When voters confront a ballot form, they must quickly figure out how to respond in order to convey their preferences. This task requires reading instructions and information about the candidates on the ballot, determining how to produce an effective response, and detecting and correcting any errors in the entire process. The butterfly ballot form in Palm Beach County was designed in such a way that it provides highly contradictory cues to voters as they undertake these tasks. As a result, it produced an unusual number of errors that created substantial biases in the Palm Beach County voting results.
My conclusions about this ballot are based upon my experience as Director of the University of California Survey Research Center and as a scientific survey researcher who has designed and undertaken surveys in Canada, Estonia, Russia, and the United States. I am knowledgeable about the design of survey questionnaires and the factors that affect response quality. Questionnaires for scientific surveys involve either responses filled in by survey interviewers or by the respondent. In either case, survey researchers are very concerned with limiting the number of errors and reducing the bias in responses. From this perspective, I can say with a very high degree of scientific certainty that the ballot form in Palm Beach County violated standard and well-known principles of form design.
The Palm Beach County butterfly ballot is designed with candidates listed in two columns with punch-holes between the two columns. The left hand column lists the Republican ticket, then the Democratic ticket, and then other tickets. The right-hand column lists the Reform ticket, the Socialist ticket, and then other tickets.
The punch-holes are for the Republican, Reform, Democratic, Socialist, and other tickets in that order. The parties are numbered (3) Republican, (4) Reform, (5) Democratic, (6) Socialist, etc. The punch-holes are in the same order. The physical layout places Republicans highest on the ballot, Reform second highest, Democrats third highest, Socialist fourth highest, and so forth.
Thus, three features of the ballot - its numbering, the order of the punch-holes, and the physical position of the party and candidate labels, suggest the following order of the candidates: Republican, Reform, Democratic, Socialist, etc. But detecting these features and determining this order requires that voters move their eyes back and forth across the page to read labels in both columns and to ascertain the position of punch-holes.
Deciphering and marking this ballot is difficult because its format violates several common conventions that allow people to understand complex forms. One convention is that people read down columns and then to the top of the next column. According to this convention, the parties are in the order Republican, Democrat, etc., but this ordering is different than the order of the punch-holes. The second convention, that the major parties are listed first on the ballot in Florida, leads to the same problem because the punch-holes include the Reform Party in-between the two major parties. A third convention, that punch-holes or "answer" boxes are typically to the right of the candidate names, is also violated. To properly cast their votes, Palm Beach County residents had to ignore these conventions. They had to go back and forth from one column to another as they deciphered and marked their ballot, and they had to look for punch-holes on the left or the right depending upon the column they were in. Consequently, voters were confronted with a very difficult cognitive task given the short time that they have to complete their ballot. Because the ballot violated conventions, there was no clear relationship between the stimulus and the response. Voters who read down the column could easily infer that the first punch-hole was for George Bush and the second punch-hole was for Al Gore. As a result, Gore supporters could inadvertently punch the hole for Pat Buchanan. Other errors are also possible, although I will not explore them here.