Tuesday, 10 May 2016

Disaster for Corbyn and the Art of Analytical Spin

We have seen it time and again, and in the midst of last week's local elections we saw it once more.

It was a disaster for Corbyn before anyone had voted on May 5, as the votes were being counted, and after results had been declared. There was, and still seems to be, little room for an alternative point of view in this suffocating narrative.

Initially, pollsters and pundits were predicting that Labour was set to lose 100-150 seats. 'Labour set for worst council defeat in opposition in 34 years' screamed the Telegraph headline. They also provided a helpful infographic so you could picture how much of a disaster it would be.

Based on a model by polling expert John Curtice, the infographic suggested:

  • If Labour was level in the polls the party would lose 120 council seats.
  • If Labour was two points behind the party would lose 170 council seats.
  • If Labour was four points behind the party would lose 220 council seats.

Well, a poll commissioned in late April showed that Labour was eight points behind. Yikes.

Then, Corbyn sputtered petrol onto the fire by claiming 'Labour won't lose any council seats'. Instead of managing expectations, he was beating the drum. The stage was set for a Labour meltdown.

Given this ongoing narrative most pundits used election night as an attractive opportunity to peer into an inevitable disaster - and who wouldn't?

Coverage was dominated by Labour's performance, and at the heart of this was the 'high bar' Miliband had set in the 2012 local elections. In what was referred to as Miliband's "best electoral performance" he won over 2000 seats and made hundreds of gains while scoring a projected national share (PNS) of 38% compared with the Tories' 31%. 

Indeed, Miliband hailed Labour's performance as a sign that Labour are "back throughout the country." 

LABOUR    | 2159 councillors (+823) | 75 councils (+32) | 38% PNS  

TORIES      | 1006 councillors (-405) | 42 councils (-12)  | 31% PNS

The media narrative at the time greatly favoured the Labour leader. It was seen as a bad night for the Tories, and according to the BBC report: "Conservative ministers shrugged off the results as typical for a mid-term government." 

While 2012 is a fair bar to set, as it was the last time these seats were contested, there are several clear flaws. First, as Steve Richards argued on last week's Sunday Politicsthe electoral map is more fractured than it was four years ago. Second, we are only one year into Corbyn's stint while in 2012 we were two years into Miliband's tenure.

Third, the two questions journalists asked throughout the night, and subsequently used to pass judgement on Corbyn once answered, are logically inconsistent. The two questions they asked were:

Q1) How did Corbyn do in 2016 compared with Miliband's 'peak' in 2012?
Q2) What does Corbyn's performance in 2016 tell us about Labour's chances in 2020?

If we are using local elections to project the outcome of a general election (Q2), then using Labour's 2012 'peak' as a bar against which to Judge Corbyn (Q1) is completely pointless, as Labour lost in 2015 anyway.

Nevertheless, of the seats being contested, the results were more or less exactly the same.

LABOUR    | 1326 councillors (-18) | 75 councils (n/c) | 31% PNS  

TORIES      | 842 councillors (-48)  | 38 councils (-1)   | 30% PNS

The meltdown never came.

Another way of judging Labour's performance is by comparing the PNS against the vote share in May 2015, and polling from when Jeremy Cobyn was elected leader in September.

LABOUR    | 30% in May 2015 | 30% in September 2015 | 31% PNS in May 2016

TORIES      | 37% in May 2015 | 42% in September 2015 | 30% PNS in May 2016

Labour's support has more or less remained the same over the last 12 months, while the Tories' support has wildly fluctuated; from a good high of 42% in September to 30% in these local elections.

Incidentally, I'm unsure how the official PNS is calculated (though I have no doubt it is accurate), but if you reverse engineer the John Curtice model the Telegraph so gleefully used to project hundreds of seat losses, Labour would be 4.08 points ahead nationally following these results.

Conventional wisdom may suggest Corbyn is in for a hammering in 2020, but I'm not so we can trust conventional wisdom at a time conventional wisdom also suggested Jeremy Corbyn was in for a hammering in the leadership contest.

Steve Fisher, election expert at Oxford University, was consulted for this article which asks what we can learn from these results, if anything. It strikes a balanced tone in a sea of hysteria. On whether you can use local election results to project a general election, Fisher says: "There is absolutely no discernible, sensible correlation." The article was written on May 1. 

Which is where I'd like to bring this back to analytical spin. If you commit to pushing a narrative based on particular assumptions, it's very easy to simply maintain that narrative by cherry-picking data, even after finding your initial assumptions were completely wrong.

Of course, it's possible to interpret these set of results to suit your pre-conceived political bias or pre-formed narrative, regardless of which measure you use, regardless of whether you're sitting in your bedroom in your pants, or in a television studio in your glistening suit.

For example, on vote share, Corbyn's critics can look to a reduction in the PNS in 2016 compared with 2012, 38% versus 31%. Corbyn's supporters, meanwhile, can look to a 12-point deficit in September being swung into a one point lead in May.


This is why, in my opinion, our political analysis can do with a little less of a reliance on constructing narratives. 

In focussing so obsessively on the  narrative of an inevitable Labour meltdown (which, by the way, never came), I feel they missed out on some interesting talking points:

  • Turnout is consistently dire, and in some areas was much lower than in 2012.
  • The Liberal Democrats are fighting back, with some success, having gained the most seats (+ 42) of any party.
  • Roughly as many people support the Lib Dems and Ukip combined (27%) as Labour (31%) and the Tories (30%).
You can see the full local election results on BBC News. Congrats to the Britain Elects Twitter account for its fantastic reporting. Their ward-by-ward results can be found here

Steve Richards, who I cited earlier, has written a great piece via Total Politics on the two figures who will determine the fate of Jeremy Corbyn.

No comments:

Post a Comment