h2g2 The Hitchhiker's Guide to the Galaxy: Earth Edition

ok, so what do they reflect then? Surely if the public are voting in polls, then the results must be what the public think. Just don't get that one...

The readers of the Sun? Or the bloke they got chatting to on the tube on the way to the office?

Still the public though?

I first spotted it on the Beeb polls, but just seen it on Doggysnaps as well.

Results reflect the opinions of (i) people with internet access, (ii) people who can be bothered to phone a premium-rate phone line, (iii) people who give a toss about the question you've just asked.

Any or all of the above.

RF

But... all those people are members of the public?

In order to *accurately* reflect public opinion (which I think is the missing word in your title) you need to take a statistically representative sample of public opinion. Psephologists spend a lot of time and effort ensuring that the samples that they use for opinion polls are a true reflection of the population whose opinions they are trying to understand, so they ensure that the sample population accurately reflects age, gender, earnings, geographic location, and any number of other characteristics of their target audience.

RF

But a very select group. Like if you did a survey and only asked people who were attending a Star Trek convention dressed as Klingons, chances are it wouldn't be representative of the general population. (I'm not slagging Star trek fans - I *am* one.)

Clicked too quick!

I meant to add...you can't control those with an online poll that's just a yes/no or limited choice answer to a single question.

RF

Aha, so the wording of the 'warning' on all these polls is wrong! They are missing 'accurately'!

I knew I was right.

Hate yes/no quizzes. Though hate the strongly agree->strongly disagree ones more, unless they have a "don't care" option..

i think it is used to distinguish between polls where the voters come to the poller rather than the other way around. for obvious reasons there are huge problems with calling it public opinion in the former instance.

Well can't they just say that they asked readers of the Daily Mail or surveyed shoppers in Kensington High Street? That way at least we'd have an idea of who gave the responses.

I always thought they said that because a few years ago it was easier to vote multiple times on online polls. Likewise you could phone premium rate phoneline multiple times. If 99% of the poll were voted for by one person I don't think it would *accurately* reflect public opinion.

There's maybe something in that, but I thought it'd be easier now to block mass voting with cookies on your computer and recognising your phone number.

not really. if you ask a question like say, 'does god exist?', only lunatics from both sides will bother to respond to that by initiating a response to a poll. if you are asking the public at random you get the positions of the other 97% of the population who aren't lunatics as well. a LOT of issues that small groups of people feel extremely passionate about simply don't rate for interest to the majority. the majority might still have an opinion on the matter tho.

"I thought it'd be easier now to block mass voting with cookies on your computer "

Have you any idea how easy it is to delete your cookies?

Well yeah, it's easy, but it still takes time to vote, delete cookie, vote again, delete again.......

And yeah, depending on what you ask, those who feel most strongly are gonna be more likely to respond. But if you're going to approach people for a more general and random group you still have to be careful. Like if you go door-to-door on a Sunday morning, you'll miss everone who's at church. But going out on a friday or Sarurday night would loose you those who go to pubs and clubs.

They have a thing in the thelondonpaper (freebie evening paper in Nodnol - and yes, that's how they spell it) where they have a "guest" columnist (i.e. unpaid MOGP) sound off about something. They then ask us MOGPs to vote "more or bore" by texting a premium number.

I've notice that the following days results often look like this:

More 50% Bore 50%
More 66% Bore 33%
More 25% Bore 75%

See if you can guess how many people texted an answer!

RF

Hmmmm. Seems a little off. Could they not tweak it a little and make it 49.3% and 50.7% or something to make it a little more realistic?

well that would be outright lying. the suggestion is that, variously, 1, 2 or three people voted.

going to decimal places means thousands at least.

imho, given the insidiousness of push polling and other media manipulations, basic information about any poll should also be pulbished. the precise question asked, how many responses were involved etc.

i've actually had telephone pollsters start by asking my age and then telling me i don't 'qualify' for the rest of the poll. i have to wonder how idiotic the people who commissioned the poll are.