The following is an extended guest post from John Koza:
Some have suggested that presidential elections would be improved if every state passed a law dividing its electoral votes in proportion to each candidate’s share of the popular vote in the state.
When you analyze the actual numbers, you find that a proportional system would work very differently than one would expect:
- Although it might appear that the proportional system would give presidential candidates a reason to campaign in all 50 states, candidates would only campaign in about half the states.
- In the 26 or so states where candidates would actually campaign, the proportional system would almost always be a battle for just one electoral vote—thus giving a small state as much influence as a big state.
- The proportional system would not necessarily deliver victory to the candidate receiving the most popular votes nationwide (and would not have done so in the most recent close election, namely 2000).
- Enactment of the proportional system on a state-by-state basis would be a self-arresting process because each enacting state would increase the influence of the remaining winner-take-all states.
The reason that presidential candidates would campaign in only about 26 states arises from three facts.
First, absent a federal constitutional amendment abolishing the office of (human) presidential elector, a state’s electoral votes can only be cast in whole-number increments.
Second, the average number of electoral votes per state is about 10 (538 divided by 51). Thus, in an average-sized state, one electoral vote would correspond to a 10%-share of the state’s popular vote. Moreover, one electoral vote would correspond to considerably more than 10% of the popular vote in the 33 states with a below-average number of electoral votes. In the eight smallest states (with 3 electoral votes), one electoral vote would correspond to a 33.33%-share of the popular vote.
candidates under the proportional system, first consider the eight states with three electoral votes. If a candidate received between 16.7% and 50.00% of the popular vote in those states, the candidate would receive one electoral vote under the proportional plan. If a candidate received between 50.01% and 83.3% of the popular vote, then the candidate would receive two electoral votes. A candidate with Obama’s 29% level of support in Wyoming in 2012 would first consider whether he had any chance of gaining or losing anything by campaigning in the state. Gaining one electoral vote would have required Obama to perform the monumental task of increasing his support by 21 percentage points (in order to reach the “breakpoint” of 50% where he would win two electoral votes, instead of just one). Romney could have taken away Obama’s one electoral vote only by performing the monumental task of cutting Obama’s support by 12.3 percentage points (below the breakpoint of 16.7%). Because Obama’s level of support of 29% was so distant from the “breakpoints” of 50% and 16.67%, both candidates would have quickly concluded that they had nothing to gain or lose by campaigning in the state. Hence, neither would campaign in the state.
In the real world, actual presidential candidates (wisely) campaign only in states where they are within three percentage points of winning something. For example, in 2012, 100% of the 253 general-election campaign events (and virtually all the ad spending) occurred in the 12 “battleground” states where Romney’s support was within three percentage points of 48% (Romney’s eventual national level of support). There were no general-election campaign events in the 38 remaining states. Candidates simply do not campaign in a state unless they have something to gain or lose.
| Romney % | 2012 general-election campaign events (out of 253) | State | Ad spending |
| 51% | 3 | North Carolina | $80,000,000 |
| 50% | 40 | Florida | $175,776,780 |
| 48% | 73 | Ohio | $148,000,000 |
| 48% | 36 | Virginia | $127,000,000 |
| 47% | 23 | Colorado | $71,000,000 |
| 47% | 27 | Iowa | $52,194,330 |
| 47% | 13 | Nevada | $55,000,000 |
| 47% | 13 | New Hampshire | $34,000,000 |
| 47% | 5 | Pennsylvania | $31,000,000 |
| 47% | 18 | Wisconsin | $40,000,000 |
| 46% | 1 | Minnesota | $0 |
| 45% | 1 | Michigan | $15,186,750 |
In fact, 98% of the campaign events (249 of 253) were concentrated in the states within two percentage points; and 82% of the campaign events (208 of 253) were concentrated in the states within one percentage point. See the full 2012 spreadsheet for more information.
Now consider what would happen when real-world political considerations are applied to the proportional method of awarding electoral votes.
In the eight states with three electoral votes (Wyoming, North Dakota, South Dakota, Alaska, Montana, Delaware, Vermont, and the District of Columbia), Obama was not within three percentage points of any of the breakpoints (at 16.7%, 50%, and 83.3%) that would have caused him to gain or lose an electoral vote in 2012. Accordingly, Obama (and hence Romney) would have ignored all eight states.
In the five states with four electoral votes (Idaho, New Hampshire, Maine, Rhode Island, and Hawaii), Obama’s level of support was within three percentage points of a breakpoint in only one state (Rhode Island). Obama’s level of support in Rhode Island was 64%, which was reasonably close to the breakpoint of 62.50% between winning two versus three electoral votes. Hence, Romney would have campaigned vigorously to push Obama below 62.50%, and Obama would have fought hard to stay above 62.50%. Meanwhile, both candidates would have ignored Idaho, New Hampshire, Maine, and Hawaii.
In the three states with five electoral votes (West Virginia, Nebraska, and New Mexico), Obama’s level of support was not within three percentage points of a breakpoint enabling him to gain or lose an electoral vote. Thus, neither Obama nor Romney would have campaigned in any of these states.
When the 25 least populous states (seven electoral votes or less) are analyzed in relation to their respective “breakpoints,” only five would have been “battleground” states under the proportional system in 2012—Rhode Island, Kansas, Mississippi, Utah, and Oklahoma. The other 21 states would have been politically irrelevant “spectator” states.
When all 50 states and DC are similarly analyzed in relation to their respective “breakpoints,” there would have been only 26 states inside the three-percentage-point window that would attracted the presidential candidates. The table shows the 26 “battleground” states arranged in order of the percentage change (column 1) needed to gain or lose one electoral vote. Because Rhode Island was discussed earlier, it is high-lighted below.
| Change needed to gain or lose 1 EV | State | EV | 2012 D percent | Breakpoint just below D percent | Breakpoint just above D percent |
| 0.11% | Alabama | 9 | 38.78% | 27.78% | 38.89% |
| 0.22% | Missouri | 10 | 45.22% | 45.00% | 55.00% |
| 0.23% | Pennsylvania | 20 | 52.73% | 52.50% | 57.50% |
| 0.37% | Utah | 6 | 25.37% | 25.00% | 41.67% |
| 0.44% | Florida | 29 | 50.44% | 50.00% | 53.45% |
| 0.48% | New York | 29 | 64.28% | 63.79% | 67.24% |
| 0.83% | Georgia | 16 | 46.04% | 40.63% | 46.88% |
| 0.85% | California | 55 | 61.87% | 60.91% | 62.73% |
| 1.03% | North Carolina | 15 | 48.97% | 43.33% | 50.00% |
| 1.06% | Minnesota | 10 | 53.94% | 45.00% | 55.00% |
| 1.08% | Illinois | 20 | 58.58% | 57.50% | 62.50% |
| 1.20% | Texas | 38 | 41.99% | 40.79% | 43.42% |
| 1.26% | Tennessee | 11 | 39.65% | 31.82% | 40.91% |
| 1.26% | Ohio | 18 | 51.51% | 47.22% | 52.78% |
| 1.52% | Rhode Island | 4 | 64.02% | 62.50% | 87.50% |
| 1.54% | Wisconsin | 10 | 53.46% | 45.00% | 55.00% |
| 1.68% | Michigan | 16 | 54.80% | 53.13% | 59.38% |
| 1.68% | Maryland | 10 | 63.32% | 55.00% | 65.00% |
| 1.76% | New Jersey | 14 | 58.95% | 53.57% | 60.71% |
| 1.97% | Virginia | 13 | 51.97% | 50.00% | 57.69% |
| 2.49% | Oklahoma | 7 | 33.23% | 21.43% | 35.71% |
| 2.50% | Louisiana | 8 | 41.25% | 31.25% | 43.75% |
| 2.53% | Mississippi | 6 | 44.20% | 41.67% | 58.33% |
| 2.69% | Massachusetts | 11 | 61.79% | 59.09% | 68.18% |
| 2.75% | Colorado | 9 | 52.75% | 50.00% | 61.11% |
| 2.78% | Kansas | 6 | 38.89% | 25.00% | 41.67% |
See the full proportional 2012 spreadsheet for information on all the states.
Similar spreadsheets for 2008, 2004, and 2000 show that only about half the states would have been politically relevant in those years under the proportional system.
Another deficiency of the proportional system arises from the fact that it would, in practice, be a “one-state-one-vote” system in 25 of the 26 states in the above table. It would be a “one-state two-vote” system in California, where an electoral vote corresponds to 1.82% of the popular vote. That is, only 27 electoral votes (out of 538) would be in play under the proportional system.
In a “one-state-one-vote” system, a win in a small state would be as rewarding as a win in a big state. Shifting 0.37% of Utah’s popular vote (about 3,600 votes) would be as rewarding as shifting 0.23% of Pennsylvania’s popular vote (about 13,000 votes) or shifting 0.44% of Florida’s popular vote (37,000 votes). However, it is considerably less expensive to shift 3,600 votes in Utah than to shift 13,000 or 37,000 votes. Consequently, candidates would almost certainly ignore all of the high-population states, such as Texas, Florida, New York, Illinois, Pennsylvania, Michigan, New Jersey, and Ohio (and probably also California).
Another deficiency of the proportional approach is that it would not necessarily deliver victory to the candidate receiving the most popular votes nationwide. If this approach had been in use in every state in the most recent close presidential election (2000), the result would have been a 269–269 tie in the electoral vote, even though Al Gore led by 537,179 popular votes nationwide. See table 4.21 in our book Every Vote Equal: A State-Based Plan for Electing the President by National Popular Vote—downloadable for free at www.NationalPopularVote.com. The election would have been thrown into the U.S. House of Representatives which (based on its partisan makeup at the time) would have elected George W. Bush.
Finally, there are two practical political impediments associated with getting states to enact the the proportional system on a state-by-state basis.
First, the only states that would be likely early adopters of the proportional system would be the relative handful of states where the governor and legislature are controlled by a political party different from the party that regularly wins the state in presidential elections. Thus, bills have been filed in recent years by Republican legislators in Michigan to proportionally divide the state’s usually Democratic bloc of electoral votes.
Second, if the proportional system advanced beyond the early-adopter stage, another impediment would emerge. Each additional state adopting the proportional approach would increase the political clout of all the remaining winner-take-all states—thereby making adoption increasingly less attractive to the remaining states. If, every state except Florida (with 29 electoral votes) adopted the proportional approach, Florida would have an overwhelming incentive to retain its existing winner-take-all law. With only 27 electoral votes at stake in the 26 “battleground” states, Florida’s 29 electoral votes would thus single-handedly control the national outcome. Thus, the process of adopting the whole-number proportional approach on a state-by-state basis would be self-arresting.
Of course, the whole-number proportional approach could be installed on a simultaneous nationwide basis if it were enacted in the form of a federal constitutional amendment. However, once we start talking about constitutional amendments, Senator Cannon’s 1969 fractional proportional approach would be a far better amendment than the whole-number proportional approach discussed above.
Cannon’s proposed fractional proportional approach would have eliminated the position of (human) presidential elector, thereby permitting the proportional division of each state’s electoral votes to be carried out to 3 decimal places. In the nation’s biggest state (California), 0.001 of an electoral vote would correspond to about 230 popular votes, so that campaigning would be rewarding to candidates in California (and every other state). Thus, the fractional proportional approach would succeed in making every voter, in every state, politically relevant. In other words, the fractional proportional approach would eliminate the major defect of both the current state-by-state winner-take-all approach and the whole-number proportional approach—namely that presidential candidates have no reason to solicit votes in large numbers of states.
However, the fractional proportional approach would not make every vote equal throughout the United States, and it would not guarantee victory to the winner of the nationwide popular vote. If the fractional proportional approach had been used in 2000, it would have resulted in the election of George W. Bush, despite Al Gore’s 537,000-vote lead in the national popular vote (as shown in tables 3.1 and 3.2 of the book Every Vote Equal: A State-Based Plan for Electing the President by National Popular Vote).
In summary, the whole-number proportional method of dividing a state’s electoral votes would not make voters in every state matter; it would not make every vote equal; and it would not guarantee the Presidency to the candidate receiving the most popular votes nationwide.
REFERENCES:
http://nationalpopularvote.com/sites/default/files/prez-2000-proportional.xlsx
http://nationalpopularvote.com/sites/default/files/prez-2004-proportional.xlsx
http://nationalpopularvote.com/sites/default/files/prez-2008-proportional.xlsx
http://nationalpopularvote.com/sites/default/files/prez-2012-proportional.xlsx
http://nationalpopularvote.com/sites/default/files/2012-campaign-events-and-results-chart.pdf