the brilliantly leaping gazelle

Today is a day of two polls, which together illustrate how political polling has as much credibility as shampoo adverts.

Today there is a poll in the Daily Mail that carries the shock news that;

Theresa May’s hopes of winning a landslide election victory suffered a major blow last night after her poll lead plummeted to 11 per cent amid the row over her tax and pension plans.

A Survation poll for The Mail on Sunday put the Conservatives on 40 per cent, followed by Labour on 29 per cent and the Lib Dems and Ukip level on 11 per cent.

It means Mrs. May’s lead over Jeremy Corbyn has nearly halved in four days: a poll immediately after she called the Election gave the Tories a 21-point advantage.

However, there is another poll, first published on the Guardian homepage as a top story, but now relegated to the politics page announcing the equally shocking news that

The Tories have more than doubled their lead over Labour to 19 points since Theresa May called a snap general election last Tuesday, according to a new poll that suggests theConservatives are heading for a landslide victory on 8 June.

Confused? Possibly, if one only reads the headlines and the first paragraph of each story. Although the Guardian to be fair does point out in the second paragraph:

The survey by Opinium for the Observer, conducted on Wednesday and Thursday of last week, puts the Tories on 45% (up seven points compared with the previous week), while Labour is down three points on 26%.

To be fair? Sorry, I meant to be vague. At least the Daily Mail put the Guardians poll in a slightly more understandable context;

Last night an Opinium poll, taken before the tax, pensions and overseas aid row, put the Tory lead at 19 points.

So who to trust? It’s a tough one! But earlier on I mentioned shampoo adverts? You know, the ones on the television that claim that 93% of women agree that this shampoo is the best shampoo in the history of ever? And for a few seconds you think ‘ Wow! 93%. That’s impressive.’ But then reason and logic kick in. And you see that buried in the smallest possible writing is the revelation that only 200 women were asked if they agreed. And you wonder how such a big claim could be based on such a little sample. Who, for example are these women?

And then you start to question the validity of ALL polls, realizing that the sample of people questioned in survey is low – in both cases above less than 2100 – and remembering that at the last election in 2015 pretty much every poll got it wrong!