Originally posted by carpedm9587
View Post
First of all - thanks for correcting my "two"
Second - much as I would like to deny that such is occurring, the range of polling is suspicious, especially when left-leaning sources have polls that lean left and right-leaning sources have polls that tend to lean right. Rasmussen is a good case in point - being off the "aggregate" to the right almost all the time. Quinnipiac is off the aggregate to the left almost all the time.
But there is another dynamic at play: the polling methodology. There is a clear difference between robo-polls and people-polls. There is a clear difference between polls that include cell phones and polls that exclude them. These are comparatively new dynamics and the pollsters do not yet seem to know how to adjust for them.
That is one of the reasons I tend to prefer using aggregators like FiveThirtyEight and RealClearPolitics. Their aggregate numbers, historically, have been some of the most accurate numbers out there. It's odd to me; it would seem to me the aggregate of flawed polls would be a flawed aggregate - but that's the lovely world of statistics for ya!
Second - much as I would like to deny that such is occurring, the range of polling is suspicious, especially when left-leaning sources have polls that lean left and right-leaning sources have polls that tend to lean right. Rasmussen is a good case in point - being off the "aggregate" to the right almost all the time. Quinnipiac is off the aggregate to the left almost all the time.
But there is another dynamic at play: the polling methodology. There is a clear difference between robo-polls and people-polls. There is a clear difference between polls that include cell phones and polls that exclude them. These are comparatively new dynamics and the pollsters do not yet seem to know how to adjust for them.
That is one of the reasons I tend to prefer using aggregators like FiveThirtyEight and RealClearPolitics. Their aggregate numbers, historically, have been some of the most accurate numbers out there. It's odd to me; it would seem to me the aggregate of flawed polls would be a flawed aggregate - but that's the lovely world of statistics for ya!
Comment