Definitions and Methodology

This page defines many of the terms found in the Security Map’s Legends.

Methodology

Data Collection

The Security Map primarily draws on news articles and data from federal court records. It also summarizes a multi-method database that relies in part on interviews and surveys. This methodology describes how newspaper articles appearing as dots on the map were collected and analyzed by The Impact Project. The latter primary sources—data from federal court records, interviews, and surveys—were not gathered by The Impact Project; those methodologies can be requested here and here.

News articles were identified through systematic Boolean searches conducted by three researchers, each assigned a set of key terms and responsible for completing searches across all 50 states and the District of Columbia.

Data Searches

This table describes individual searches constructed by including one term from columns 1 and 2. Terms appearing on the same row were not exclusively paired together.

1. Action 2. Public Servant Type
threatpublic official
warngovernment representative
swatgovernment worker
doxelected official
stalkjudge
harrasscivil servant
assaultpoll worker
violencemilitary
abusepolice
menaceBoard of education
damagefederal worker
rapecongress
legislator
senator
law enforcement officer
judicial officer
school board
election worker
politician
detective
sheriff
lawmaker
prosecutor
librarian

Searches were organized by “subject of threat,” and researchers conducted Boolean searches using combinations such as (“threat” OR “violence” OR “assault” OR “stalking”) AND ("federal worker" OR "federal employee") AND (“injury” OR “firearm”). Searches were performed in Google News for the period beginning on January 1, 2015 and ending on September 1, 2025. The searches we used were recorded, and the spreadsheet of results was deduplicated by URL to ensure each story was reviewed only once.

Data Coding

To ensure consistent coding across researchers, the three researchers independently coded assigned blocks of new stories in dropdown menus in the database. For each story, the following information was recorded to the extent it was included in the news stories: area affected (city or county), state, individual threatened category (e.g., judge or law enforcement), job of the threatened person, agency of the threatened person, agency program of the threatened person, political affiliation of the threatened person, type of threat, whether or not the threat was acted upon, actions taken in response to the threat (e.g., investigation, arrest), number of people threatened, and source. Quotes describing threats were also included. Coding was based strictly on the story’s content—e.g., a mention of “stalking” was coded as “stalking”—to minimize subjective interpretation and enhance consistency. Multiple threat types were coded when applicable; for example, if a judge experienced harassment and stalking, the story would be tagged as “judge,” “harassment,” and “stalking.” 

When researchers found sufficiently detailed data, each threatened individual received their own entry into the database. If the precise number of threatened persons was not included in an article, then each threatened category received its own entry (e.g., if an article reported that county judges were threatened, but the number of those threatened was unclear, researchers coded one entry for “county judges”). Researchers assigned only one threat category (e.g., judges, elected officials) per entry, unless the threat includes a primary target and their staff, office, or family. In these cases, we coded the primary, secondary, and tertiary threats in a single row (e.g., judge & family members). Researchers coded news articles to the location of the threatened individual when that data was available. If either the location of the threatened individual was unknown or the perpetrator threatened a large number of people over a diverse geographic area (e.g., 100 judges across 50 states), researchers coded the locations of the perpetrator. After initial review, each researcher was assigned a different set of articles to review coding accuracy and reconcile discrepancies.

Below is a list of categories available in the map’s dropdown menus, which can be used to filter and sort incidents by various attributes.

Type of Threat Public Servant type
HarassmentElected Official
Threatening StatementPublic Official
StalkingElection Official
DoxxingLaw Enforcement & Military
SwattingJudge
Physical AttackSchool Board Official
Suspicious PackageFamily Member
VandalismOffice
Death to Threatened PartyOther
Other

Baseline Data

Baseline data describes what users see when hovering over a state rather than an individual dot. The baseline data includes a summary of threats collected by The Impact Project, as well as data from Princeton University’s Bridging Divides Initiative (BDI), Chapman University, and the National Counterterrorism Innovation, Technology, and Education Center (NCITE) at the University of Nebraska Omaha.

“Local Threats”: Local Threats refers to the data collected by Princeton University’s Bridging Divides Initiative (BDI) in their Threats and Harassment Dataset (THD). BDI collects secondary information stemming from open-source investigation of traditional and social media; and integrates a variety of data from BDI partner organizations to track reported threats and harassment incidents against local officials around the country, from elected officials at the municipal, county, and township levels to appointed officials and election workers. BDI updates the THD dataset each month. In addition to the THD, BDI also seeks to understand threats and harassment against local officials via surveys (in collaboration with CivicPulse) and via interviews with local officials as part of a larger mixed-methods project. BDI’s data is not included as dots on the map, but is included in the summaries that appear when users hover over a state. BDI’s assessment of news articles is distinct from the methodology described above.

“Cases Filed”: Cases Filed refers to the data collected by Chapman University and the National Counterterrorism Innovation, Technology, and Education Center (NCITE) at the University of Nebraska Omaha. Chapman University and NCITE built this archive using federal court records from the Public Access to Court Electronic Records (PACER) database from across all 94 federal districts, as well as publicly available data from the Federal Bureau of Prisons and Department of Justice. Between 2013-2022, they identified more than 500 individuals charged in federal court for threatening public officials.

Impact Project News Analysis

The total number of Federal, State, and Local government employees as described in The Security Map is pulled from the U.S. Bureau of Labor Statistics’ website at BLS.gov.

Definitions

Public Servant Type 

Elected Official. A local, state, or federal employee who was elected to their position

Public Official. A local, state, or federal employee who was appointed to or selected for their position without an election

Election Official. A local, state, or federal employee whose job involves managing elections at the county, state, or federal level

Law Enforcement & Military. A local, state, or federal employee responsible for enforcing laws (includes state police, local police, the FBI, ICE agents, prosecutors) and military; notably, this definition excludes judges

Judge. A local, state, or federal employee appointed to decide court case outcomes

School Board Official. Local or state employees who serve on school boards or boards of education

Family Members. Members of a public servant’s family.

Office & Staff. A public servant’s staff or physical office.

Other. Any local, state, or federal employee who does not fall within one of the above discrete categories.

Threat Category

Harassment. Repeated threatening statements and actions, including numerous phone calls, voicemails, emails, or letters or verbal abuse, that cause annoyance, distress, or frustration

Threatening Statement. Expressions of intent to harm that may be verbal, written, or online

Stalking. Unwanted, obsessive attention from an individual that may include the perpetrator following the victim, cyberstalking, repetitive messaging, etc., that causes the victim to fear for their safety 

Doxxing. Release of a victim’s private or identifying information, which may include their full name, home address, office location, etc.

Physical Attack. A physical action on a public servant, including assault, battery, attempted kidnapping, kidnapping, attempted murder, and sexual violence

Vandalism. Deliberate destruction of or damage to public or private property related to a public servant’s presence, job, or actions

Suspicious Package. The delivery of unidentified material to a public servant or government office with the intent to cause annoyance, distress, or harm

Swatting. A hoax call to emergency services in an attempt to cause (whether it happens or not) an armed law enforcement dispatch to an address

Death to Threatened Party. Death of threatened public servant related to threat

Other. Threats that do not fall into one of the above discrete categories

Response to Threat

Case filed. Case filed against alleged perpetrator in federal court (PACER via NCITE 2013-2022). A follow-on version of The Security Map tracking local responses to acts of political violence will be released in the coming months.

Government Level

Federal. Employees who work for the federal government (e.g., members of Congress and their staff, federal judges, postal workers, etc.)

State. Employees who work for the government at the state level (e.g., governors and their staff, state legislators, state police, etc.)

Local. Employees who work for the government at the local level (e.g., county officials, mayors and their staff, school board members, etc.)

** Note that threats also occur on Tribal lands and against tribal government officials. However, these incidents are not well represented in newspaper articles accessed for this initial study.