The online Nielsen poll in today’s Fairfax papers has Labor ahead 58 to 42. Mention “online opinion poll” and a certain Disney cartoon character with big ears and red shorts (first name Mickey) comes to mind. And for good reason. Most, particularly of the “click here to tell us what do you think?” variety, are just fluff and entertainment.

But there are online polls and there are online polls. Britain’s Yougov has a very good record by all accounts.

The devil, as with all polling, is in the sampling. Allowing respondents to “self-select” — that is, with a general invitation to participate in this political survey if you want — is a no-no, not least because you’ll get people overly interested in politics.

The idea behind Nielsen’s online surveys, clearly explained in this short paper written in 2004, is to first recruit a pool of willing respondents who have some access to the internet. They don’t have to be particularly net savvy, and they might be surveyed on all sorts of topics (not just politics). Importantly, much of the recruiting is done offline, for example by telephone. In this way Nielsen has compiled some 90,000 names.

You then select a sample from that pool, and survey them. Of course, Nielsen’s 90,000 would still not be representative of the general population – young people would presumably be over-represented, oldies under-represented – so the sampling is twiddled to allow for this.

Broadly speaking, perhaps the assumption is that whichever type of person you are, being able to use the internet does not correlate with voting inclinations. That is, a pig-shooter or environmentalist who can use the net is no more likely to vote Labor, Liberal, National or Green than one who can’t. (Presumably Nielsen doesn’t weight to such detail.)

Another way of putting it: not everyone has a telephone, but pollsters reckon they can adjust their data to allow for this. It’s the same with online polling – only more so. Is online polling as “good” as telephone polling? Perhaps that’s like asking if telephone polling is “better” than face to face. Both have advantages. It depends how well you do it – but we can certainly expect to see more online polling in the future.

Here’s the latest poll-mix including the Nielsen online poll.