Quote:
Originally Posted by Tyrone Slothrop
Most polling seems like an art of speculative content generation, not a science. The measurements they are taking are not falsifiable, so how can it be scientific? If a poll shows x +6% a month before Election Day, and x wins by 2%, how can anyone tell if the poll was right? If the final election results vary from the polls, it's attributed to things like ground games and weather and last-second decisionmaking, and this means that there is really no feedback mechanism to test a poll's accuracy.
There obvious is some use to tracking polls that apply the same methodology over times, in that they show shifts, but that value of those measurements is relative, not absolute.
This seems like something everyone knows on some level, but everyone wants the polls to be something they're not, so we collectively suspend disbelief.
|
These days, the polls are almost all within the margin of error. They literally mean nothing beyond it is going to be very close.
Or, as I am hoping, they are very wrong and it won't be.
