Two interesting pieces of Opinion Poll news

The first is from America and relates to the US Presidential elections. From the BBC coverage of these elections you would assume that a Barack Obama victory was a dead-cert as almost every interviewee expresses a preference for Obama. However I see on Gallup that

"Obama Has Modest 4-Point Lead" and that "Barack Obama maintains a modest 46% to 42% lead over John McCain among registered voters nationally in Gallup Poll Daily tracking, with no sign yet of a significant "bounce" from Obama's high-visibility world tour."

Meanwhile I noticed a very telling remark on Political Betting where in an article on the latest MORO opinion polls I read (my emphasis)"

These are the figures with comparisons on a month ago CON 47% (+2): LAB 27% (-2): LD 15% (-2). These figures, like all MORI surveys are based on those “certain to vote”. The fieldwork ended on July 20th - so it’s up to date.

The pollster which is the only one that does not applying a political weighting to its samples. has undergone a major methodological review which took place following the London Mayoral election. All surveys are now done by telephone and the pollster takes measures to ensure that it is not over-representing public sector workers in its samples.

Now why would not over-representing public sector workers be important? Could it be that Gordon Brown's client state is more likely to vote for the hand that feeds it than the rest of the population?