We are grateful to our friends from the Left for helping to improve the monthly ConservativeHome survey in future.
They have done so in the following way. Last month, a group of them tried to rig the results. Since they advertised the attempt on Twitter, it was self-defeating. It was not difficult for us to work out which replies were theirs, close the survey, and discount these responses when we calculated the results.
We were able to make this assessment and shut the survey because the disruption took place three days after it was originally issued: the overwhelming majority of replies come in on day one or two. In essence, the interference took place after nearly all the responses from readers had been received. But the story might have had a less happy ending. Had the attempt taken place a few hours after the survey was issued, the findings would almost certainly have been invalidated.
Which got us thinking: how could this possibility be headed off – and the survey improved in the process?
We have decided to proceed as follows. The response rate to recent surveys has been high, as tends to be the case during election campaigns: for example, our general election survey in April gained 6244 responses, 1515 from readers declaring themselves to be party members.
So during the next few days, we will be e-mailing those 6000 or so people, plus others who have responded to recent surveys, to ask them if they want to reply to the survey each month by responding to an e-mail which we will send. This should provide a serviceable response rate. After all, polling companies often use samples of a thousand people to gauge the views of millions. We hope to get at least 750 resposes to gauge the views of about 150,000 party members.
It is important to note that the survey is not a scientific poll. The latter would be far more likely to pick up the view of relatively inactive party members. But over the years it has been very consistent – see here, here and here, for example. There is thus no reason to believe that a significant proportion of those who have declare themselves to be Party members are not so. They also tend to produce responses that are nearer the mainstream than those of our non-Party member readers.
All in all, we believe that the survey is a good rough guide to trends among those Party members – picking up last month, for example, a certain resistance to the main present future Party leader contenders. And from now on it will be drawing from our regular pool of such respondents, which will make it less vulnerable to fluctuations, and improve the consistency still further. Sooner or later this will need to be replenished, but that’s an issue for another day.