Most people think journalists are liars. Believe me, it's the truth.

But the sceptics among you - and, if you trust the results of a new poll, that's the vast majority - will probably choose not to.

According to the Mori survey, politicians rank only slightly higher than journalists in terms of trust, while doctors then teachers top the league.

Like it or not, and that probably depends which profession you're in, it's what we believe.

But look again. The issue is not quite as cut and dried as it might seem, since languishing in the bottom half of the "truth table" are opinion pollsters themselves.

Less than half the public think the likes of Mori, Gallup and NOP tell the truth, while a sizeable share actually class them as fibbers.

The latest figures reflect a slow but steady slip in people's estimations of pollsters. In 1993 52% expressed trust in polling companies and in 1983 it was 55%.

Not surprisingly, Bob Worcester, chairman of Mori, is fiercely defensive of his profession.

"People don't understand statistics," he says. "They cannot believe that 1,000 people can actually reflect the views of 43 million. But they can.

"Ninety-five times out of 100 we have got the results right, with our margin of error of plus or minus 3%."

Basic maths lesson

Of course polling relies on nothing more than primary school maths, but, as Mr Worcester sees it, "most people were not paying attention at school".

Bob Worcester: It's a question of trust, and simple maths

He sums up the job thus: "Opinion polling is a marriage of the art of asking questions and the science of sampling."

The other half of the story lies in the aftermath of the 1992 general election, when John Major's outright victory over Neil Kinnock took everyone by surprise, says Mr Worcester.

It had been a close-run campaign and an NOP exit poll predicted a hung parliament.

Mr Worcester blames the media, particularly the BBC, for blowing the issue out of all proportion.

"Again and again it was said we got it 'horribly wrong'. But if only one in 200 people hadn't voted Conservative but voted for their second choice it would have been a hung parliament."

Spread the blame

Evidence shows that the blame cannot be pinned on one particular polling firm. According to the Independent, pollsters were, on average, nine points wide of the mark.

The result sent a shockwave through the political polling side of the industry and many companies went back to the drawing board in an effort to refine their sampling techniques.

Meanwhile, Mr Worcester has been on a personal mission to set the record straight.

"In September 1995, from reading the polls, I made four predictions at the Liberal Democrat's conference.

"One, the election would be on May 1st, 1997; two, it would be the lowest turnout since the War; three, the Lib Dems would double their representation in the Commons; and four, it would be a Labour landslide."

"I thought we did brilliantly in the 1997 election. I predicted a 159-seat majority for Labour and it was 179. How much more accurate can we get?"

When yes means no

By Mori's own admission - its latest poll findings - the public remains unconvinced.

But that may not be entirely its own fault. After all, one of the hardest-learned lessons from 1992 was that of the so-called built-in Labour bias. Many of those polled said they would vote Labour when, in fact, they had no intention of doing so.

This type of denial was still apparent long after Mr Major's victory. Several post election surveys found more people claiming they voted Labour than could possibly have done. It's a similar phenomenon to when a ropey pop song occupies the number one slot for weeks on end and yet no-one you know will admit to buying it.