So, what can read from the weekend's various opinion polls? One quite in favour of Yes, and another quite in favour on No.
However, the poll giving a majority Yes has been rubbished, on the grounds that there was only a sample of 705 compared with the other's 1,000. The Yes poll was deemed to be unrepresentative.
Now from what I recall from my days of dealing with statistics, both samples are unrepresentative. I seem to recall that you needed a sample of at least 1,900 when dealing with polls of this ilk before you could start talking turkey with any form of authority.
Sampling size is important. I play a lot of Spider Solitaire on the PC. Over any given run of 30 consecutive games, I can average as high as 55% or as low as 30%, depending on the cards. But over 100 games, I'm consistently achieving between 40 and 49% (and yes, I'm mad enough to keep records)!
Which brings me onto a particular bugbear of mine - L'Oreal hairspray and Dove moisturiser, etc. All those TV adverts flash up in the small print "87% of 131 women agreed that this product gave extra volume", and so on. Absolute baloney if you're coming at it as a statistician. I'm amazed that the Advertising Stands Agency aren't all over them.
Anyway, can't stop on here to talk to you good folk. A recent survey of 1 Ostrich found that 95% of it was about to head for the biscuit barrel. The other 5% is going along for the ride .....