In 2000, half a million more Americans voted for Al Gore than George W. Bush, yet through the math of the electoral college, Bush became president. Is this a likely occurrence in the event of a close national race? And furthermore, does this phenomenon just hurt Democrats? Are we doomed in close elections by the math of the electoral college, even if we narrowly win the popular vote?
First, a definition. A "tipping point state" is the closest state that a losing candidate needed to win to become president. In 2000 it was obviously Florida. In 2004, John Kerry just needed to do a little bit better in Ohio to have won the necessary 270 electoral votes. But what about in non-close elections? In 2008, Obama won the electoral vote 365-173. What if McCain had done just a bit better? We can assume McCain would have easily held Indiana and North Carolina, which Obama just barely won. That would not have given him the presidency though. Even giving McCain the big prize states of Ohio and Florida (won with about 5 and 3% respectively) still doesn't put him over the edge. Gradually shifting every state towards McCain by margin of victory doesn't make him president until Colorado's electoral votes flip red. So had 2008 been a close election, Colorado should have been the tipping point state...give McCain that state and every closer state, and McCain wins.
How difficult would that have been for McCain to do? Well, Obama won Colorado by 9%, a significant margin, and even greater than Obama's national popular vote victory of 7.2%. If McCain had improved nationally at equal levels by state, it is conceivable that the Arizona Senator could have narrowly lost Colorado (and the electoral college) while winning the popular vote by almost 1.8% (9 - 7.2)! A big caveat to this analysis however: it is not expected that a candidate would have improved by exactly equal margins nationwide. Perhaps a more popular McCain would have made up more votes in states like Arizona and Michigan, while he still would have gotten crushed in New York by only a slightly smaller margin. So tipping point state analysis in such a case can only ever be a theoretical approximation of what a closer election would have looked like.
Still, let us run with this assumption. In that case, the electoral college actually worked to Obama's advantage in 2008, as his 9% margin in Colorado gave him a larger electoral vote cushion than his 7.2% national margin. We'll call that a 1.8% EV advantage, and this will be shown to be fairly high compared to recent elections.
As a close election, 2004 is easier to analyze. Kerry lost Ohio by 2.1%, and nationally by 2.4%. So an electoral college victory was actually slightly easier than flipping the national vote margin. He gets a 0.3% EV advantage, though this is just barely an advantage at all.
2000 is also simple. Gore loses Florida by effectively 0%, and wins nationally by 0.5%. He suffers a 0.5% EV disadvantage, costing him the presidency.
What follows is the same math done for every presidential election going back to 1916, flipping every close state to the loser one by one by winning margin until achieving electoral college victory. However, pretty much everything before 1960 is more for fun than for meaningful insight, as the parties and their geographic bases were quite different before then. For convenience, positive numbers mean Democratic victories and margins, while negative numbers mean Republican victories and margins.
Year |
Tipping Point State |
State Margin |
National Margin |
EV Advantage |
2008 |
Colorado |
9 |
7.2 |
1.8 |
2004 |
Ohio |
-2.1 |
-2.4 |
0.3 |
2000 |
Florida |
0 |
0.5 |
-0.5 |
1996 |
Pennsylvania |
9.2 |
8.5 |
0.7 |
1992 |
Tennessee |
4.7 |
5.5 |
-0.8 |
1988 |
Michigan |
-7.9 |
-7.7 |
-0.2 |
1984 |
Michigan |
-19 |
-18.2 |
-0.8 |
1980 |
Illinois |
-7.9 |
-9.7 |
1.8 |
1976 |
Wisconsin |
1.7 |
2.1 |
-0.4 |
1972 |
Ohio |
-21.6 |
-23.2 |
1.6 |
1968 |
Illinois |
-2.9 |
-0.7 |
-2.2 |
1964 |
Iowa |
24 |
22.6 |
1.4 |
1960 |
New Jersey |
0.8 |
0.2 |
0.6 |
1956 |
New Mexico |
-16 |
-15.4 |
-0.6 |
1952 |
Maryland |
-11.5 |
-10.9 |
-0.6 |
1948 |
Illinois |
0.8 |
4.5 |
-3.7 |
1944 |
New York |
5 |
7.5 |
-2.5 |
1940 |
Pennsylvania |
6.9 |
10 |
-3.1 |
1936 |
Ohio |
20.6 |
24.4 |
-3.8 |
1932 |
Iowa |
17.7 |
17.7 |
0 |
1928 |
Illinois |
-14.7 |
-17.4 |
2.7 |
1924 |
Illinois |
-25.5 |
-25.2 |
-0.3 |
1920 |
Connecticut |
-29.7 |
-26.2 |
-3.5 |
1916 |
California |
0.3 |
3.1 |
-2.8 |
Conclusions:
1) In modern politics, there is no statistical EV advantage to either party. It simply fluctuates back and forth, even for an incumbent seeking re-election.
2) The electoral college advantage is small, typically less than 1%. Unless the general election is within a percent (itself a rare event), we shouldn't even worry about the electoral college overturning the results of the popular vote.
3) Prior to 1952, a Democrat could expect to rack up ridiculous margins in the South, distorting the national popular vote in their favor (and consequentially, giving the GOP an electoral vote advantage compared to popular vote, since all those extra Dem votes in the South didn't deliver any additional states). It must be pointed out that this was the era of literacy tests, poll taxes, and KKK thugs hanging around the voting booths. South Carolina "voted" over 98% for FDR twice.
4) Historically, the tipping point state tended to be midwestern or northeastern by location...this can be partly attributed to their larger electoral vote shares of the past, but I think more to just how swingy the now-Democratic region was.
Some other thoughts...claims that Perot handed the presidency to Clinton are common, yet consider this further evidence to the contrary. Bush Sr. would have needed to win all the way up to Tennessee to be re-elected, a state in which Perot ran quite weakly (the South has never taken to independent candidates). Clinton won Tennessee 47% to Bush's 42.4%, and it is doubtful that enough Perot voters would have A) still voted in his absence, and B) picked Bush Sr. over Clinton to tip that state the other way.
As the South used to inflate Dem totals, it apparently did the same for Nixon in 1972, causing that divergence. In 1964, Goldwater also appears to have narrowed the expected national vote margin somewhat through his overperformance in the South. Basically, blame the South's monolithic voting habits for causing most of the large EV advantages.
Finally, back to 2008. What explains this relatively large divergence? It seems to me that Obama was uniquely strong in a number of swing states, while scoring more typical numbers elsewhere in the country. Had Kerry beaten Bush by 7.2%, as Obama did to McCain, I doubt the Massachusetts Senator would have scored such large victories in states like Virginia and Colorado, though he probably would have done better in the Deep South and Appalachia.
So what can we expect for 2012? National polling has been all over the map, showing everything from large Obama leads to small Romney leads. Yet one state poll after another, especially in crucial swing states like Colorado, Virginia, and Ohio, shows Obama in a commanding position. Could we see an even larger EV advantage for Obama this fall? Or will it reverse yet again and give Romney the potential of repeating Bush's 2000 electoral college reversal of the popular vote?