Can market research be trusted?

For an event as important as this, pollsters would have enlisted their best experts to work with the latest technology and methodology. Money would have been no object.

If they couldn't get it right with all those advantages and huge budgets, what hope is there for humble marketers trying to establish whether a new baked beans ad for Koo will be effective against an All Gold campaign?

Most advertising depends on market research at some stage. You use it to determine the size of the market for a new product, to test the effectiveness of an ad campaign, or to establish the attitude of the consumer. But what it can't tell you with any certainty is whether it will sell.

When a fast-talking young geek with a laptop buttonholes a voter for his or her opinions about an election candidate, the voter can refuse to disclose, tell a lie or change his mind by the time he votes. Polls are not allowed in Britain on the day before the election, because they can have an influence on the result.

Evidence shows that voters will switch their support to a candidate who seems to be leading in the run-up to polling day. And there was a dramatic late surge favouring the Conservatives.

Ronald Reagan looking on... © iofoto – 123RF.com

When Ronald Reagan was elected US president in 1980, the polls said the race against Jimmy Carter was "too close to call." The result? A landslide for Reagan. Ring any bells? A similar thing happened in 1948, when Harry Truman unexpectedly beat Thomas Dewey for the US presidency and in the UK in 1992, when John Major beat Neil Kinnock. But this one is probably the biggest research failure of them all.

In between, there were long periods when the polls acted reliably as indicators of public voting intentions. Then something happened that was outside normal experience. In 1980, Reagan's support heartland was the mid-west, a land of down-to-earth conservatives, farmers and small-town hicks. There was no way they would reveal their voting intentions to a city slicker with a clipboard (no laptops in those days) who turned up on his doorstep.

In the current British case, what has changed is the use of technology and its impact on how people think and act, and the multiplicity of viable-looking political parties. The pollsters are busy working out what went wrong, and how they need to change. Exactly what this means to their methodology has still to be established, but they probably need to ask their questions in different ways.

But you can be sure they won't be conducting any opinion polls to find out.