Entry tags:
That Whole "Race" Thing
Disclaimer: I don't believe in "race", but I do have to acknowledge the existance of racism. Because of this, I am prone to defining "racism" as "the belief in 'race'", which leads me to make phrases that the population at large does not make.
Now, there is a wide-spread and recognized phenomenon about pre-election polling concerning race (which, from this point on, I will type without the quotes), especially with Democrats (since Reagan made it okay to be a Republican in the South): Democrat voters will tell pollsters that they will vote for a black candidate, but when they actually get into the voting booth, they don't. This has happened to gubernatorial candidates, congressional candidates, and (thinking mostly of Jesse Jackson) presidential candidates.
Usually, when we see a significant difference between pre-election poll numbers and the election results themselves, the first thing to do is recount the election results. However, in the case of black candidates, the election results don't tend to change all that much from the initial count. We therefore look at the polling methods from before the elections, and ask "How did the pollsters get it so wrong?".
This year, thanks to Zogby and Rasmussen using automated telephone polling machines, and Suffolk and other university-based pollsters still using human callers, we may have a clue. The polling outfits which used machines which sounded like machines on the telephone predicted Clinton victories in California, New York, and Massachusetts by ratios insignificantly different from the actual results. Polling outfits who used human volunteers predicted much better results for Obama than he actually got.
We still have the problem that Democrats do not wish to be perceived by humans as being racist in their choice of candidate (including the meta-racism of, "Well, I would like to see Obama as president, but there are too many racists who won't vote for him, so I'll vote for the shrieking shrew instead"). But when those Democrats being polled do not believe they're talking to a human, they'll answer the way they'll actually vote.
I don't know what to do with this information, except to recommend that the Obama campaign use mechanically-voiced polls if they want accurate results.
Now, there is a wide-spread and recognized phenomenon about pre-election polling concerning race (which, from this point on, I will type without the quotes), especially with Democrats (since Reagan made it okay to be a Republican in the South): Democrat voters will tell pollsters that they will vote for a black candidate, but when they actually get into the voting booth, they don't. This has happened to gubernatorial candidates, congressional candidates, and (thinking mostly of Jesse Jackson) presidential candidates.
Usually, when we see a significant difference between pre-election poll numbers and the election results themselves, the first thing to do is recount the election results. However, in the case of black candidates, the election results don't tend to change all that much from the initial count. We therefore look at the polling methods from before the elections, and ask "How did the pollsters get it so wrong?".
This year, thanks to Zogby and Rasmussen using automated telephone polling machines, and Suffolk and other university-based pollsters still using human callers, we may have a clue. The polling outfits which used machines which sounded like machines on the telephone predicted Clinton victories in California, New York, and Massachusetts by ratios insignificantly different from the actual results. Polling outfits who used human volunteers predicted much better results for Obama than he actually got.
We still have the problem that Democrats do not wish to be perceived by humans as being racist in their choice of candidate (including the meta-racism of, "Well, I would like to see Obama as president, but there are too many racists who won't vote for him, so I'll vote for the shrieking shrew instead"). But when those Democrats being polled do not believe they're talking to a human, they'll answer the way they'll actually vote.
I don't know what to do with this information, except to recommend that the Obama campaign use mechanically-voiced polls if they want accurate results.