Do we believe pollsters are biased?

Discussion in 'West Mall' started by HornHawk, May 5, 2008.

  1. HornHawk

    HornHawk 250+ Posts

    I think the way pollsters phrase questions or the order in which certain questions are asked can lead to bias in the results, but does that mean the pollsters are biased? Take a look at some results from today on Real Clear Politics poll collection site:

    North Carolina Democratic Primary Insider Advantage Obama 48, Clinton 45 Obama +3.0

    North Carolina Democratic Primary Zogby Tracking Obama 48, Clinton 40 Obama +8.0

    North Carolina Democratic Primary PPP (D) Obama 53, Clinton 43 Obama +10.0

    Indiana Democratic Primary InsiderAdvantage Clinton 48, Obama 44 Clinton +4.0

    Indiana Democratic Primary Suffolk Clinton 49, Obama 43 Clinton +6.0

    Indiana Democratic Primary PPP (D) Clinton 51, Obama 46 Clinton +5.0

    Indiana Democratic Primary SurveyUSA Clinton 54, Obama 42 Clinton +12.0

    Indiana Democratic Primary Zogby Tracking Clinton 42, Obama 44 Obama +2.0

    Democratic Presidential Nomination Gallup Tracking Obama 50, Clinton 45 Obama +5.0

    Democratic Presidential Nomination Rasmussen Tracking Obama 46, Clinton 45 Obama +1.0

    Democratic Presidential Nomination USA Today/Gallup Obama 44, Clinton 51 Clinton +7.0

    Democratic Presidential Nomination CBS News/NY Times Obama 50, Clinton 38 Obama +12.0

    General Election: McCain vs. Clinton Rasmussen Tracking Clinton 47, McCain 43 Clinton +4.0

    General Election: McCain vs. Clinton USA Today/Gallup Clinton 46, McCain 49 McCain +3.0

    General Election: McCain vs. Clinton CBS News/NY Times Clinton 53, McCain 41 Clinton +12.0


    The Indiana poll difference between Zogby and Survey USA is really interesting. How can there be a 15 point differential? That spread is too large to be statistically insignificant. Same is true of USA Today vs. CBS/NY Times McCain v Clinton match-up. Again a 15 point spread. The only poll I seem to trust is Rasmussen, which seems to be the closest historically to calling the races correctly within the margin of error.
     
  2. eflow24

    eflow24 1,000+ Posts

    There is not doubt that polling is biased. I think both parties acknowledge that and accept it. That is why most of us always look at poll data with skepticism. They are about as reliable as exit polling.
     
  3. Kwisatz

    Kwisatz 500+ Posts

    I don't doubt that some are biased, but lets say you're a biased pollster. Do you overstate your candidates support to try and give the illusion of momentum prior to the election? Or do you understate it to help your candidate win the "expectations game"? (obviously this strategy only works with staggered primaries)

    Ever since the California primary I've been paying the most attention to Survey USA. They were the "outlier" in CA predicting a big win for Hillary and they were right. They have a very good track record since I've been paying attention.
     
  4. RollingwoodHorn

    RollingwoodHorn 500+ Posts

    I'm sure we can all agree that the way questions are asked can push results in one direction but that seems difficult to do unless it's being done fairly blatantly, when the question is simply who would you vote for. I mean, surely none of the pollsters are conducting surveys where the question is 'Are you planning on voting for that ***** Hillary or Obama' or 'Are you planning on voting for that Muslim Obama or Hillary'.
     
  5. HornHawk

    HornHawk 250+ Posts

    Push polling is done pretty blatantly by the individual candidate's pollsters. I think the more typical explanation with variations has to do with methodology and the sampling group, but that wouldn't explain 15 point swings because the margin of errors between any two sampling groups (if the polls are done scientifically) shouldn't be larger than six-seven points.
     
  6. BattleshipTexas

    BattleshipTexas 1,000+ Posts

    The typical way to influence a poll is by the series of questions immediately preceding it. For Obama you would ask whether his lead of delegates makes him the inevitable choice. For Hillary, you ask whether it is fair to exclude Florida and Michigan from such an important decision and whether, as a democrat, you trust your party leaders who make up the superdelegates.

    In Britain, at one time compulsory 2-year military service for 18 year olds was considered. Someone was able to show that asking:

    Do you think discipline is important for young men?
    Would following positive male role models be a good thing?
    Do you think teaching respect to young people is important?
    etc etc etc
    followed by
    Are you for compulsory service at age 18? This got a 75%--- "Yes" answer.


    Then they asked the same question to a similar group, but preceded it with:
    Should young men be given guns and taught to use them?
    Do you think teaching all of Britain's youth how to kill with a knife is a good thing?
    etc etc etc.
    Are you for compulsory service at age 18? This then got a 75%--- "No" answer.

    Even worse, they called the exact same people a period of time later, and were able to get them to answer the exact opposite, just by switching the lead up questions.
     
  7. RollingwoodHorn

    RollingwoodHorn 500+ Posts

    Thanks Battleship, I didn't even think about the questions they could ask preceding the 'who would vote for' question. I was just thinking about the question in a vacuum.
     
  8. Ag with kids

    Ag with kids 2,500+ Posts

    Dammit....

    I had a whole thing written up about how subtle bias is more powerful and then Battle supertrumps me...he's bloody well right...

    [​IMG]


    [​IMG]
     
  9. Joe Fan

    Joe Fan 10,000+ Posts

    Perhaps the worst thing about the modern political poll is the manner the questions are written. Too often it seems that they were drafted either by (a) ESL students, or (b) liberal arts school grads unable to express themselves in just one sentence.
     
  10. HoosierHorn

    HoosierHorn 500+ Posts

  11. UTViking

    UTViking 250+ Posts

    I used to get calls from pollsters, but I always had a hard time answering the yes/no questions.

    "Should we raise taxes to pay for X?"

    "It depends on which tax you want to raise: income, sales, estate, property, gas, alcohol, or tobacco."

    "Do you favor spending more on public transportation?"

    "What kind of public transportation? How much more? Would we have to raise taxes? Would we cut some other program to pay for it? Which program would we cut?"
     

Share This Page