Why are there so many Buchanan Voters in Palm Beach County?

There are two obvious explanations for the large Buchanan vote. One is that for some reason, Palm Beach County has an unusually large number of Buchanan supporters who voted for him. The second is that some feature of the ballot caused large numbers of non-Buchanan supporters to inadvertently cast their votes for Patrick Buchanan.

Several pieces of evidence have been offered to suggest that Palm Beach County has an exceptional number of Buchanan supporters. One is that Buchanan obtained 8,788 votes in Palm Beach County in the 1996 Republican primary in Florida. This number is larger than the 3,407 he received in the 2000 Presidential race. However, in 1996, Buchanan's support from Palm Beach County was 5.4% of his statewide total, whereas in 2000 it was 19.6% of his statewide total. These figures can only be explained by believing that forces that were reducing Buchanan support elsewhere in 2000 were not working in the same fashion in Palm Beach County. Furthermore, in 1996, Buchanan's support in Palm Beach County was not statistically significantly different from what could be predicted from simply looking at the relationship between Buchanan support and Dole support in all other Florida counties. Yet, as pointed out above, in 2000, Buchanan's support was much, much greater than expected.

Another argument is that there is an unusually high registration of Reform Party members in Palm Beach County. In fact, in 2000, there were only 336 registered Reform Party members in Palm Beach County. Clearly, these are not enough people to account for the 3,407 votes recorded for Patrick Buchanan. In addition, there were only 1,282 voters in Palm Beach County for Joel Deckard, the Reform Party candidate for the Senate.

Moreover, statistical analysis shows that there is no relationship between the number of voters who voted for Buchanan in Palm Beach County precincts and the number of voters who voted for Deckard - even though there is a statistically significant relationship between the two votes in the adjoining Broward County. This very strongly suggests that there is something different about Buchanan's vote in Palm Beach County.

The support for the proposition that Buchanan votes resulted from something other than ballots cast by Buchanan supporters is increased beyond any reasonable doubt by the following analysis. In two nearby counties for which I have obtained data (Broward and Miami-Dade), the Buchanan vote among absentee and election day voters is very similar. In Broward County Buchanan support is .13% among election day voters and .17% among absentee voters. In Miami-Dade County it is .09% among election day voters and .11% among absentee voters. In both cases, the absentee voters are somewhat more likely to vote for Buchanan, but the increase in Buchanan support between the election day voters and the absentee voters is no more than about 31%. (Thus, in Broward County we calculate $\frac{[.17 - .13]}{.13} = .307$ or 31%).

The comparable figures for Palm Beach County are astonishing. Three things stand out on Figure 1 which presents the data for all three counties:

(1) The Buchanan vote among absentee voters is .23% and this is well within the error bounds of what almost all models have predicted for the level of Buchanan support among Palm Beach County voters. That is, the Buchanan vote among absentee voters looks like what researchers around the country would expect from Palm Beach County as a whole.

(2) The Buchanan vote among election day voters of .85% is, unlike that in the other two counties, larger than the Buchanan vote among absentee voters of .23%.

(3) Most incredibly, the Buchanan vote of .85% among election day voters is almost four times bigger than the Buchanan vote of .23% of absentee voters.

In short, while the absentee vote for Buchanan in Palm Beach County is what we expected based upon our models, the election day votes in Palm Beach County are extraordinarily anomalous.

A look at the absentee ballot (Figure 2) makes it clear why this is so. The absentee ballot in Palm Beach County was designed much differently than the election day ballot. It lists candidates in columns with controls to the right. As a result, there is virtually no chance that a voter will inadvertently circle one candidate when intending to vote for another.

In short, I believe far beyond a reasonable doubt that it was the election day ballot that created the anomalous Buchanan vote percentage in Palm Beach County.

In addition, a comparison of the Buchanan vote among those who cast absentee ballots with the vote of those who cast election day ballots provides an excellent measure of the number of mistakes caused by the butterfly ballot. In the other two counties, the percent Buchanan vote among absentee voters is higher than that among the election day voters. Thus, a very conservative estimate of the proportion Buchanan voters in Palm Beach County is the percent Buchanan among the absentee voters which is .23%. The difference between the vote share for election day voters of .85% and this figure of .23% for absentee voters is an excellent and very conservative measure of the mistakes caused by the Butterfly ballot. This approach yields an error rate percent of .62% (calculated from 85% - .23%) on the 386,731 votes that were initially reported as cast on election day. This amounts to about 2,400 Buchanan votes that were cast in error and only about 1,000 votes from true Buchanan supporters.