Thursday, August 11, 2005

My Condorcet Re-analysis

     I believe in Condorcet voting, a mathematical system for voting first described by the Marquis de Condorcet.  I decided to look back into history and see which US Presidential elections would have had different outcomes if we had had this mathematically based voting system.

     To be the subect of further research, an election had to have three or more candidates.  If a state has more than 50% of the vote for one candidate, Condorcet would not change the results.  This assumes sincere voting.  No allegations of vote fraud, rigging, stacking or miscounting are discussed or considered.  If the vote was thrown to the US House (see 1824) the House could use Condorcet voting.

     I was very pleased with the outcome.  In many instances, America would have been a less racist, even as it was being truer to its electorate.  Two important elections before the Civil War would have been won by opponents of slavery.  Wilson would have been unlikely to bring segregation to Washington, DC. 

     Famously, A Jackson would have won the election of 1824, even though most other famous elections, including Lincoln's, remain the same.

     More recent results favor both parties.  WJ Clinton probably won in 1992, and GW Bush would not have won in 2000.  This is, with mathematical certainty, the best way to vote for Presidents.  It did not exist when America was founded, although it was invented, later, by a good from of Jefferson by the name of the Marquis de Condorcet.

YearNew VictorThe new victor is in bold face if I am 100% positive of the result
1824 A Jackson JQ Adams vs A Jackson vs W Crawford vs H Clay. There are experts on this matter, it is too complicated for anything less than expert analysis.  It appears that, even though Clay hated Adams and Jackson personally, he would have favored the policies of Adams.  However, it is also likely that people who favored Crawford would have favored Jackson.  Using this bare minimum of facts, combined with an original EV count of Jackson 99, Adams 84, Crawford 41 and Clay 37, the result of Condorcet balloting in the US House (where the election was thrown) would have favored Jackson to a great degree.  Considering how almost each state had its own policy at that time, and many of the states split their electors at first, it really is not easy to say.  Based on the Poole/Rosenthal technique, but relying on the existing voting system, authors J Jenkins and B Sala show that the "corrupt bargain" thesis is likely wrong, in any event.
1844 H Clay H Clay/T Frelinghuysen's vote was split by the abolitionist Liberty Party. Assuming 100% support for the Whigs over Democrats JK Polk and G Dallas, the votes for the states of Michigan and New York would have switched columns under a Condorcet system. The original electoral vote victory was 170 for Polk to 105 for Clay. The new total would be 146 Clay to 129 Polk.  Ignoring Michigan's 5 electoral votes, let's examine New York and their 36.  The actual vote total for the State is 237,588 for Polk, 232,482 for Clay, and 15,812 for Birney. To win, Clay would have needed 10,460 of the Liberty Party's vote, or almost exactly 2/3rds. Considering the Liberty Party is a radical abolitionist party, and the Democrats at that represented the southern slave-holding aristocracy, this is more than likely. Clay was a lifelong enemy of slavery, although he was not radical about it.
1848 L Cass Free Soil Party (Liberty + some Whigs), with Martin Van Buren as their candidate, splits the Democrat vote away from long-time politician Lewis Cass (appointed Governor of Michigan by James Madison) and elects Zachary Taylor, who had never held political office before.  This split is well known to political historians.
1884 JG Blaine SG Cleveland vs JG Blaine.  BF Butler, a Radical Republican (think liberal democrat) was the Democrat Governor of Massachusetts (think M Bloomberg, Republican Mayor of New York), who entered the race on the Greenback ticket in order to split the Democrat vote and cause Cleveland to lose.  The Republican, Blaine, would have been the second choice of a Greenback voter.  It also appears, if the infamous "Rum, Romanism and Rebellion" statement had any truth to it, that the Prohibitionists would have favored Blaine. 
1888   I need more information.  B Harrison was seen as close to the elites, so probably wouldn't have gotten the two labor party votes (Union Labor and United Labor) while I imagine the Prohibition vote would have gone for B Harrison.  I'm not sure.
1912 T Roosevelt WH Taft and T Roosevelt split the Republican vote.  I calculate that almost all Republicans would have picked the other Republican, Taft or Roosevelt, above Wilson.  I also imagine that the Socialist vote breaks for the Republicans.  Using this, Wilson only wins the eleven states of the former Confederacy.  This result is, by far, the biggest change in any of the elections in terms of number of states changing hands.  An estimate of the new electoral vote total goes with Roosevelt 227, Taft 164, and Wilson 140.  This doesn't give Roosevelt the required EV votes to win a majority, which means the race would be thrown to the US House.  The Democrats had just taken control of the House.  One can use this page to see that 23 State delegations were majority Democrats, and 21 majority Republican.  Three, Rhode Island, New Mexico and Maine, were evenly split.  I simply can't guess.  It'd be nice for me to think of Wilson losing.
1916   The problem here is the first World War.  Hughes was seen as slightly more pro-war than Wilson (preparedness).  Yet Socialist AL Benson criticized Wilson for his preparedness position, and left the Socialists (with a faction) when they blamed the US for the war.  Hughes would have needed only 55% of the Socialist vote in California to win.  To add confusion, however, the Prohibition party got enough votes in California that it would take Hughes 86% of the California vote to lose 100% of the Prohibition vote.  If the Prohibition vote is pro-Hughes, well, it still is up in the air.  Only three states weren't won by a majority, and only California can change the result.
1992 GHW Bush GHW Bush needed 101 more electoral votes to win.  If he had, nationwide, more than 66% of Perot voters behind him, he would have won the election.  Actually, he just needed 2/3rds in Georgia, New Hampshire, Ohio, Montana, Nevada, New Jersey, Colorado, Wisconsin, Kentucky, Maine, Connecticutt and Iowa.
2000 A Gore Even if every other third party broke 100% for GW Bush, A Gore needed about 71% of the R Nader vote in Florida or 82% of the Nader vote (losing 100% of all others) in New Hampshire to win those states, either would have given him the victory.

     Close and/or Famous Races: which might leave certain readers wondering; for which details are provided.

1796 Only one elector didn't choose J Adams or T Jefferson as their first choice, and so couldn't have changed the 71 to 68 result.
1800 This famous election, if Democratic-Republican and Federalists stayed true to form, would not have been changed if Condorcet voting methods had been used.
1852 The Free Soilers might have changed the result in Delaware, Connecticutt and Ohio, but that wouldn't have been enough to put W Scott ahead of F Peirce.
1856 The Whig-American Party, a major, anti-immigrant, anti-catholic, pro-bible, third party that nominated former President M Fillmore, could not have J Fremont the win over J Buchanan even if they supported J Fremont 100%.
1860 Lincoln vs Breckenridge vs Douglas vs Bell.  The only two states that Lincoln won by plurality (with less than 50%, but still the greatest number of votes) were California and Oregon.  Since they only had seven EVs between them, and the EV tally was Lincoln 180, Breckenridge 72, Bell 39 and Douglas 12, it would not have changed the result.
1876RB Hayes vs S Tilden was not swayed by Greenback Party nominee P Cooper
1892 The Populist Party was more pro-Democrat, and the Democrats won, so J Weaver didn't change the SG Cleveland victory over B Harrison.
1948 There's no chance the progressives or dixiecrats would have voted for Dewey, is there?
1960 If Nixon received 95% of the third party vote in New Jersey and Illinois, he might have won.  That seems very unlikely.  The total of third party votes in Illinois was under a quarter of one percent.
1968 Four out of five G Wallace voters were RM Nixon supporters.
1976 I imagine that most E McCarthy voters would have been JE Carter supporters.
1996 WJ Clinton could lose 92 electoral votes if the HR Perot vote goes 81% for R Dole, and the other third parties split evenly.  However, that still isn't enough in this 379-159 race.  The fourth biggest party, the Greens, would likely have broken for WJ Clinton.

     List of elections for which Condorcet voting systems could not possibly have effected the outcome: 2004, 1988, 1984, 1980, 1972, 1964, 1956, 1952, 1944, 1940, 1936, 1932, 1928, 1924, 1920, 1908, 1904, 1900, 1896, 1892, 1880, 1872, 1868, 1864, 1840(Liberty), 1836(Four Whigs, did not compete with each other), 1832, 1828, 1820, 1816, 1812, 1808, 1804, 1800, 1796, 1792

No comments: