With the Scottish result on a knife-edge, the spotlight falls on the undecided

http://www.theguardian.com/commentisfree/2014/sep/15/scottish-independence-vote-dont-knows

Version 0 of 1.

There has been much discussion about the true number of “don’t knows” in opinion polls over the past week or so. Recent online surveys (including YouGov for the Times and Sun and ICM for the Sunday Telegraph) have put them at 6-9%. This compares with 18% in a separate ICM telephone poll last week for the Guardian, and 23% in a face-to-face poll by TNS. What explains the difference?

There are five kinds of “don’t know”; the trouble is, we can’t be certain how many people belong to each group.

1. People who, through circumstance, culture, language or lack of interest, seldom, if ever, vote or answer opinion polls. They are completely hidden from view, but their views, or lack of them, make no difference to the outcome of any election or referendum.

2. People who seldom, if ever, vote, and don’t join online panels such as YouGov’s, but will talk to telephone or face-to-face pollsters. In practice, they have little effect on who wins any election or referendum.

3. People who are willing to join online panels but won’t talk to pollsters who ring them up, knock on their door or stop them in the street. They are happy to give their views anonymously, but not to a stranger. (This group includes both “don’t knows” and people who do have clear views and are willing to express them online.)

4. People who are willing to give their views both online and to a stranger, but react in different ways. A telephone or face-to-face poll may catch them at a moment when their minds are on other things. They may say “don’t know” with complete honesty to a question put to them out of the blue. When they complete an online survey, they do so at a time of their choosing. If they face a question to which they need to give some thought, they can pause and think about it, and then give their answer. They are under no pressure to respond immediately.

5. Other people will say what they truly think online but don’t want to offend strangers. Political scientists have a clumsy term for the phenomenon of people who seek to avoid offence to telephone or face-to-face polling field-workers: social satisfying. It’s possible that some of the recent TNS and ICM/Guardian “don’t knows” are people who know perfectly well how they will vote, but would rather avoid taking sides on such a divisive issue as independence when asked their views by someone they have never met.

There is yet another group that must be acknowledged: people with clear views and definite voting intentions who insist on the secrecy of the ballot and never talk to pollsters, whether online, by telephone or face-to-face. If this group tilts strongly one way or another, it can make a mockery of all polling results.

In the final analysis, all pollsters assemble the best sample they can, and seek to extrapolate as accurately as possible from the people they reach to the voting population as a whole. It is not, and never will be, an exact science. We all rely on our judgment. And in a close race, quite separate from the factors listed above, we always risk being caught out by random sampling fluctuations: none of us can repeal the laws of probability. That is why our final survey, for Thursday’s Sun and Times, will have a much larger sample than normal. This reduces the risk of random error.

Each of us will know on Friday whether or not our judgments have been vindicated.