Polling and the Conservative Loss of Political Ascendancy
– James Kanagasooriam, Head of analytics, Populus
In the aftermath of the surprise General Election result, its fall out, and the Prime Minister’s poorly received conference speech, the Conservative ecosystem has gone into overdrive attempting to explain how the party unexpectedly lost its majority and sense of political ascendancy. The sheer number and range of theses (which overlap with each other) suggest that clarity over why the result happened is far away. With such dramatic political inventions going on seemingly every day, it’s perhaps useful to go back to the election and briefly evaluate the five of the most commonly expressed theses on why the Tories didn’t get a majority, to see if any, hold up best. Perhaps in doing this, we might alight on clues as to where the polls might turn from here on in.
1) The structural decline thesis
Commentators, such as CapX’s Robert Colville have for some time been arguing that the markers/behaviours which indicate a higher propensity to vote Conservative are in structural decline. Polling data, and analysis of seat results have shown that Conservative voters are older (see Exhibit 1), less likely to have a degree, more likely to be white, much more likely to be home owners, and more likely to be married. On every single one of these five key markers the country is moving away from the Conservatives. Britain is becoming more ethnically diverse, higher education access has exploded (see Exhibit 2), each generation is less likely to be a home owner than the last, and fewer people get married. Whilst the electoral dividend for Labour may not be immediate from these changes, the seats which have undergone substantial demographic change in the last 10-15 years did show a larger than average propensity to swing to Labour in May. The political fortunes of Canterbury, Bristol North West, Warwick, Reading East, Cardiff North, Oxford West and Abingdon all point to the demographics of the big cities beginning to bleed out into mixed/suburban areas, and not to the benefit of the Conservatives.
Exhibit 1 – the Conservative and Labour party vote shares GE 2017 by average age of seat (England and Wales)
Exhibit 2 – Level of qualification by age group (England and Wales)
The most critical factor here though is the age point. The average age of the Conservative voter is such that the steepness of its “age curve” (the increasing probability of a person at 2017 voting Conservative given their age) is now almost certainly steeper than the natural degree to which people “get” more Conservative as they age. This is important as it suggests that new cohorts of voters cannot replace and replenish the ranks of the Conservatives, even if they do naturally get more Conservative over time. (See Exhibit 3)
Exhibit 3 – Average age of the Conservative/Labour voter vs the adult population
Certainly the issue over younger voters is one that is likely attract much further analysis, and to continue to vex the Conservative eco-system. The evidence is clear however, that for a variety of motives, defecting younger 2015 Conservative voters were decisive in denying the Conservative party a majority.
Exhibit 4 – Understanding which age groups defected/joined the Conservatives (net) from 2015-2017 (Great Britain, YouGov data)
2) The culture wars thesis
This thesis (discussed here on Unherd.com) holds that entire areas of public life (in particular the public sector, education, arts and entertainment sectors) have been ceded to the left, leaving fertile ground for Labour. This thesis holds that university is a place where students routinely leave more liberal then they arrived, and that this process is exacerbating the dividing line between graduates and non-graduates, and by extension the division between young and old (of the latter whom less than 1 in 5 went to university). Whilst there is little data to verify/falsify this particular thesis. The left did hold a comparable sway in the 1980s over similar areas of cultural life, but during an era of Conservative electoral dominance which slightly diminishes the potency of this argument, albeit the 1980s were without the enabling influence of social media and campaigning groups such as Momentum.
3) The unfree market thesis
Matthew Elliot and I recently co-wrote a paper for the Legatum Institute which covered in detail the lack of public support across almost every demographic group for classical free-market policies. We found that:
“On almost every issue, the public tends to favour non-free market ideals rather than those of the free market. Instead of an unregulated economy, the public favours regulation. Instead of companies striving for profit above all else, they want businesses to make less profit and be more socially responsible. Instead of privatised water, electricity, gas and railway sectors, they want public ownership. They favour CEO wage caps, workers at senior executive and board level and for government to reign in big business. They want zero hours contracts to be abolished.”
Exhibit 5 – Legatum/Populus polling on the free market
If figures like these had been more widely known or asked prior to the 2017 election, the evaluation of Jeremy Corbyn’s potential appeal may have been quite different. The faulty assumption of some was that Corbyn was too left wing for voters. This assumption itself can be traced to the flawed thesis that Ed Miliband failed to win the 2015 election because he was too left wing. A post-election YouGov poll revealed that the two principal reasons given for his loss was his that people did not consider as a good candidate for PM (53%) and their distrust of Labour on the economy (42%).
4) The austerity thesis
Whilst our Legatum paper indicated that the public still narrowly backed austerity, the declining support for this economic approach has been significant since the 2010 Coalition government. Some data suggests this decline is a cyclical effect relating to how much the government is actually spending as a proportion of GDP. In other words, support for austerity is declining, as austerity itself is enacted in a process that resembles political homeostasis. Exhibit 6 below indicates that since the 1980s public attitudes on government spending (as measured by the British Social Attitudes Survey) track government spending itself almost perfectly.
Exhibit 6 – British Social Attitudes data on austerity vs public spending data (ONS)
4) The Brexit thesis
One of the more striking headlines from the election was that the result was the “revenge of the remainers”, with younger, middle-class, metropolitan voters deserting the Conservatives in large numbers. One of the more interesting statistical patterns of the election was how sensitive the pick-up/loss of Conservative votes was in relation to the Remain/Leave voting patterns in each seat. The most interesting thing about Labour’s large and widespread rise in vote share however, was the degree to which it was only loosely correlated to the EU Referendum result. Whilst voters were very much joining, staying and leaving the Conservative party with a view to Brexit, the change in Labour’s vote suggests this wasn’t necessarily the case. Indeed a recent NatCen poll has revealed it was this asymmetry of patterns; the Conservatives engaging in a Brexit dominated campaign in England and Wales, whilst Labour campaigned on core left-wing issues, which meant that whilst the Conservatives only had differential appeal in regards to new voters from Leave orientated voters, whereas Labour could pick up votes across the demographic and geographic board.
Exhibit 7 – The swing in Conservative and Labour vote shares 2015-2017 vs the 2016 Leave vote (England and Wales)
5) The Manifesto and leadership thesis
The previous 4 theses don’t and can’t explain fully the election result, simply by virtue of the fact that the Conservative party were some 10-20 points ahead of Labour at the beginning of the election campaign. Certainly each of the issues discussed provided the mood music behind the result, and may have provided the necessary conditions behind the reversal in fortunes during the campaign, but they weren’t sufficient for the hung parliament result.
Exhibit 8 – Public polls a year before up until the General 2017 election
What the data suggests however, is that the Conservatives, even with certain structural demographic issues around their vote, a differential Brexit approach, the waning of public tolerance for austerity, and the weight of public opinion set against the free market, could still have obtained a majority. The clearest evidence for this was the Prime Minister’s IPSOS Mori satisfaction ratings just a day prior to the launch of the Conservative manifesto. Leadership ratings are good statistical proxies for likely election outcomes (with 2017 no exception). What’s striking is that the Prime Minister’s ratings and those of Jeremy Corbyn just a day before the Conservative party manifesto was launched indicated a 15% lead for the Conservatives. By the election the respective party leader ratings implied a Conservative lead over Labour of only 2.3% (the result was 2.4%)
Exhibit 9 – IPSOS Mori leader ratings vs election outcomes
What this all means in regards to current polling
Inputting the latest IPSOS Mori leader ratings into this leadership model suggests that nationally Labour is around c.3% ahead of the Conservatives, broadly in line with where public polls are, which in October had Labour at 42%, and the Conservatives at 40%; a switch around of their vote shares at the General Election. Over the next couple of months it will be interesting to observe, not just the movement on party preference polling, but on Brexit polling, and whether they will start to correlate/become stapled to each other. The most recent YouGov poll was striking as it suggested 64% of voters thought the government was doing a bad job of negotiating Brexit and that 47% thought that voting for Brexit was the wrong decision vs only 42% who believed it was the right decision. It was just one poll however, and it will be worth keeping an eye on how that develops as EU negotiations continue.
After the dramatic and fluid voting intention polling leading up to Election Day, the lack of movement in Westminster voting intention polls now, given the dramatic political events unfurling and various party conferences suggest that voters, perhaps fatigued by events, are tuned out more so than the people who have to cover them.