A quick reply to LKS’s recent post on the efficacy of online polls, here. My comment is reproduced here and I will be checking back for other comments on LKS’s post to update this post should it be required, just to keep the conversation going.

A short note on online-polls. I had said something about this earlier, here. I really do think that polling is a small concern, but an important one nonetheless depending on the uses to which the said poll may be put to. Garnering opinions is a tricky business, and there are so many ways to cast information in a negative or favourable light, depending on your party political affiliation, or even depending on your beliefs: it’s always a matter of interpretation, and the “opinion on the ground” such as it is, can become a political tool.

I do realize that small polls don’t represent much, and in the specific context of LKS’s poll, losing sleep over its accuracy or efficacy is pretty ridiculous: we have to be pragmatic and common sensical about things.

What’s important to me personally is that we hold ourselves to a higher standard of proof, and that we start demanding this of all our representatives: whether they are newspapers, government bodies or politicians. We need to know what’s at stake: a democracy of opinion.

My reply:

Actually, I think a happy compromise would be to make the poll more comprehensive. CNN conducts “Quick votes” wherein they state categorically that the results should not be taken as indicative of anything concrete.

These same websites (CNN, MSNBC too I believe, and the BBC) do make some attempts annually or every six months or so to conduct brief surveys on readership, requesting information on income levels, professions, locality and such. With clear indications of readership, they can then proceed to cast a more precise light on poll results.

A quick check with dictionary.com would provide a vivid definition of a ’straw poll’. That notwithstanding, I think it may be more useful to aim for a greater level of objectivity that would be helpful to readers.

The notion that blog polls provide a quick check with the ‘pulse of society’ should be disabused because all these polls do, on blogs or even websites which cater to political opinions of a specific cast, is take the pulse of its own *partisan readership*.

We could argue whether readers of your blog or any partisan blogs are unthinking partisans themselves or not, but the fact that there is no way to determine partisanship — not to mention income gaps, education levels, etc. etc. — leaves such polls open to the criticism nonetheless: it’s always been a matter of perception, sir.

It’s like taking a poll on issues where people who read your blog agree by default with your views — at least, that’s the kind of perception people can have.

To which I think a better solution would be the one described above: if we can have a comprehensive poll on the type of readership we have here, their political persuasion, their income-levels, their education levels, their locales, the amount of time they spend surfing the net, etc. etc., and then present this sort of information at the close of the poll would make it truly persuasive.

A short note on polling questions and such also: I’m no expert, and I think stats majors or even soci majors are able to better relate the kind of questions that would prove useful, but maybe we could try to improve the kind of responses we might want from polls:

For example, on polling about corruption, we could ask also where we think the government is failing in fighting corruption, or where government will is lacking in follow through, etc.; my point is specificity can be really helpful too.