Data Sources and Methodology

Analytical Approach

Our analysis identifies patterns in Maine school closures by examining:

• Historical closure data: Where and when schools have closed
• Building characteristics: Age, size, and location of school facilities
• Enrollment trends: Changes in student populations over time
• Economic context: Construction costs and funding availability
• Geographic patterns: Rural vs. urban distributions

By mapping these factors, we identify common characteristics among closed schools and highlight open schools that share similar profiles.

Tools and Technologies

• DuckDB: Data processing, cleaning, and analysis (used initially to examine large files)
• Observable Framework: Interactive web-based visualization platform
• GeoJSON: Geographic data format for mapping school locations
• JavaScript/D3.js: Custom visualizations and interactivity

National Center for Education Statistics (NCES)

NCES Education Demographic and Geographic Estimates (EDGE) - School Locations

• URL: https://nces.ed.gov/programs/edge/Geographic/SchoolLocations
• Coverage: 2015-2016 through 2023-2024 school years
• Includes: Geographic data for public and private schools
• Updated: Annually

NCES Common Core of Data (CCD) - Public School Data

• URL: https://nces.ed.gov/ccd/
• Includes: School names, addresses, enrollment, grade levels, operational status
• Updated: Annually

NCES Private School Universe Survey (PSS) - Private School Data

• URL: https://nces.ed.gov/surveys/pss/
• Includes: School names, addresses, enrollment, grade levels, religious affiliation
• Updated: Biennially (every 2 years)

NCES Open GIS Data - Public School Locations 2023-24

• URL: https://nces.ed.gov/opengis/rest/services/K12_School_Locations/EDGE_GEOCODE_PUBLICSCH_2324/MapServer
• Map Service ID: 0
• Format: GIS web service for direct mapping integration

School and District Navigator

• URL: https://nces.ed.gov/ccd/schoolmap/
• Used for: Cross-referencing school locations and verifying data accuracy

---

Maine Department of Education

Summary of Maine School Building Inventory

• URL: Maine School Building Inventory Report
• Includes: Comprehensive facility data including building age, size, and condition
• Used for: Assessing infrastructure characteristics and identifying at-risk facilities

Maine DOE Data Warehouse

• URL: https://www.maine.gov/doe/data-reporting/warehouse
• Includes: State-specific school information, district boundaries, enrollment trends
• Used for: State-level context and verification of NCES data

Maine Open and Closed Schools Database

• URL: Maine Open and Closed Schools Database
• Format: Excel spreadsheet (downloaded)
• Provides: Comprehensive list of closed schools with closure dates
• Note: Critical primary source for identifying closed schools

---

Geographic Reference Data

Maine Town and Townships Boundary Polygons

• URL: Maine Town and Townships Boundary Polygons
• Format: GeoJSON/Shapefile
• Used for: Municipal boundary overlays and rural/urban classification

Maine Long-Range Facilities Master Plan for Schools

• URL: MSAD 75 Public Forum - Community Feedback Summary
• Format: PDF
• Note: Recent community vote on policymaking for schools

---

Secondary Sources: Economic and Policy Context

Essential Programs Services Formula

• URL: https://themainemonitor.org/essential-programs-services-formula/
• Used for: Understanding state funding mechanisms and allocation formulas

Governor's Commission on School Construction Interim Summary (April 15, 2025)

• URL: Governor's Commission Interim Summary
• Key data: $11 billion total infrastructure need, construction cost trends, funding gap analysis
• Used for: Economic context in narrative

Why is Building and Renovating Schools So Expensive? (Central Maine, September 2024)

• URL: Central Maine Article
• Key data: Construction costs rose from $270/sq ft (2015) to $661/sq ft (2024)
• Used for: Cost escalation visualization and narrative context

Maine School Building Renovation Expenses (News Center Maine)

• URL: News Center Maine Article
• Provides: Additional construction cost context and case studies
• Used for: Supporting narrative on economic pressures

$156 Million Middle School Under Construction (Maine Monitor, July 2025)

• URL: https://themainemonitor.org/windham-156-million-dollar-school/
• Key data: State funding success rate (9 of 74 applicants), project cost examples
• Used for: Funding competition context

Maine Forms Commission to Address Aging Infrastructure (News Center Maine)

• URL: News Center Maine Article
• Key data: $580 million invested since 2019, ongoing demand exceeding resources
• Used for: Policy context and state response

Bill to Add Hotel Sales Tax for School Construction (News Center Maine)

• URL: News Center Maine Article
• Provides: Funding formula inequities, rural vs. urban disparities
• Used for: Understanding structural barriers facing rural communities

Portland Press Herald: 5 Takeaways from our Reporting on Maine's School Construction Backlog

• URL: Portland Press Herald Article
• Used for: Comprehensive overview of infrastructure challenges and state response

---

Supplementary Research: Individual School Closures

The following sources were used to verify closure dates, fill data gaps, and provide case study context for specific schools:

• Maine Girls Academy closure
• Catherine McAuley High School
• Bangor Hilltop School
• Cocoons Day School
• School Around Us Community
• MSAD 49 Albion Elementary School
• Clinton and Benton School History
• Juniper Hill School
• Alna's Juniper Hill School Reopening
• MMCFC Falcon Auburn
• Margaret Murphy Center Potential Cuts
• Toddle Inn Child Care Reviews
• Maine Legislature Testimony
• Toddle Inn

Development Process

This interactive visualization was developed with assistance from GitHub Copilot (Claude Sonnet 4.5) through an iterative prompt-driven workflow. The AI assisted with technical implementation while all analytical decisions, data interpretation, and narrative framing were directed by the research team.

Major Development Phases

Initial Setup and Data Integration

• Integrated multiple data sources (NCES school locations, demographics, EPA sites, enrollment time series)
• Created data loader scripts to compile data into CSVs (see data/scripts_backup folder)
• Developed helper functions for distance calculations and data formatting

Risk Algorithm Development

• Designed weighted risk scoring algorithm combining:
  • Enrollment decline trends (greater than 10% decline threshold)
  • Child population decline in census tracts (greater than 5% decline threshold)
  • Poverty rate (normalized 15 to 35% range)
  • Education attainment (bachelor's degree rates)
  • Unemployment rates
• Implemented configurable weight controls for each risk factor

Interactive Map Creation

• Implemented color-coded school markers by risk score (green to yellow to red gradient)
• Added zoom functionality with scale-invariant dot sizing using D3 customization
• Created county labels at geographic centroids with zoom-responsive text sizing

Summary Panel and Statistics

• Built compact summary panel showing filtered school counts
• Displayed risk distribution (low/medium/high categories)
• Calculated enrollment statistics (total students, averages)
• Integrated environment and demographics metrics (EPA proximity, public/private counts)

Detail Panel Implementation

• Designed slide-out detail panel triggered by school marker clicks
• Displayed comprehensive school information including:
  • Risk score with color-coded background and contributing factors
  • School profile (type, status, enrollment, grades, locale)
  • Nearby EPA brownfield sites (within 2-mile radius)
  • Census tract demographics (income, poverty, unemployment, education)
  • Time series charts comparing enrollment vs. child population trends
• Implemented enrollment/population trend convergence/divergence analysis
• Added close button and off-screen hiding behavior

Filter Controls

• School level filter (Elementary, Middle, High, Multi-Level, Other)
• School type filter (Public, Private)
• EPA proximity filter (Near sites, Not near sites)
• Locale filter (12 categories: City/Suburb/Town/Rural x Large/Midsize/Small/Fringe/Distant/Remote)
• Minimum risk score slider (0 to 10 range)

User Interface Enhancements

• Added page margins for better layout
• Created map header with instructional text
• Implemented algorithm tooltip button showing risk calculation formula
• Styled Observable Inputs with compact, cohesive design

Visual and Interactive Features

• Implemented hover tooltips on school markers showing name and risk score
• Added click event handlers for detail panel population
• Created smooth panel slide-in/slide-out animations
• Added county boundary visualization with labels
• Implemented responsive scaling for markers and labels during zoom

Styling and Polish

• Styled range sliders with custom thumbs and gradients
• Styled checkboxes with blue accent colors
• Added hover effects on interactive elements
• Implemented form container styling with borders and backgrounds
• Created detail panel CSS for tables and sections

Transparency Note

This iterative development approach allowed for rapid prototyping and incremental refinement of both functionality and design. All analytical judgments, methodological decisions, and interpretations remain the responsibility of the research team.

Data Quality Challenges

Inconsistent Reporting and Incomplete Records

We encountered significant challenges with data quality and completeness during this project. Maine's school closure data is fragmented across multiple sources with inconsistent formatting, missing closure dates, and incomplete facility information. We have submitted data requests to the Maine Department of Education and continue to advocate for better public data infrastructure.

Known Gaps and Constraints

Closure Timing:

• Some closure dates are approximate or based on last available enrollment data
• Private school closures are significantly under-reported compared to public schools
• Exact closure dates are often unavailable in public records

Building Condition:

• Facility condition assessments are not comprehensively available for most schools
• Deferred maintenance data is limited to publicly reported information
• Inspection reports are often not publicly accessible without FOAA requests

Financial Data:

• Individual school district budgets and bond vote results require extensive manual research
• State funding application outcomes are not always publicly accessible in real-time
• Historical funding decisions are difficult to track systematically

Predictive Limitations:

While our pattern analysis identifies key risk factors, it cannot account for:

• Sudden policy changes or emergency funding
• Unexpected funding sources (federal grants, private donations)
• Community-specific factors not captured in quantitative data
• Successful grassroots efforts to prevent closures
• Political will and local decision-making dynamics