Keep in mind I’m talking AVERAGES (individual polls can be and sometimes were all over the place).
NATIONAL POPULAR VOTE: Obama by 7.6% / Obama by 6% (that could still spread some as remaining numbers from CA, WA, and OR are added in).
STATE POLLS: Let’s compare before and after, for the traditional “Battleground” states and a few (like GA) that had only recently moved close enough to be considered “Battleground”.
RealClearPolitics Projected (average of all polls 11-4-08) / Actual Results:PA: Obama by 7.3% / Obama by 10.3%.
NM: Obama by 7.3% / Obama by 14.7%.*
NV: Obama by 6.5% / Obama by 12.4%.*
CO: Obama by 5.5% / Obama by 6.6%.
VA: Obama by 4.5% / Obama by 4.5%.
OH: Obama by 2.5% / Obama by 4.2%.
FL: Obama by 1.8% / Obama by 2.5%.
NC: McCain by 0.4% / Obama by 0.2%.
IN: McCain by 1.4% / Obama by 0.9%.
MO: McCain by 0.7% / McCain by 0.2%.
MT: McCain by 3.8% / McCain by 3.4%.
GA: McCain by 4.0% / McCain by 5.3%.
* Okay, Obama "over-performed" a little in NM and NV. But, although there is no MOE for “averages”, most of the final numbers were WELL within “MOE” (from individual polls) of the immediate pre-election projection. The final RCP poll averages also “picked the winner” in every case, except two of the very closest toss-ups, IN & NC (and the latter is still TBD).
http://elections.nytimes.com/2008/results/president/map.htmlhttp://www.realclearpolitics.com/epolls/maps/obama_vs_mccain/?map=10If you break down the EV count, it, too, ended up
almost exactly as RCP predicted based on these polls. MSNBC currently has Obama 349-McCain 174, with NC still too close to call. If you look at RCP’s pre-election “no toss-ups” projection (333-200), and switch the razor-thin IN to Obama, and the razor-thin NC to the current toss-up, then they match almost perfectly (349-173)… the difference being one “partitioned” EV from either Maine or Nebraska (not sure if they’ve split that up in these results yet).
Electoral-Vote.com was very similar as well.
http://www.electoral-vote.com/evp2008/Pres/Maps/Nov04.h... Long story short: the polls immediately pre-election, when using a good average of all of them, were
VERY ACCURATE!No one (with any sense) ever said a single poll is reliable.
But the good news is that we CAN rely on the polls, when compiling all of them into an average.