After the American Century
In the aftermath of the election, one can judge how accurate public opinion polls were. Real Clear Politics developed a useful poll of the polls, giving an average figure that combined all the various efforts to quantify the public mood. This proved quite accurate. On the eve of the election, this average was 52.1% for Obama, who actually received slightly more, 52.6% of all votes cast. McCain was predicted to receive 44.5% but actually got 46.1%.
To put this another way, the poll of polls predicted a difference between the candidates of 7.6%. The actual difference was 6.5%. This could be considered an error of 1.1%, but some voters did change the minds during the last week of the contest, and this was a running average.
Which polling service did the best? None predicted a McCain victory. I hate to admit it, but FOX News predicted Obama by 7%. CNN also had Obama by 7%, while the PEW Trust concluded that the difference would be 6%. This suggests that however rabid the FOX news department may be, their polling experts do a good job. Less successful were Reuters and Gallup, as both predicted a massive Obama victory, with a margin of 11%. CBS News and ABC News gave Obama a 9% margin of victory - too much, but only 2.5% away from the correct result.
In terms of electoral votes, the polls also were pretty accurate. While John McCain tried hard to win in Pennsylvania, the polls kept saying that this was not going to happen. In fact, the race there was not particularly close. All the hype about the "Bradley Effect" proved mistaken. Likewise, the polls said Obama would likely win in Virginia, which he did, and that it was too close to call in North Carolina. In fact, Obama did win there, but by a margin of less than 1%. The polls were also right to say that Obama was leading in Florida, that he would take Colorado, New Mexico, and Nevada, and that the only swing state where McCain had a good chance was Missouri. In fact, we are still not absolutely certain he won there, but it appears that did did, by less than 8,000 votes. The poll of polls put McCain ahead by 0.1% - amazingly accurate.
On the whole, I suppose this is a good result, but one always hopes for a little bit of unpredictability to keep things interesting.
In the aftermath of the election, one can judge how accurate public opinion polls were. Real Clear Politics developed a useful poll of the polls, giving an average figure that combined all the various efforts to quantify the public mood. This proved quite accurate. On the eve of the election, this average was 52.1% for Obama, who actually received slightly more, 52.6% of all votes cast. McCain was predicted to receive 44.5% but actually got 46.1%.
To put this another way, the poll of polls predicted a difference between the candidates of 7.6%. The actual difference was 6.5%. This could be considered an error of 1.1%, but some voters did change the minds during the last week of the contest, and this was a running average.
Which polling service did the best? None predicted a McCain victory. I hate to admit it, but FOX News predicted Obama by 7%. CNN also had Obama by 7%, while the PEW Trust concluded that the difference would be 6%. This suggests that however rabid the FOX news department may be, their polling experts do a good job. Less successful were Reuters and Gallup, as both predicted a massive Obama victory, with a margin of 11%. CBS News and ABC News gave Obama a 9% margin of victory - too much, but only 2.5% away from the correct result.
In terms of electoral votes, the polls also were pretty accurate. While John McCain tried hard to win in Pennsylvania, the polls kept saying that this was not going to happen. In fact, the race there was not particularly close. All the hype about the "Bradley Effect" proved mistaken. Likewise, the polls said Obama would likely win in Virginia, which he did, and that it was too close to call in North Carolina. In fact, Obama did win there, but by a margin of less than 1%. The polls were also right to say that Obama was leading in Florida, that he would take Colorado, New Mexico, and Nevada, and that the only swing state where McCain had a good chance was Missouri. In fact, we are still not absolutely certain he won there, but it appears that did did, by less than 8,000 votes. The poll of polls put McCain ahead by 0.1% - amazingly accurate.
On the whole, I suppose this is a good result, but one always hopes for a little bit of unpredictability to keep things interesting.
