Poll numbers from the initial days after British Prime Minister Theresa May’s triggering of snap elections for June 8th were perhaps even more dire than hankerers after New Labour could have prayed. Corbyn-led Labour started off the campaign with headlines, based on data from multiple scientific polling firms, blaring out their twenty some point deficits to May’s Conservatives. While even those numbers should be considered deceiving, a consistent drum beat of better news from many of the same pollsters – and with the numbers read more accurately – suggests May’s lead could have shrunk to single digits already with plenty of time and room for further Labour growth.
Without getting into the weeds too much, there is a simple problem with most of the headlines and even the few poll aggregators active to date in the United Kingdom election cycle. Like the poll aggregators and data experts who gave Hillary Clinton far better chances than even Nate Silver and FiveThirtyEight, insufficient attention is being paid to the huge number of undecided voters recorded by nearly every pollster. (See Silver’s “The Invisible Undecided Voter” from January 2017. As a side note, our final analysis here of the November 8th U.S. election was more accurate on both a national and state by state level than FiveThirtyEight’s. FiveThirtyEight was generally more accurate for the Democratic Primary cycle.)
Could Corbyn pull-off another Brexit or Trump-style shocker? We will not know for five more weeks, but what we do know should give succour to electorally minded leftists while causing serious trepidation to supporters of May and backers of former Prime Minister Tony Blair’s ongoing tantrum around the elevation of Corbyn.
Here, in one chart, is May’s lead over Corbyn from all scientific polling to date with and without undecideds included:
By simply removing undecided voters from the calculus, the numbers used by most media outlets artificially inflate May’s numbers by five points. This happens because the net result works to assign undecideds to the parties according to their numbers strength otherwise in the polling data. Roughly speaking, on average, in the data above May’s Conservatives are assumed to garner just under two undecided voters for every one assigned to Corbyn’s Labour.
But is this a fair assumption?
Not in the least based on the data that we do have. As with Trump’s initial soft support in the U.S. elections, there are far more Labour voters from 2015 who are considering not voting for Labour this time around. Perhaps they will stay away or vote for another party, but just as Never Tump voters, Bernie or Busters, and 2008’s PUMAs (Party Unity My Ass supporters of Hillary Clinton), the overwhelming majority of party loyalists disappointed in the primaries or leadership races do eventually wind up coming home by election day. Indeed, the clear data we have to date from YouGov polling, for instance, shows that this explains much of the gap Corbyn has already closed. Eighteen percent of 2015 Labour voters were undecided in the first full YouGov poll [pdf] after the elections were called (versus just 12% for 2015 Conservative voters). That same number fell to just 11% for Labour for the two YouGov polls in the field April 25-28th [pdf], while falling just a tick to 10% for Conservatives.
Over the course of those same three polls, May’s lead over Corbyn, including the undecideds fell from 16% to just 8%, not very far off from the results (a 6.6% Conservative advantage) of the 2015 election. Meanwhile, the trendlines are all pointing toward continued tightening in Labour’s favour. All five pollsters (YouGov, ICM, ORB, Opinium, Panelbase) with multiple polls out since April 18th have seen Corbyn inch closer with each new poll. (Update as of early Thursday a.m.: Early reports suggest May has increased her lead in the latest YouGov poll and with a sixth pollster that would then have multiple polls, Kantar. We do not have full results, especially including undecideds as of this writing.)
Clustering the polls by date (with undecideds included) over the first week and a half of the campaign, we see that May’s large lead is far from stable . ICM’s poll out Tuesday shaved another point off May’s lead. If the polls that join ICM in making up the next cluster follow this pattern, we are likely to have an average of recent polls in the single digits.
If this trend continues, the Conservative majority in parliament may look very similar to what it has been since 2015.
More troubling for May’s Conservatives, not all of this lead shrinkage can be chalked up to undecided Labour voters moving to the Corbyn column. May also looks to be bleeding support. She could, in fact, end up with no majority at all.