Wednesday 17 September 2008

Tories At Fifty-Two Per Cent?

If so, then fifty-two per cent of what?
Posted by at

8 comments:

  1. Anonymous17 September 2008 at 17:46

    52% of the adult population likely to vote, having weighted for political neutrality.

    So it doesn't mean 52% of the population. But it does act as a barometer for what is likely to happen. For example, MORI were 0.67% out at the GE05, exactly on for number of labour seats in GE05, and within 1% of the vote for London mayor in 2004.

    ReplyDelete
  2. David Lindsay17 September 2008 at 17:48

    Define "neutrality". A lot of dedicated non-voters are anything but "neutral". There are now huge numbers of disenfranchised people just waiting to be reached, but they are discounted by pollsters before headline figures are calculated.

    ReplyDelete
  3. Anonymous17 September 2008 at 17:56

    Polictically neutral means asking people on the basis of how the voted last time, which is not prfect but over a large enough data sample allows for a good indicator of the way people are likely to vote next time.

    Non voters are excluded yes - because all the evidence is that they are less likely to vote next time. Your argument (they're there if we just ccan attract them) is one that was used by the Liberal Democrat party, the Green party, the Referendum party, the natural Law party...

    The key question you have to answer is why, if polling companies are wrong to exclude or weight less the don't knows and won't votes, they consistently predict to a large degree of accuracy the actual votes cast on polling day?

    ReplyDelete
  4. David Lindsay17 September 2008 at 18:01

    Insofar as that is true at all, it is because they have done what opinion polls are there to do, which is not to measure public opinion, but to influence it.

    Also, note that most polling is now conducted by telephone. In other words, they know who they are asking.

    ReplyDelete
  5. Anonymous17 September 2008 at 18:09

    Sorry, you're wrong:

    "Insofar as that is true at all"

    It is true. I've provided you above with three examples that prove it, and there are hundreds of others which I could quote.

    "they have done what opinion polls are there to do, which is not to measure public opinion, but to influence it"

    Complete rubbish. If that was the case, why do polls move? If MORI, say, wanted a Tory win, why was it predicting a 100 seat Labour majority this time last year?

    "Also, note that most polling is now conducted by telephone. In other words, they know who they are asking"

    Yes, they do. Proper opinion polls then weight this accordingly to make sure they have a balanced sample representative of the population. This is all easily verifiable in their data tables which they produced, and also on their regulator's home page. They have to do this to be approved by the regulator.

    Unverified polls do exist (polls on blogs, say). No one takes them seriously becaus they are not balanced.


    I suspect this is another issue where your predjudices and your lack of detailed knowledge on this are coming to the fore. This really is a very regulated profession, and the results bear them out.

    ReplyDelete
  6. David Lindsay17 September 2008 at 18:12

    The results bear out that they are good at their job, that's all.

    Well, they sometimes do. Remember that hung Parliament in 1992?

    Still, I'm sure that things have moved on since then. These days, they just ring carefully selected people at home.

    ReplyDelete
  7. Anonymous17 September 2008 at 18:17

    Do you know *anything* about polls?

    The 1992 example is an interesting one - what happened was that the polls didn't pick up Tory voters. Of course, if we believe your example, and the polls influence pubhlic opinion, then that shouldn't have happened. Should it?

    "they ring carefully selected voters at home"

    Factually incorrrect. They use random number dialling. Then when people answer they ask them their details so they can weight appropriately. Its exactly the same thing as when people used to be asked on the street - they also asked background data, past voting etc that was needed to weight

    Just because you don't like what the polls tell you, don't attack them for being unscientific and flawed unless you have a background for criticising them methodologically. What is your level of statistical qualifications, by the way?

    ReplyDelete
  8. David Lindsay17 September 2008 at 18:47

    About as much as theirs, I'd say.

    No, of course they don't ring people at random! How stupid do you think that other readers are, to suggest such a thing?

    Dear readers, have you ever been polled? Do you know anyone who has been? Of course not. It's just the same carefully selected people over and over again.

    And even they, if they give the wrong answer - "I'm not voting for any of them, because none of them represents my views" - are simply discounted for headline purposes, leaving anyone who holds that view - never less than one third, and increasing around forty per cent, *even of those permitted to be polled* - to assume that they are the only one in the country.

    ReplyDelete

Subscribe to: Post Comments (Atom)