in their original sampling; and the polling industry has not yet settled on a total overhaul that might make us entirely confident that all of this 'bias' has been eliminated. So techniques are not quite as different now as you may imagine, as compared with the those deployed in the last Parliament. But let's be generous to Labour for a bit. No-one else is at the moment.
Still with us? Thanks for getting this far. The mean overstatement for all Oppositions in the polling is only 0.7% since 1979, but for Labour alone it's 1.7% - a small but in some ways slightly heartening part of the party's failure to turn polls into votes that might just have been created by pollsters' sampling or other methods all along. If we apply this overstatement from each Parliament (ranging from a pretty large 4.2% in 1992-97 to an actual understatement of Labour's position of 2.7% in 1979-83), we get rather less of a fall from here to the next General Election: of 5.9% (the figure is 2.8% if we look at all Oppositions). So from this 'adjusted' data, we can say that Labour's vote if we look only at Labour's performance from this point might be around 25.3%, or 28.4% if we use the slightly more generous measure of 'all Oppositions', including Conservative Oppositions, performance. So on the basis of past polls and past Parliaments, adjusting in a very crude way for polling error? Labour might get between about 25% and 28% at the next election.
There's another way of confirming this - via local election results. Now, yes, we know that the 2016 local elections haven't happened yet. But the eagle-eyed among you will have spotted that the most respected experts in this field, Colin Rallings and Michael Thrasher, are predicting that Labour will only get about 31% as a National Equivalent Share of the Vote in those contests (as against 32% for the Conservatives). Here again we can make a straightforward, crude and linear extrapolation of what that might mean if it does happen. Here there is no 'error' to look out for: there are real votes in proper ballot boxes that we're dealing with. On average, since 1979 there has been a 3.2% fall in Opposition's poll numbers from their first set of local elections to the next General Election. Labour's fall has, again, been worse here - of about 4.5%. If we use the 'all Oppositions' score to suggest from here Labour's support in 2020, that might indicate a vote share of 27.8%; if we look just at Labour's numbers, the party might receive 26.5% of the popular vote. Now, again, this is only an average, but only once has an Opposition done better in a Parliamentary election than it did in its first set of local elections - in 1993 looking forward to 1997, a period which was of course marked by the election of a charismatic and centrist new Labour leader. We can't recall his name right now, but we think he was pretty successful at this elections business. Anyway, the headline is this. If Rallings and Thrasher are anything like right, Labour in 2020 will be in very much the same range, on a historical basis, that the polls now suggest - these numbers giving us projections very tightly inside the bounds that the polls right now suggest.
So here's the headline: we have an upper bound of 28.4% on Labour's performance in 2020 (not, perhaps, quite by coincidence almost exactly their vote share in 1983); and a lower limit of 25.3%. If as in November we limit the Conservatives to a very modest 40%, that would produce a Conservative majority in a 'reformed' 600-seat House of Commons of about 130, with Labour reduced to maybe just around 160 or so MPs. In the best-case scenario based on these past historical relationships, there would be a Conservative overall majority of 90, and about 185 Labour MPs might get into Parliament.
We said in November that Labour was likely, on these assumptions, to receive between 26.4% and 28.7% of the vote in 2020. Things have moved on two months, and we've tried to refine our techniques. The result? The range is 25.3% to 28.4% - a little less optimistic for Labour in terms of both its mean and its range. The party is sinking in the morass that its own leadership deficiencies and divisions represent - running a series of scenarios, remember, that have been designed to be favourable to Labour. We've not included the crudest, simplest calculation - of past declines in the polls just mapped on to Labour's present score. We've adjusted for the fact that they might have been overestimated in the past, by working out a gap between polls and performance that might just reflect voters' second thoughts on the day rather than some distance from an unreadable 'true' score; we've assumed in those calculation that many of the problems of over-egging Labour's ratings have mostly been solved; we've not included the latest polls, which are slightly worse for Labour.
Ah, we can hear you thinking. But it's not as simple as that, is it? The polls usually overstate Labour when we actually get to elections, and given that the apparent 'fall' from this point in each Parliament might be a statistical artefact - a trick of the polling light. If pollsters have now fixed that propensity to overstate Labour's position, then we need to adjust for the fact that their poll ratings will not 'fall' by quite so much - since they will never have been quite so strongly ahead in the first place. Now that's totally fair enough. A more-than-justified objection. So what we propose is this. We can use the average 'error' at each General Election to look at the size of that polling gap, and then subtract that from Labour's supposed decline from this point in each successive Parliament. Then we might get a truer picture of the 'real' fall away in voters' inclination to put their trust in Labour between the end of Parliaments' first year and their end. It's a very gritty and quite nasty way of looking at the problem, but it is one way of trying to anchor projections in reality - a benchmark for polling error, if you will. The adjustment probably flatters Labour, as telephone polls are probably still overrating the party in their original sampling; and the polling industry has not yet settled on a total overhaul that might make us entirely confident that all of this 'bias' has been eliminated. So techniques are not quite as different now as you may imagine, as compared with the those deployed in the last Parliament. But let's be generous to Labour for a bit. No-one else is at the moment.
Still with us? Thanks for getting this far. The mean overstatement for all Oppositions in the polling is only 0.7% since 1979, but for Labour alone it's 1.7% - a small but in some ways slightly heartening part of the party's failure to turn polls into votes that might just have been created by pollsters' sampling or other methods all along. If we apply this overstatement from each Parliament (ranging from a pretty large 4.2% in 1992-97 to an actual understatement of Labour's position of 2.7% in 1979-83), we get rather less of a fall from here to the next General Election: of 5.9% (the figure is 2.8% if we look at all Oppositions). So from this 'adjusted' data, we can say that Labour's vote if we look only at Labour's performance from this point might be around 25.3%, or 28.4% if we use the slightly more generous measure of 'all Oppositions', including Conservative Oppositions, performance. So on the basis of past polls and past Parliaments, adjusting in a very crude way for polling error? Labour might get between about 25% and 28% at the next election.
There's another way of confirming this - via local election results. Now, yes, we know that the 2016 local elections haven't happened yet. But the eagle-eyed among you will have spotted that the most respected experts in this field, Colin Rallings and Michael Thrasher, are predicting that Labour will only get about 31% as a National Equivalent Share of the Vote in those contests (as against 32% for the Conservatives). Here again we can make a straightforward, crude and linear extrapolation of what that might mean if it does happen. Here there is no 'error' to look out for: there are real votes in proper ballot boxes that we're dealing with. On average, since 1979 there has been a 3.2% fall in Opposition's poll numbers from their first set of local elections to the next General Election. Labour's fall has, again, been worse here - of about 4.5%. If we use the 'all Oppositions' score to suggest from here Labour's support in 2020, that might indicate a vote share of 27.8%; if we look just at Labour's numbers, the party might receive 26.5% of the popular vote. Now, again, this is only an average, but only once has an Opposition done better in a Parliamentary election than it did in its first set of local elections - in 1993 looking forward to 1997, a period which was of course marked by the election of a charismatic and centrist new Labour leader. We can't recall his name right now, but we think he was pretty successful at this elections business. Anyway, the headline is this. If Rallings and Thrasher are anything like right, Labour in 2020 will be in very much the same range, on a historical basis, that the polls now suggest - these numbers giving us projections very tightly inside the bounds that the polls right now suggest.
So here's the headline: we have an upper bound of 28.4% on Labour's performance in 2020 (not, perhaps, quite by coincidence almost exactly their vote share in 1983); and a lower limit of 25.3%. If as in November we limit the Conservatives to a very modest 40%, that would produce a Conservative majority in a 'reformed' 600-seat House of Commons of about 130, with Labour reduced to maybe just around 160 or so MPs. In the best-case scenario based on these past historical relationships, there would be a Conservative overall majority of 90, and about 185 Labour MPs might get into Parliament.
We said in November that Labour was likely, on these assumptions, to receive between 26.4% and 28.7% of the vote in 2020. Things have moved on two months, and we've tried to refine our techniques. The result? The range is 25.3% to 28.4% - a little less optimistic for Labour in terms of both its mean and its range. The party is sinking in the morass that its own leadership deficiencies and divisions represent - running a series of scenarios, remember, that have been designed to be favourable to Labour. We've not included the crudest, simplest calculation - of past declines in the polls just mapped on to Labour's present score. We've adjusted for the fact that they might have been overestimated in the past, by working out a gap between polls and performance that might just reflect voters' second thoughts on the day rather than some distance from an unreadable 'true' score; we've assumed in those calculation that many of the problems of over-egging Labour's ratings have mostly been solved; we've not included the latest polls, which are slightly worse for Labour.
What all this would mean for the scale of Labour's defeat is rather easier to say. It's by no means quite certain yet that the present government will achieve its aims of reducing the Commons seat total to 600, thus weakening Labour's position by reducing the number of smaller English urban and Welsh seats that it holds now. But it is likely. And in that case - there's no nice way to say this - between 54 and 79 Labour MPs are going to be looking for new jobs (we put that minimum number at 'only' 42 just two months ago). They can keep their heads down for now. They can avoid confrontation with Left-wing activists and their own Leader's office if they want. But every day brings them closer to a General Election. They can't avoid their rendezvous with the electorate - which, on present trends, looks for many Labour MPs like it is going to be a very short, sharp clash with public opinion indeed.
Now we know that stuff happens - one of the reasons for all those poll declines for most Oppositions struggling through most Parliaments. In 1982-83, the economy began to recover, and the Government enjoyed a bit of a boost from the Falklands War. In 1987, the economy boomed. The damage to Labour's reputation wrought by the Scottish independence referendum and the SNP scare in England could not have been foreseen in 2014-15. And in the next few years there is almost certainly going to be a new Conservative leader. Perhaps there'll be a recession. Maybe the Conservatives will be riven by internal strife. All quite possible. But all things being equal and on average we would not expect Labour's position to improve from here. Quite the opposite. They are in a big, big hole - and they just keep on digging.
Everything confirms it. Local by-elections confirm it, with Labour's vote down across the board from the last Parliament. Leader ratings confirm it. Private donors' reluctance to waste their money confirms it. Everyone knows what is going to happen in 2020. We know it. The betting markets know it. You know it, dear reader, in your heart of hearts. The dogs in the street know it.
There is only one conclusion that we can possibly draw, on exactly the same lines as Prof. Bale's: if nothing changes radically between now and 2020, Labour is headed for disaster. And yet the party's membership appear to welcome that fate. They resemble nothing more than a pirate ship storming directly towards the rocks, the crew high on and thrilled by the impending wreck as if enjoying their dizzying participation in some dark revel that they don't want to end - all the better, perhaps, to prove their bravery and bravado. It is a strange sight. It is hard to take one's eyes off it. But the impact, the destruction? Well, that'll be much harder to look at - even if anyone is left, or wants, to sort through the debris.
