THE ORPINGTON POLL
SIR,—In so far as the comments in 'Press and People' in your issue of March 27 refer to Mr. Gregory, I am sure he can and will look after him- self. In so far as they refer to National Opinion Polls I should like to make the following observa- tions. We took our sample of the voters in Orpington on the Saturday and Sunday before the election was held and were quite happy with it. Between our sample taking and polling day several things happened. In the first place there was the Blackpool result, secondly Mr. Goldman himself was getting a very bad press (vide the Sunday Express), thirdly there was the publication of the poll itself. I can quite see that under certain circumstances the pub- lication of a public opinion poll can have an effect on an election result. It does not follow that it will have an effect nor is the actual effect predictable. In some circumstances it may result in confirming the 'forecast' and in other circumstances it may work in the opposite direction. In the case of Orpington it is our view that our forecast was accurate at the time we made it and that subsequent to our taking the sample certain events had their effect and the publication of the poll was almost certainly one of them.
I think you do us less than justice in two respects.
In the first instance there was a very high turnout on polling day, 80 per cent, in fact. This is excep- tionally high for a by-election and although we had estimated there would be a high turnout we did not expect it to be as high as this. Nor did anyone else. This in itself threw out our prediction on total votes. In the second place we always ask the question 'Leaving your own party sympathies on one side, who do you think will win?' and publish the re- sult. In every poll we have carried out this has tallied with the overall findings in voting intention. It didn't at Orpington. We continued our practice of publishing the expectation but our forecast was made on the basis of voting intention and we were certainly not `hedging' in publishing this figure but conforming to our normal practice. The fact of the matter was that most people expected the Conserva- tive to win at the time we took our poll and on past voting they had much to support their views. It has long been recognised by pollsters that one is at risk between carrying out the sample and polling day and in this particular case there were factors which acted to throw out the actual predic- tion. We try to take our sample as near to polling day as possible but some interval is unavoidable.
In short, under the circumstances our forecast for Orpington must be considered reasonably good and 1 can find no record of any other organisation which was forecasting the Liberal would win, let alone attempt to say what the majority would be.
National Opinion Polls Ltd., Tallis House.
Tanis Street, EC4
Previous page
