Dean Brook Blog

Here’s my analysis of the voting patterns from the May 6, 2006 Shutesbury School Committee election.

These data were collected by observing the counting of ballots after the election. In the case of an election where more than one seat is being filled the data we collected show more detailed patterns of voter preference and intent than the summarization of that data available after the votes have been counted.

We collected data on the pattern of votes for school committee for each ballot. That means that we wrote down the votes for school committee for each ballot. We were later able to tell for example how many people voted for the two incumbents as a group on a single ballot or how many people voted for Dan Greenberg and chose not to cast a second choice.

The law states that the counting of ballots is open to the public and that observers must be able to hear the tallying as it takes place. These data had to be collected during the counting of votes as it is not accessible afterwards. Ballots are not actually public records and after the counting the ballots are kept sealed and may not be opened except with a court order (usually for a recount). The ballots were tabulated at three tables. At each table there were two people, one person read the ballot and the other person kept track of the tally. To observe the counting we had one person sit behind the barrier close enough to each table to hear the tallying.

I’d argue that there are two purposes for reporting ballot results: 1) to publicize the results and 2) to present the intentions of the electorate. The data available after a race where there is only one seat open can provide both results and information on voter preferences. However in races where two or more seats are being filled much of the data from which voter intent and preference may be determined is lost after the counting happens.

In a single winner race a ballot report that lists total votes for each candidate per precinct (Shutesbury is a single precinct) gives a pretty complete picture of both the results and the intentions of the electorate. In a race where there are two winners while reporting just the vote totals per candidate represents the results accurately the intentions of the electorate can still be obscured.

There were 418 ballots cast. The final vote totals for School Committee were:

Daniel Greenberg: 231
Marianne Jorgensen: 218
Anne Lundberg: 202
Stephen Bannasch: 112

In addition to these votes 70 ballots were cast with only one candidate selection for the school committee race. Without collecting the additional data this would be the only voting data from which to draw inferences.

Besides running a good campaign and being an attractive candidate I think Daniel Greenberg won because of the increased turnout in the Baker Road area and in the successful use by his supporters of a strategy to vote only for Daniel.

Of the 231 people who voted for Dan Greenberg 105 also voted for me as a write-in. Additionally 58 people voted only for Daniel. I was not running a campaign, did not describe myself as a candidate, did not ask anybody to vote for me, and had no expectation of winning. I think the people who wrote my name in knew this and so I would clump these 105 votes for me with the 58 ballots cast for only Daniel together into a group of 163 votes and characterize the intention of the voters in this group as a strategic choice made in order to give Daniel the best chance to win. If instead these 163 votes had been evenly spread between Marianne Jorgensen and Anne Lundberg the results would have been quite different:

Marianne Jorgensen: 300
Anne Lundberg: 284
Daniel Greenberg: 231

Here’s the actual rough data (some numbers may not be exact due to errors in our data transcription):

Of the 231 people who voted for Daniel Greenberg:

105 voted for Stephen Bannasch
58 cast a blank as their second choice
38 voted for Marianne Jorgensen
and 30 voted for Anne Lundberg.

Of the 218 people who voted for Marianne Jorgensen:

168 voted for Anne Lundberg
38 voted for Daniel Greenberg
11 cast a blank as their second choice
and none of them voted for Stephen Bannasch

Of the 202 people who voted for Anne Lundberg:

168 voted for Marianne Jorgensen
31 voted for Daniel Greenberg
1 person voted for Stephen Bannasch
and 1 cast a blank as their second choice.

Of the 112 people who voted for Stephen Bannasch:

110 voted for Daniel Greenberg
1 voted for Anne Lundberg
none of them voted for Marianne Jorgensen
and one person cast a blank as their second choice*.

* Why anybody would vote for just me and not Daniel is beyond my comprehension.

One person voted for Marianne Jorgensen, Anne Lundberg, and Daniel Greenberg. The votes on that ballot were discarded because there was no way for the people tabulating votes to determine intent for just two candidates.

There was a block of 168 people who voted for both Marianne Jorgensen and Anne Lundberg. In addition there was effectively a block of 163 people who voted for just Daniel Greenberg (105 of these people voted for me but for purposes of analysis I consider these votes just symbolic). Considering that these two blocks of voters basically balanced each other out the difference between Marianne and Anne was due to slightly more support for for Marianne among the people who voted for Daniel and the fact that 11 people voted for only Marianne. Daniel outpolled Marianne and Anne because he had some support from both people who voted for Marianne and from people who voted for Anne.

In a few weeks data wll be available from the town clerk that will list the names and addresses of people who voted. It will be interesing to compare turnout by area of town in this election to previous elections.

I’d like the town to collect and report the data we collected for any election where more than one seat is being filled in a single race. I find it odd that the data are public for a brief period of time and then become inaccessible.