What was the Real Popular Vote?

No, I’m not referring to the Florida recount of 2000 or even 1876. In this post, I’m discussing a challenge faced by any person attempting to present results of older Presidential Elections. The problem has its root cause in the U.S. Constitution – a small excerpt from Article II, Section I reads: Each State shall appoint, in such Manner as the Legislature thereof may direct, a Number of Electors, …. It turns out, that most states, for a long period of time, chose to appoint Presidential Electors through a method direct popular vote of the Electors themselves. As an example, the residents of the State of Ohio, in 1904, had to cast votes for 23 names among 138 Presidential Electors. The names of the actual Presidential Candidates to whom the long list of Electors are pledged, were not on the ballot.

Given such an unwieldy voting task, the totals amongst the electors of the same party often varies. In our Ohio example from 1904, the 23 Republican Electors’ totals vary by over 3,000 votes. In several cases, including MD 1908, CA 1912, and WV 1916, the closeness of the election combined with the variation in totals for the electors resulted in a split ticket of chosen Presidential Electors!

This variation in Electors is due to many factors, including errors, the desire for some voters to vote for the actual electors that they know, a lack of understanding of the voting procedure, and even the choice of an elector with the same name as another party’s Presidential Candidate such as ‘Alfred E. Smith’ , a Socialist Elector from Colorado in 1928 – he received about 1,000 more votes than any other Socialist Elector.

With all these votes recorded by the various election agencies of the time, many publications created by these agencies summarize the results – inconsistently from state to state and even within a single state from year to year. Examples of summaries include: data shown for the elector whom received the most total votes, data shown for a “generalized average” of electors, data “given is that cast for the elector receiving the highest aggregate vote cast in each county for elector by the party represented”… and so forth. When compiling all of these together – especially for a close election – the result is a “total popular vote” that is not statistically correct. Did Kennedy really win the popular vote in 1960? How about the split result in Popular/Electoral vote of 1888?

I am attempting to collect the data for all Electors in all the states for these years, so that the results may be presented consistantly (and perhaps even variably (meaning that I might be able to have an option to show the “total popular vote” by different statistical methods such as average, highest total elector, median elector, total of highest aggregate elector in each county, total of median electors in each county, etc.)) This is a big job, so don’t expect it too soon.

Comments Welcome.