methods

Methods

In November 2026 we collected the documents on antisemitism issued by cities and other local governments in North America, Europe, Australia, Latin America and elsewhere, from their official websites and council records.

Coverage

The corpus is the set of documents we collected that name antisemitism and were issued by local governments. The form of local government differs by country: cities and municipalities in most of Europe and Latin America, borough and city councils in Britain, local councils in Australia (Glen Eira and Waverley among them), and in the United States both cities and counties, such as Miami-Dade County. Four kinds of document qualify: (1) plans and strategies, (2) council decisions that set out measures, (3) mandates for an office, commissioner or task force, and (4) adoptions of a definition of antisemitism, which we kept even where the adoption comes with no other measure. Berlin, Hamburg, Bremen and Vienna are each both a city and a state, and we kept their documents and marked them as city-state documents.

State, provincial and national plans are not part of the corpus. We list those we came across, marked as such, so that a city's documents can be compared with the plan of its state or national government. The document list shows them only when that level of government is selected.

Gathering the documents

With the help of AI agents (Claude, a model made by Anthropic), we searched in successive rounds in November 2026: the largest cities of North America, Europe, Australia and Latin America first, then the cities not yet searched, then Eastern and Central Europe, Asia and Africa.

  1. We searched the city's own website, its council records and the wider web, in the city's language, for plans, strategies, motions and executive orders naming antisemitism.
  2. We took each document from the issuing government's website, or from the council agenda or minutes where the document was adopted as an attachment. Where a document exists only as a web page, we saved its text with its address.
  3. We opened each saved file to confirm that its title and date match the document named, rather than relying on the file name.
  4. We recorded the title in its original language, an English gloss, the issuing body, the date, the kind of document, the official address and, for long agendas, the page or item number.

The team has not yet checked the entries by hand, document by document. We copied titles and dates from the documents, and where we could not confirm a date from the document itself, we say so in the entry's notes.

Kinds of document

We sorted each document into one of nine kinds by its form, without assessing its content.

Definition of a plan

A plan, in this project, is a document issued or adopted by a local government that gives a narrative and states the government's own intentions. The narrative is the document's own account, in prose, of the problem it addresses: what antisemitism is, how it appears in the place, or why the government acts. A plan may state its intentions as goals and priorities, as specific measures with an owner and a budget, or at any level of detail between the two. A plan that sets goals without naming measures is still a plan, and we recorded the level of detail of each plan as a variable.

We kept a document that fails either test as a related document of the place that issued it. A directive, such as an executive order, a resolution or a council motion, lists actions without a narrative. We counted a motion as a directive, however many actions it lists, when its prose is a background or summary of about a page or less. A statement, a report or a progress report gives a narrative without the government's own intentions. We typed each related document as a directive, a statement, a report, a progress report on a plan, a mandate for an office or post, a request to another level of government, a funding decision or a meeting record.

Plans that a national or regional government issued for a territory, such as the Paris prefect's plan of 2017 and the Brussels-Capital Region's plan of 2023, are not local plans, whereas a city that is also a state, such as Berlin, Hamburg or Bremen, is a local government. We included a plan on racism or discrimination only when it names antisemitism in its own text, since we did not set out to catalogue anti-racism plans as such. We kept a plan proposed and never adopted, such as the Bogotá bill of 2024, and marked it as not adopted. We treated the draft, amended and adopted texts of one plan as versions of a single plan.

We read each candidate document against the two tests and recorded the passage that shows the narrative and the passage that shows the intentions, each with its page. We read every motion, resolution and council decision first sorted as a plan a second time under the test for motions. The team has not yet checked the sorting by hand.

IHRA decisions and pledges

For the decisions on the IHRA definition, we combined the documents our searches found with the lists of local adoptions kept by the Campaign Against Antisemitism for Britain, the American Jewish Committee for the United States, and the Combat Antisemitism Movement for other countries. We marked a row taken only from one of these lists as taken from a secondary source, and where two lists give different dates, we recorded both in the notes. For Britain we also listed every principal council that existed between 2016 and 2026, from the Office for National Statistics' lists of local authorities, so that councils that did not adopt are counted too. We drew a random sample of 60 rows dated from the lists and found a council record to compare for 16 of them. Of those 16, 12 matched to the day and 4 to the month. We could not check the others, mostly because the council's site blocked automated access.

We took the pledges from the signatory lists of the American Jewish Committee's campaigns and from the programmes and press releases of the Mayors Summits Against Antisemitism. The campaign lists give no signing date for any city, and no summit published a list of signatories, so a summit row records that a city attended.

We loaded the text of every saved document, page by page, into the search page. We took the text from each PDF's own text layer. For the nine scanned documents without one, we ran the Tesseract text-recognition program in each document's language, and their text may contain errors. The search ignores capitals and accents on every page.

Word profiles

For the word profiles, we counted the words of 15 topics in the original-language text of each local plan on antisemitism and each wider local plan that names antisemitism, and gave each count per 1,000 words. The topics include Israel, the Holocaust, schools, the police, money and sport. For each topic we wrote a list of word stems in English, French, German, Dutch, Danish, Swedish, Spanish, Catalan, Italian, Portuguese and Polish, and matched the stems after lower-casing the text and stripping its accents. The lists are printed with the word profiles. A script made the counts, which show which words a plan uses and how often. The coders, working from the coding framework, recorded what a plan says about each topic. We did not profile plans that a national or regional government issued for a territory, such as the Paris prefect's plan of 2017 and the Brussels-Capital Region's plan of 2023, since neither was issued by a local government.

Definitions named in the text

For every saved document with a text layer, we searched the text, in the document's own language, for three definitions of antisemitism by name: the working definition of the International Holocaust Remembrance Alliance (IHRA), the Jerusalem Declaration on Antisemitism and the Nexus Document. We recorded, for each document, the number of mentions of each definition and the first passage that names it. The search finds which definitions a document names. A document may name a definition in order to adopt it, to set it aside or to describe a debate, and New York City's 2026 strategy, for instance, names all three. We could not search saved files that are scans without a text layer, and we say so in their entries. The coders recorded the definition a document adopts as a separate variable in the draft coding framework.

The same search counted how each document spells two words: antisemitism, closed or hyphenated as anti-Semitism or anti-semitism, and anti-Zionism, closed or hyphenated, together with the equivalent forms in French, German, Dutch, Spanish and Portuguese. In German, Dutch, Swedish and Danish the closed spelling is the standard form, so the count varies mainly across English and French documents. The search does not count a word broken across a line in a PDF.

Jewish population

For each place in the corpus we recorded the size of its Jewish population as context, with the geography, year, source and kind of measure for every figure. The kind of measure differs by country: census counts of religion for local governments in Britain, Canada, Australia, New Zealand, Switzerland, Mexico and Chile, and for Brazil from the 2010 census; county estimates from the American Jewish Year Book for the United States, where the census asks no question on religion; registered membership of the Jewish communities in Germany; and metropolitan estimates by Sergio DellaPergola for a few other cities. A membership count includes only registered members, and a metropolitan estimate covers the whole metropolitan area, so neither equals a census count of the place, and the figures compare more readily within a country than across countries. Where we found no figure for the place itself, we say so in the entry and give no borrowed national figure.

For three countries we also listed every local government, so that adoption of the IHRA definition can be given as a share of all councils: the 448 principal councils of Britain since 2016, the 737 Canadian census subdivisions with 5,000 people or more in 2021, and the 541 Australian local government areas with a council in 2021, each with its Jewish share from the national census (tables on the Diffusion page).

We did not record counts of antisemitic incidents for each place, since police forces and monitoring bodies record incidents differently in each country and each force. The coders recorded instead, as a variable in the coding framework, the figures that each plan cites as evidence of the problem.

Editions

Many plans have been renewed, and a search for a city's plan often returns an older edition first. We therefore keep each edition as its own entry with its own date. We marked the newest edition we found as current, and where we could not establish that no later edition exists, we marked the entry unconfirmed. The Hague's plan, for instance, has two entries: the February 2025 edition, which used the IHRA definition, and the June 2025 revision, which uses the Jerusalem Declaration on Antisemitism. We listed separately the documents New York City issued under Mayor Adams and those it issued under Mayor Mamdani, since Executive Order 1 of 1 January 2026 revoked the earlier orders and Executive Order 2 of the same date re-established the Mayor's Office to Combat Antisemitism.

Gaps

We could not save some documents, because the website showed an automated-access check or refused the download. Their entries link to the official address and are marked not saved. Where a site blocked automated downloads but opened in an ordinary browser, we saved the document through a browser, and where a server did not answer at all, we saved the Internet Archive's copy and recorded the date of its capture in the entry's notes. For each official address we also asked the Internet Archive for a dated copy, and an entry links to that copy where one exists, so that a document remains available if the government removes the page. Other entries are leads: documents reported in council minutes or the press that we could not retrieve. We marked the leads as unconfirmed.

Places searched without result

For each place listed, we searched and found no city document on antisemitism. The absence of a place from the corpus means only that our search found nothing, and the notes against each place in the inventory state what we checked.

Qualifications

a project of

This site is a team project of the Lab for the Global Study of Antisemitism, housed in the Anne Tanenbaum Centre for Jewish Studies at the University of Toronto.

Ron Levi and Sol Goldberg
ron.levi@utoronto.ca

Research Team: Cypress Chernik, Lir Josefsohn, and Arianna Seijts, all of whom are undergraduate interns at the Lab for the Global Study of Antisemitism.

This is a work in progress. Corrections, missing documents and requests to reuse the data are all welcome.