Forgot to mention this earlier – a article I wrote for Total Politics last month giving tips on the best way to get the most of opinion polls http://www.totalpolitics.com/magazine_detail.php?id=1127
 | 25 Oct | 40 | 24 | 15 | 10 | Con +16 |
 | 25 Oct | 36 | 23 | 18 | 12 | Con +13 |
 | 21 Oct | 37 | 24 | 19 | 11 | Con +13 |
| 21 Oct | 37 | 22 | 19 | 11 | Con +15 |
 | 18 Oct | 32 | 24 | 21 | 13 | Con +8 |
 | 18 Oct | 36 | 27 | 17 | 11 | Con +9 |
| 17 Oct | 37 | 24 | 16 | 12 | Con +13 |
 | 17 Oct | 33 | 29 | 18 | 12 | Con +4 |
 | 15 Oct | 39 | 25 | 18 | 8 | Con +14 |
| 15 Oct | 37 | 22 | 18 | 11 | Con +15 |
| 11 Oct | 33 | 30 | 17 | 12 | Con +3 |
| 10 Oct | 33 | 27 | 18 | 12 | Con +6 |
| 9 Oct | 35 | 22 | 20 | 12 | Con +13 |
 | 7 Oct | 35 | 29 | 16 | 11 | Con +6 |
| 6 Oct | 33 | 27 | 19 | 13 | Con +6 |
| 4 Oct | 38 | 23 | 15 | 12 | Con +15 |
 | 4 Oct | 31 | 26 | 20 | 11 | Con +5 |
| 1 Oct | 34 | 21 | 23 | 12 | Con +11 |
| 27 Sep | 33 | 22 | 21 | 13 | Con +11 |
| 27 Sep | 36 | 24 | 20 | 11 | Con +12 |
| 25 Sep | 33 | 22 | 22 | 14 | Con +11 |
| 25 Sep | 27 | 24 | 22 | 16 | Con +3 |
| 20 Sep | 37 | 22 | 17 | 12 | Con +15 |
| 20 Sep | 30 | 23 | 22 | 14 | Con +7 |
| 19 Sep | 29 | 27 | 20 | 13 | Con +2 |
Good points AW, especially about taking into account all the polling on a given subject (different question wordings etc).
Btw, the sentence beginning “In contrast, anti-hunting campaigners would normally….”, it looks like it has acquired a glitch.
Yes, the title of the piece ended up in the middle of the sentence. Luckily the sense emerged anyway,
A good piece mine host. I’ll just crawl away now..
I printed off the article to pin above my desk at work. I understand the key point about using the Scottish or Welsh figures in UK polls as the numbers are small but isn’t there some reliance that can be put on a cumulative base using a number of sequential polls? Is there a formula or anything?