h1

The pollsters’ big fear: Thursday could be a disaster for them like 1992

September 15th, 2014


GE1992 Even the exit poll got it wrong

Martin Boon, head of ICM the pollster with the best track record, has voiced on several occasions in recent weeks that the Scottish referendum could be a disaster for them on the scale of 1992. This was when all the firms totally understated the eventual CON lead of 7.5% and only one of them had a lead at all albeit of just 0.5%

For GE1992 was the election when the pollsters got it wrong and led to big changes in polling methodology in the years that followed. This is one of the reason why those who try to predict polling trends by looking back are on the wrong track. The pollsters have moved on.

As can be seen from the screen-shot from the BBC 1992 election results programme even the exit poll was very much out.

The big challenge of the IndyRef is that there hasn’t been one before so there’s nothing to fall back on. Add to that turnout predictions of 80%+ and you create a huge headache for pollsters used to doing general election polls where only 65% actually vote.

The measures that are used to filter out potential non-voters from their samples could mean that they are understating YES. On the other hand there’s what have become known as the “shy Noes” – those who don’t want change but are reluctant to admit it.

So another 1992? My reading is that Thursday could produce a YES victory or a NO win by up to a 10% lead. Hedging my bets? You bet.

Mike Smithson

2004-2014: The view from OUTSIDE the Westminster bubble