Massachusetts’ Address Lists

Note from Will B: This post was hand-written by me. It’s a bit long. You can read an AI generated summary here. I’ve reviewed the summary and it’s fair.

This post focuses on how address lists maintained by state and local governments in Massachusetts support a complete count of Massachusetts residents in the Decennial Census and highlights the challenge of improving those lists. Raising the quality of address lists across the state is a project that we should be focused on now to prepare for the 2030 census. Good quality address lists going into 2030 will help the commonwealth and municipalities obtain their fair share of funding over the coming decade and could also help protect Massachusetts’ representation in Congress.

Background

For determining representation in Congress, the U.S. Constitution requires an actual enumeration of persons in each state. The Decennial Census implements that requirement. Our state’s access to federal funds, as well as our state’s representation in Congress, depend on a complete count of our residents. Assuring a complete count in the next Census is one of the concerns of the state senate’s Committee on the Census.

When Census enumerators seek to count the residents of each state, they work from a list of residential addresses. That list includes both individual units, like houses and apartments, and group quarters, like dormitories. Through email, mail, direct visits, and sometimes other records, enumerators attempt to find out who is living at each listed address. A complete count depends on the accuracy and completeness of the address list, known as the Master Address File or “MAF.” The Delivery Sequence File maintained by the U.S. Postal Service is one of several sources that the Census uses to maintain the MAF. The National Academies’ assessment of the 2020 Census (Chapter 5) offers a deep dive about the process for creating and updating the MAF.

Our focus is on how state and local government can help improve the MAF. Substantial undercounts in the 1990 Census led to calls for reform and ultimately to the The Census Address List Improvement Act of 1994 (P.L. 103-430). That legislation required the Census to give local governments the opportunity to review and provide comment on the address list within their jurisdiction. The Census Bureau implemented this law through the Local Update to Census Address (LUCA) program. The LUCA program offers municipalities a suite of tools to compare their own address list with the MAF, identify discrepancies, and provide comment.

LUCA does not officially open for the 2030 Census until 2027, but the Census encourages municipalities to begin preparing now. The state has a compelling interest in assuring that municipalities have all the tools and staffing that municipalities need in order to begin preparing. The primary official involved in supporting municipalities is the Secretary of State. The Massachusetts Population Estimates Program at the UMass Donahue Institute provides important technical services to the Secretary and to Municipalities. The state’s geographic information services department, MassGIS, has created a state level Master Address Database that can help municipalities who have not fully developed their own database to prepare for and participate in LUCA.

Massachusetts’ Master Address Database

Legal Framework

Chapter 7D of the General Laws establishes the Massachusetts Office of Information Technology and within it the Bureau of Geographic Services, endowed with standard setting powers:

Section 5. There shall be a bureau of geographic information within the office which shall develop, maintain, update and distribute geographic information, technology, data and services for use by state agencies, municipalities and the public. The office shall coordinate all geographic information activities in state and local government and shall collect, manage and distribute geographic information maintained by state agencies and local government agencies. Subject to sufficient appropriation, the office shall provide technical services related to geographic information to state agencies and municipalities. The secretary may, as needed, set standards for the acquisition, management and reporting of geographical information and for the acquisition, creation or use of applications employing such information by any state agency and the reporting of such information by municipalities.

G.L. c. 7D, s. 5. See also MassGIS overview.

The bureau of geographic information, commonly known as MassGIS, standardizes, maintains and presents a wealth of geographic information which can be conveniently viewed online.

Among its many valuable data products, MassGIS has implemented a statewide master address database to support emergency response. The database includes street addresses — “24 Beacon Street, Boston” — and geographic coordinates for each. Although MassGIS has the mandate to standardize a statewide address database, it is clear that municipalities are the underlying authority for address definition. Several statutes are relevant:

• Chapter 41, sections 73 through 79 authorized the creation of local boards of survey for laying out town plans and public ways and sections 81A through 81GG created the more modern concept of a planning board with more extensive powers. These sections also recognize the implicit power of municipal governing bodies to control local planning (see sections 77 and 81G).
• Chapter 82 further regulates the laying out of public ways. Since public ways may continue across municipal boundaries chapter 82 interweaves state, county, and municipal authority.
• The conveyance of property ownership requires precise definition of the actual boundaries of each property. Property boundaries are defined by surveyors following standards partially defined by Chapter 97. Property boundaries are recorded in the registries of deeds which are organized by county.
• Local property taxation is based on these county records. Section 11 of Chapter 59 provides that : “Taxes on real estate shall be assessed, in the town where it lies, to the person who is the owner on January 1, and the person appearing of record, in the records of the county, . . .”. Local assessors maintain an inventory of the properties in their municipality for the purpose of assessing taxes on each property.
• The requirement for display of a street number on structures originates in the state’s firecode. This section also requires the state 911 department to enter addresses into the state’s emergency response database which is established under Chapter 6A, section 18A through 18J. However, these are to be supplied voluntarily by municipalities: Paragraph (g) of section 18B provides that “The [state 911] department shall initiate a voluntary program in which municipalities may contribute timely address information to support the enhanced 911 database.”
• MassGIS has encouraged municipalities to enact bylaws fully defining their process for assigning official names to streets and numbers for parcels along the streets. The authority by which municipalities could enact these bylaws seems implicit in their planning powers and could also be authorized under the catch-all authorization of Chapter 40, Section 21, paragraph (1).

None of these statutes standards for how numbers and street names should be assigned. In a document titled “An Address Standard for Municipalities” published in 2016, MassGIS offers guidance to municipalities for address assignment. This document speaks to several kinds of issues:

• Conceptually, how should physical streets and parcels be named and numbered to avoid confusion? This may seem straightforward enough along city blocks — the only basic rule being to avoid duplication. But where streets cross and re-cross municipal boundaries or branch out into a series of dead ends, confusion may arise — confusion that could be dangerous in an emergency. The standards speak specifically to a number of possible on-the-ground layout scenarios. Additionally MassGIS standardizes community names within which street duplication should be avoided. This is especially important within some municipalities, notably Boston, which do have duplicate street names across different neighborhoods.
• How should streets and street numbers be referred to on lists? We are all familiar with various address abbreviations that are used for mailing. The document discourages use of postal name abbreviations (too various) or reliance on postal zip codes (may shift over time). Additionally, of course, a P.O. Box is of no use for defining a parcel. The document encourages municipalities to establish a complete official list of full street names and to follow specific naming standards.
• How should municipalities go about implementing standardized addresses? The document recommends establishing clear departmental authority within each community for address standardization and suggests that official address assignment can be enforced through the permitting process. “Interested parties” should work together to create use of consistent addresses across municipal functions.
• The document offers model bylaws creating address authority and model regulations for implementing address standardization.

Creating the Master Address Database

MassGIS describes the creation of the statewide address database as a compilation of data from three main sources:

• Most importantly, the statewide address database is derived from parcel data from the local assessors. Within some larger parcels (condominium complexes, college campuses), MassGIS has had research individual structures which may have distinct addresses. MassGIS has helped assessors standardize to create a complete map with digitized parcel information across the state.
• Public utilities — phone, cable, and electric — provided anonymized service/billing lists identifying most housing units that may include geographic information.
• Public voter or resident lists compiled by election officials include no geographic information (latitude and longitude) and the personal names they include are irrelevant to the address database. However, the voter lists may help identify individual structures within parcels and apartment numbers within structures (although not all housing units include voters).

The Master Address Database benefits from continuing improvement. The clean up and reconciliation of incompletely standardized address references (is it “Arsenal Court” or “Arsenal Court Drive”?) is an ongoing effort. New addresses emerge as properties are developed and MassGIS hopes that municipalities will provide timely updates.

Access to the Master Address Database

MassGIS publishes four different views of the state master address database that can help municipalities perfect their own master address databases. Only one of the four views includes point location information for the City of Boston. Three of the four views include a master_address_id that can be used to link across them. Two of the views include an identifier that can be linked to the federal Department of Transportation’s National Address Database.

Test Match of Voter list to Master Address Database

As an exercise to gain some insight into the challenges in reconciling local lists with the state’s Master Address Database, we matched the address set for voters in my senate district (extracted from recent community voter lists in my senate district) to the Statewide Address Points for Geocoding, one of the four MassGIS data products that are based on the Master Address Database.

To match addresses from the voter lists to addresses on the Master Address Database (MAD) we had to take the following progressive matching steps, fixing breaks along the way:

• Replace the postal mailing abbreviations with full street names — for example, “Gilbert Rd” matches to “Gilbert Road.” This was straightforward since the voter lists follow postal service abbreviation conventions.
• Replace approximately 20 base street names that differed slightly from the voter lists to the MAD — for example “Arsenal Court Dr” matches “Arsenal Court” on the MAD. At this stage, we identified one instance where the MAD appears to be incorrectly placing a new building on an existing street (“Nevins St”) instead of on its newly created street (“Nevins Hill Way”). We also had to split addresses showing on the voter list on Irving Park into North Irving Park and South Irving Park according to whether they were odd or even and finally to correct two individual street addresses that had obvious street name typos on the voter list.
• At this stage, 101529 (99.3%) of 102,288 street/street_number/municipality addresses on the voter list matched to at least one point on the MAD. Using Google maps, we investigated the 24 unmatched addresses accounting for the highest number of voters and found that in most instances there was a plausible match point on the MAD. Most commonly the number on the voter list was a plausible number for a building with multiple doors, but the number on the MAD was for a different door. We left 195 addresses from the voter lists unmatched.
• Note that we ignored zip code as a match variable. We had clipped the MAD points to exclude points outside my senate district using mapping software. Within my senate district, there are no duplicate street names within community, so there was no need to include zip code in the match — if we had all of Boston in the MAD point set, we would need to differentiate by zip code because a few street names do repeat across Boston neighborhoods. However, we did compare the voter list zip codes with the MAD zip codes for the matched points and counted 52 breaks — we spot checked 25 of these and found them to lie on boundary lines.
• So far we have only been matching on street number, street, and municipality. Additional address components are suffix and unit. Omitting those last two components is a soft match — “26 Main St” would match “26A Main St Apt 6.” We explored the extent to which this relying on this soft match for geocoding the voter list would introduce errors in geolocation. We were surprised to find that on the MAD there were groups of address points that soft match to each other on street number, street, and municipality, but vary in lat/lon location by hundreds of feet, in one case over a mile. In some cases these were institutional street addresses like McLean Hospital or the Perkins school that have big campuses, but in others, it appears there is a geocoding error in the MAD. This geocoding error goes along with a geocoding error in Google maps and that may be the source of the error. As an example in Cambridge, there is a voter address of the form “123A Main Street” which both the MAD and Google locate over a mile from “123 Main Street” at a location that would correspond to “A Main Street” in street number sequence.
• Our last step was to do a series of progressively softer matches starting with strict match using all the voter list address components — street number, street number suffix, street name, unit. In the strictest match, if the fully-defined voter address matched to multiple points on the MAD, they had to all have the same geolocation. In successive steps, the matched points could be wider apart. The differences in match rate across communities in the first (strictest) pass match rate largely reflects the extent to which the neighborhoods include larger multi-unit buildings — if voter addresses that include a suffix or unit are excluded, all four communities have first pass match rates exceeding 98.3%.

Percentage of voter address list matched to Master Address File by progressively softer matches

Strictest level of match achieved	Belmont	Boston	Cambridge	Watertown
Exact full voter address match to address points with single location	99.2	85.7	96.2	93.1
Exact full voter address match to address points within 20 feet	0.0	1.7	0.0	0.0
Exact full voter address match to address points within 200 feet	0.0	0.1	0.0	0.0
Number only match to address points with single location	0.5	8.9	3.5	5.2
Number only match to address points within 20 feet	0.0	1.1	0.0	0.0
Number only match to address points within 200 feet	0.0	2.1	0.2	0.5
Soft match number match (2 off, same side) within 100 feet	0.3	0.1	0.1	0.7
No match achieved	0.1	0.2	0.1	0.5
Total	100.0	100.0	100.0	100.0

A further useful comparison to explore locally would be between the MAD and Local Assessor data, which is also publicly accessible. The UMass Donahue Institute makes comparisons like this statewide and in-depth to support Census preparation. For additional discussion of the matching process, see this 2020 Municipal Training Presentation for 2020 from the UMass Donahue Institute.

Discussion and Next Steps

All four of the communities that I represent have invested substantially in geographic mapping and are respected for their work in this area. Even in these communities, glitches do appear in the mapping process. These glitches are to be expected — address matching is intrinsically messy. But the observed glitches in the single comparison included in this post highlight the challenges that communities that have not made such substantial investments may still face. Raising the quality of address lists across the state is a project that we should be focused on now to prepare for LUCA in 2027. Address matching often requires painstaking manual investigation which does take time. This is a municipal challenge which the Secretary of State and UMass Donahue Institute are very aware of and they have consistently assisted municipalities in their efforts. The state senate’s Committee on the Census will be looking for additional ways to help.

Resources

• Census Resources
  • Census Bureau Presentation to Senate Committee on the Census (June 16, 2025)
  • National Conference of State Legislature’s Presentation to Senate Committee on the Census (June 16, 2025)
  • National Academies’ assessment of the 2020 Census
  • Local Update to Census Address (LUCA) home page
  • Census geocoding tools — get latitude, longitude, and identifiers for particular addresses
  • Census Housing Unit Change Viewer
  • Census Address Count Listing Files Viewer
  • Two additional Census programs that solicit municipal input (distinct from addresses listing)
    • Block Boundary Suggestion Program
    • Boundary and Annexation Survey 
• Other federal resources
  • Federal Geographic Data Committee — Address Standards
  • Federal Department of Transportation National Address Database
• State Technical Assistance Resources
  • Secretary of State
  • UMass Donahue Institute Population Estimates Program
    • Municipal Training Presentation for 2020 (video)
    • Municipal Training Presentation for 2020 (handout)
    • Boston Foundation 2020 Census Forum including UMass Presentation at 41:28 (See NoteGPT generated summary here)
• MassGIS Resources
  • MassGIS Home
  • Downloadable Layers
  • Master Address Database Layers
  • Statewide Interactive Map
• Municipal GIS Systems
  • Belmont
  • Boston
  • Cambridge
  • Watertown
  • Massachusetts Municipal GIS Gallery
• QGIS — Open Source Mapping Software (Mapping software can be expensive, but this open source product is free, installs easily on a desktop, and it works well with the layer formats offered by MassGIS.)