Objective

Use PCA and Clustering techniques to segment the demographic data.

Background

Like many communities in the northeastern United States, the town of Sharon, MA (2020 pop. 18,575), governs itself with an Open Town Meeting (OTM), an elected Select Board of three members, and several other officials, boards, and departments. The annual Open Town Meeting is the legislative body of the town. During an OTM, any adult citizen of the town can attend, speak, and vote on policies and expenditures. Sharon’s Select Board of three members and annual OTM have been the basic governance model since 1765.

Recently, citizen participation has been declining and the Select Board appointed a Governance Study Committee (GSC), charged with reviewing the structures and procedures of town government. The Select Board has asked the GSC to recommend modifications aimed at stimulating citizen engagement in the coming years. The GSC has reviewed state statutes, interviewed and surveyed citizens and local officials, and researched the practices in other states.

Under state law, the municipal legislative function can be performed by

• an Open Town Meeting,
• a Representative Town Meeting (RTM), in which citizens elect a group of neighbors to represent their interests, or
• a smaller Town or City Council.

The Select Board has identified two top-priority issues: 1) should the town adopt an alternative legislative form, and 2) should the Select Board grow from three to five members? The committee’s mandate includes other areas, but these are its highest priorities.

Committee members want to interview officials from towns like Sharon to exchange ideas about improving communication with the increasingly diverse citizenry to stimulate involvement in town committees, streamline Town Meetings, use technology more effectively, and so on. Some current town boards have a list of 22 peer towns that they use as a comparison group when setting competitive salaries for municipal employees. Many of these towns are geographically close to Sharon and hence in the same labor market.

The GSC has concluded that “similarity” goes well beyond physical geography and that the factors that might influence civic participation are different from those that drive competition in hiring. Rob is the lone statistician on the GSC, and he has volunteered to locate publicly available data and develop a more suitable list of comparative towns. Imagine that Rob has invited you to assist in this task, and along the way learn about cluster analysis techniques.

Thus far, Rob has gathered data from several sources. The committee has brainstormed a list of variables that might influence voter engagement in local political affairs, and Rob has already done some analysis to narrow down the list of variables and is ready to move forward with clustering.

The Task

At this stage of the project, the following tasks remain:

• Prepare statewide summary statistics: The GSC needs a baseline starting point for comparing Sharon to other communities.
• Data reduction: The number of candidate explanatory factors is large, with several redundant variables available. We need to reduce the number of columns.
• Clustering: Use hierarchical and K Means methods to develop clusters.
• Evaluate and interpret clusters: Examine resulting clusters from the perspective of municipal governance.
• Summarize the governance structures used by similar towns: Compare Sharon to the most relevant other communities.
• Prioritize a list of towns to interview: The GSC has limited time and resources to conduct interviews and must select a small number of communities from a ranked list.