> “...converting a non-random sample into a representative sample through weighting of demographics.”
Impossible.
All pollsters are statistics flunkies. They are a joke to genuine statistical science.
Only a pollster could think a distorted sample could be converted to a representative sample via weighting.
Such pollsters would believe a cube could be converted to a sphere by downweighting its corners. And they would be foolish enough to believe their distorted cube conclusions are valid.
in the real world, we often work with approximations. As to where to draw the line, it depends on the cost. Perfection is simply too costly.
Having said this, polls are outside the margin of error more times than they should be, and the bias is against us. If we’re indicated to be down by a few points the weekend prior to the election, we actually have a good chance of winning.