I’m not suggesting that opinion polls are necessarily or deliberately subsversive but even at their best they are not inert. Its well known in science that in measuring a process you alter the conditions. For instance, in order to measure a temperature you must absorb some of the heat into your measuring device, thereby affecting the system. In most cases the effect is so insignificant as to be irrelevant but the principle still stands.
Polling is a bit of a black art and even pollsters will acknowledge the weaknesses and, after all, they are only indicators with an extremely short shelf life. As in many tests and studies, polling predictions are dependant, to a great extent, on the stability of their presumptions. So the more volatile the situation the less reliable they are. To reliably predict the weather, mathematicians had to come up with the chaos theory and employ very powerful computers to reliably predict the weather over the next week, never mind the next month.
The best we have is a poll of polls, hopefully compensating for the bias of each but any mathematician will tell you that averaging averages is unsafe, especially when the distribution is wide (the average of 10% and 90% is 50% but if that were favourability ratings it would be utterly meaningless). Given a volatile system we also need to factor in how a minor intervention can have major repercussions (put a random…
