35 ms·

Too bad polls haven't been accurate for a while now.

samdoidge 10y ago
As someone who made money betting on a Trump victory - certain topics / polls. Calling Brexit / Trump voters racist leads to a certain % of the population not disclosing their vote, oversampling of Democrats was another issue in the US polls.
- ZeroGravitas 10y ago
  That not what oversampling means. http://www.pewresearch.org/fact-tank/2016/10/25/oversampling-is-used-to-study-small-groups-not-bias-poll-results/ http://www.pewresearch.org/fact-tank/2016/10/25/oversampling...
  - samdoidge 10y ago
    I'm aware of what oversampling means. I was listing the 'hidden vote' and oversampling as a reason.
    - ZeroGravitas 10y ago
      That still makes no sense.
      - samdoidge 10y ago
        Can you elaborate on what doesn't make sense?
        ZeroGravitas 10y ago
        Oversampling is a standard thing to do with polling. There's a confused conspiracy theory that it means "adding extra votes to the Democrat column". This doesn't even stand up to the most cursory of examinations, as if that was really happening then people would say "they're intentionally fudging the results to favor the Democrats" and kick up a fuss about it. Instead they say "oversampling Democrats", in the hope that people who don't know what that is will think it sounds like cheating while not actually looking like idiots to people who do know what it means. A classic dogwhistle approach to say two things to two audiences. You claim to know what it is, yet you say it's an "issue" (which it isn't) and you say that at the end of a sentence claiming that there's a "shy voter" factor leading to an undercount of Trump support. Oversampling has no logical connection here unless you mean it in the "Democrats are cheating in the polls" meaning. You claim that you mean it in the actual statistical meaning. The two obvious explanations are that you don't know what it is and have been fooled by all the chatter around it, or that you're trying to fool others despite knowing it's nonsense. I think you don't know what it is, that's why I provided the explanatory link a couple of comment above. Also, the polls did not materially undercount Trump voters. The polls were more accurate than at the last election. Some people badly mapped the accurately predicted 3 million popular vote win to electoral college results, but that's a different thing than polls being bad, Trump voters not wanting to admit who they were voting for, or being intentionally skewed. On a state by state basis polls were wrong in both directions, for various reasons, but that cancelled out overall.
        samdoidge 9y ago
        Ok, so when you were telling me 'that's not what oversampling means', you meant 'I disagree with what you are saying'. I would be more inclined to debate if you approached the disagreement in a more open fashion. Your partisan paragraphs further make me think I will be wasting my time here, so let's just say we have different views on the polls :).
- JCzynski 10y ago
  Polls were accurate in both those cases. Media coverage of them was wrong, but the polls were right.
anoonmoose 10y ago
Brexit was too close to call and Hillary won the popular vote so it ain't like things in the polling world are all that wild.
pmyteh 10y ago
Polling is hard; there are lots of ways for polls to be inaccurate (see the British Polling Council postmortem of the last general election for the gory details). Those errors don't add up to 20 points. And I say that as a card-carrying member of the Labour party.
easytiger 10y ago
That's not true. Last Uk election was very close and that was reflected in the polls. Same with brexit.

As someone who made money betting on a Trump victory - certain topics / polls. Calling Brexit / Trump voters racist leads to a certain % of the population not disclosing their vote, oversampling of Democrats was another issue in the US polls.

ZeroGravitas 10y ago
That not what oversampling means. http://www.pewresearch.org/fact-tank/2016/10/25/oversampling-is-used-to-study-small-groups-not-bias-poll-results/ http://www.pewresearch.org/fact-tank/2016/10/25/oversampling...
- samdoidge 10y ago
  I'm aware of what oversampling means. I was listing the 'hidden vote' and oversampling as a reason.
  - ZeroGravitas 10y ago
    That still makes no sense.
    - samdoidge 10y ago
      Can you elaborate on what doesn't make sense?
      - ZeroGravitas 10y ago
        Oversampling is a standard thing to do with polling. There's a confused conspiracy theory that it means "adding extra votes to the Democrat column". This doesn't even stand up to the most cursory of examinations, as if that was really happening then people would say "they're intentionally fudging the results to favor the Democrats" and kick up a fuss about it. Instead they say "oversampling Democrats", in the hope that people who don't know what that is will think it sounds like cheating while not actually looking like idiots to people who do know what it means. A classic dogwhistle approach to say two things to two audiences. You claim to know what it is, yet you say it's an "issue" (which it isn't) and you say that at the end of a sentence claiming that there's a "shy voter" factor leading to an undercount of Trump support. Oversampling has no logical connection here unless you mean it in the "Democrats are cheating in the polls" meaning. You claim that you mean it in the actual statistical meaning. The two obvious explanations are that you don't know what it is and have been fooled by all the chatter around it, or that you're trying to fool others despite knowing it's nonsense. I think you don't know what it is, that's why I provided the explanatory link a couple of comment above. Also, the polls did not materially undercount Trump voters. The polls were more accurate than at the last election. Some people badly mapped the accurately predicted 3 million popular vote win to electoral college results, but that's a different thing than polls being bad, Trump voters not wanting to admit who they were voting for, or being intentionally skewed. On a state by state basis polls were wrong in both directions, for various reasons, but that cancelled out overall.
        samdoidge 9y ago
        Ok, so when you were telling me 'that's not what oversampling means', you meant 'I disagree with what you are saying'. I would be more inclined to debate if you approached the disagreement in a more open fashion. Your partisan paragraphs further make me think I will be wasting my time here, so let's just say we have different views on the polls :).
JCzynski 10y ago
Polls were accurate in both those cases. Media coverage of them was wrong, but the polls were right.

anoonmoose 10y ago

Brexit was too close to call and Hillary won the popular vote so it ain't like things in the polling world are all that wild.

pmyteh 10y ago

Polling is hard; there are lots of ways for polls to be inaccurate (see the British Polling Council postmortem of the last general election for the gory details). Those errors don't add up to 20 points. And I say that as a card-carrying member of the Labour party.

easytiger 10y ago

That's not true. Last Uk election was very close and that was reflected in the polls. Same with brexit.