t*b 发帖数: 850 | 1 Pollsters developed all kinds of polling model or methods. Their goal is not
to make the pollling more accurate but easier for them to manipulate the
poll results.
For example, all polls left or right will interview citizens using a self-
developed population sampling theory. That is, if we poll 100 citizens, we
will have to poll x% registered democrats, y% republicans and (100-x-y)
independents. z% of the total interviewed has to be blacks, w% has to be
latino.....
Who determines x and y? the pollster themselves. DEM will say x is 38, GOP
will say x is 30. This swings the results dramatically. This is for national
poll.
For state poll, the errors are even bigger. we will have to poll a% in
Cleaveland area, b% in Columbus area, c% in eastern Ohio, d% in one county
and e% in another county. It is the same, the pollster decides these a, b, c
, d ,e....
These percentages are not based on actual population distribution, but based
on some skewed exit polls of one of the previous elections after revision.
polls also could pick people based on many other factors like these.
1) age, 2) color of skin, 3) education, 4) county or city, 5) religion
Basically any apparently ideology-leaning lolls are useless particulalry in
a close and polarized race like 2012. | f*********g 发帖数: 1637 | 2 You have made many poll prediction before, haven't you? Are they accurate? |
|