The ZigZag Conjecture – Another reason to scrap our lever-like ballots

With anecdotal evidence and rational speculation, we introduce the ZigZag Conjecture which may randomly “help” determine election winners and losers in Connecticut. We doubt we are the first to suggest such an effect. Perhaps someone has or will use statistics to study it.

Our Lever-Like Ballots

Like several states, Connecticut does uses a ballot design that could be best described as a the “face of a lever machine on paper”, with boxes for each candidate’s name and bubble. It takes up a lot of space, when there are a lot of parties and unaffiliated candidates. It can include a lot of blank space when some parties only offer a few candidates.  For examples see the ballots from November 2011 <view>  It can include a lot of blank space when some parties only offer a few candidates or there are unaffiliated candidates, for example <here> and <here (be sure and scroll down)>

Party order is specified by the State (the current Governor’s party 1st). Candidate order in vote for multiple races is selected randomly by lottery.

It used to be that you had to use a party lever to vote in Connecticut. Vivian Kellems took care of that, eventually the party levers were removed, and now you cannot vote for a party without filling all the bubbles.

Other states use a clearer ballot that wastes less space, such as this sample from Minnesota <view> It takes away the party emphasis, presumably it would take a voter a bit longer, a bit more care, and preparation to register a party line vote.

The ZigZag Conjecture

Conjecture: Many voters vote in ZigZags: voting from left to right across the ballot they want to vote for a mix of candidates, unaffiliated and from various parties. In vote for multiple races many tend to vote for one candidate in each column, even thought they could vote for more than one in the same column.This can hurt candidates in certain locations on the ballot.

Anecdotal Evidence:

  • This possibility first came to our attention when one particular candidate was unexpectedly the lowest vote getter in a vote for multiple race. It was quickly suspected that the cause was her location on the ballot directly below the highest vote getter in the race.
  • Observing a post-election audit in another town, officials were wondering why a traditional high vote getter did not lead his party in the election. Inspecting the ballot and the results he was located above the highest vote getter in the other party.
  • A similar anecdotal cases were seen in that election and others

Certainly these anecdotes are not conclusive. Perhaps a statistical study could be performed evaluating a large number of local elections to determine the extent to which results support this conjecture. Since ballot position is completely random within a party, if the conjecture were false then we would expect in each municipality on average each party candidate would receive and average number of votes for their party without regard to candidates in the same column – if a statistically significant correlation were shown  based on candidates in the same column between the higher vote getters  in each party vs. the lower vote getters in the other dominant party, then the ZigZag conjecture would be supported.

Other possible problems and implications.

Additional research might include:

  • Does the ZigZag Hypothesis apply between races as well as in a single race?
  • Does the ZigZag Hypothesis increase or drop-off from left to right in a race or across the ballot?
  • Does the first column in a race or ballot effect the direction of the entire ZigZag – or is there correlation between columns?
  • Is it advantageous to be to the right and above or below a high vote getter?
  • Are less informed, less interested voters, especially in local races, more likely to ZigZag in a well intended attempt at fairness and balance?
  • Would successful efforts toward higher turn-out through convenience actually increase the ZigZag effect, bringing more less informed voters to the polls?

Our own speculation would be yes to all of these questions, adding more weight to the need to scrap our lever-like ballot design.


2 responses to “The ZigZag Conjecture – Another reason to scrap our lever-like ballots”

  1. mattw

    Researchers have been unable to conclusively determine if being the first listed on the ballot confers a benefit on a candidate, so I’m not sure if this will ever be solved. The challenge will be to ascertain what a voter would have done if the ballot were laid out differently (such as in the style used in Vermont / California / Minnesota), and which reflects their “true” preference.

    This line stands out to me:

    “Are less informed, less interested voters, especially in local races, more likely to ZigZag in a well intended attempt at fairness and balance?”

    Designing reforms to regulate the conduct of “less informed” voters is silly, but it points out how challenging it is to determine what people really wanted: is a voter less informed because they are zig-zagging, or are they on a par with a party-line voter who doesn’t know the candidates but knows their preference for party control? What about someone who bullet votes “all the Irish names”, or votes a straight party line except for a preference for female candidates?

    In truth, the range of knowledge about candidates is as varied as the range of knowledge about the election system itself, and what happens on election day is a snapshot of a moment in time — who wanted to vote, their willingness to cast votes for people they don’t know anything about, what was on the news that morning, who has transportation and/or free time to get to the polls, who campaigned to which voters and how, as well as things like ballot structure, things that repurpose the vote like count like minority party representation rules or the electoral college, the thoroughness of the Registrar’s canvass procedures, and how close pollworkers stick to the defined election day procedures.

    If your aim is to make the outcomes of elections more perfectly align with the desires of the American citizenry it’s not clear to me that the form of the ballot is the most obvious target for reform — especially as there truly are people who just circle a party name in the mistaken belief (decades after it was true) that Connecticut still has a party lever feature.

    That’s anecdotal, of course, but keep in mind that reforms erase the education that voters have accumulated over time about the election system. Certainly some outcomes would change if you altered minor variables in the system, but I would hope that the necessity for changes would provide a certain improvement (as optical scan provides the ability to hand recount over the lever machines).

Leave a Reply

You must be logged in to post a comment.