Discussion»Statements»Rosie's Corner» Maybe polls are rigged if you're winning and unrigged if you're not. How do we know which way a rigged poll swings?
Maybe polls are rigged if you're winning and unrigged if you're not. How do we know which way a rigged poll swings?
Is it only rigged(fake phony baloney) when you lose and always real true blue when you win? Does the poll cooperate and is it a complicit conspirator or is it an innocent bystander to the fraud?
All the polls are rigged because the rigged polls end up being best matching the final result at times. A fair poll would be to randomly contact 10000 or so people and ask who they would vote for but it has been found that doing such does not match the final result very well because of people not voting or changing their minds at the end. So poll have many adjustments and some of those adjustments mean that some groups are oversampled and undersampled because the poll wants to look at likely voters not just all possible voters and the poll wants to be rid of people who will change their mind. This is the rigging. There is also a problem of sample sizes being very small in many polls because that is fast and easy. Do you think asking 300 people would give an accurate view of what 300 million people are thinking? Of course not. In fact it can be seen that polls are flawed because they do not agree within their margin of error. If two polls are done at the same time with similar methods and do not agree within margin of error then those margins of error are too small. That is most likely because of sample sizes too small or the methods being too different. If you believe the poll you have to decide if you believe thing such as first-time voters are most likely to not vote and democrats and minorities will have much higher turnout than republicans and white people. Also you must believe that supporters of third party candidates will change their mind equally to support trump or clinton, even though libertarians are fare more likely to pick trump and the green party candidate has actually endorsed trump.
Here is an example Rosie...yesterday a poll came out that showed Hillary was up by 11% with Millenials....sounds impressive doesn't it? Not so impressive when it is revealed that only 27% of those polled were registered Republicans.
Can you give me an example of what the ideal poll would be like Cruiser? Should the sampling be made up 50% from each party? What about the sample size? Are there preferable sample sizes m'dear? For example would a poll consisting of 5000 people, 50% Republican and 50% Dems, be more likely substantive than a poll such as you describe? I think you can create/manipulate a poll that will give you whatever result you are looking for depending upon the configuration of it which means all polls could quite easily be skewed. AARRGGHH! :( Thank you for your reply Cruiser! :)
I think "ideal poll" would be pretty simple to achieve. Yes 50/50 sampling of Dem Rep would be best as well as equal number sampling across all age/income/race demographics...the bigger the sampling the better. To make it really legit, the polling should also be equally done by phone, in person and electronically as in online polling. Also the poll should be done by a neutral party/organization. I have not yet seen a poll that comes close to reflecting these specifications and doubt I will.
I find it somewhat amusing some of the answerers of this question are absolutely convinced they know more about polling than people/institutions who have been doing it for decades.
I submit there are a few possible explanations for this:
1) They really are polling geniuses who should quit wasting their time on Answermug and should go open their own polling firms to show everyone else how it's done.
2) They are ideologically motivated to disbelieve what polls tell them because they don't like what they are hearing
Benedict Arnold 