For each pair of states, every selected resolution both voted on scores 1 if they voted the same,
½ if one abstained and the other voted yes or no, and 0 if one voted yes and the other no.
A pair's coincidence is the average over those shared votes, from 0 to 1 (the agreement index; yes = 1,
abstain = ½, no = 0). Absent states, states present but not voting and states without the right to vote
(arrears, South Africa 1974–94, Yugoslavia 1992–2000) are skipped for that vote, as are non-members.
A pair needs at least 20 shared votes for a score, or half the selection (at
least 5) when it has fewer than 40 resolutions; below that
it shows “too few shared votes”.
Time series. A year's score uses that calendar year's votes (sessions run September to
December) and needs 5 shared votes. The 3- and 5-year lines are centred windows pooled
over all their votes, not averages of yearly scores, so a thin year weighs what its votes weigh. The
dashed line is the comparison country's mean score with every state. Biggest movers compares
each state's score over two periods, each needing as many shared votes as a pair needs for a score.
Options. Yes/no votes only drops, pair by pair, every vote where either state abstained.
Exclude lopsided votes drops resolutions that passed with 90% or more yes: about four in ten
recorded votes did, and on those almost every pair looks aligned. Cohen's κ subtracts the
agreement two states would reach by chance, given how often each votes yes, abstains and no on the
votes they share (Häge 2011): 0 means no more alike than chance, 1 identical, below 0 less alike than chance.
What is counted. The 5,694 General Assembly resolutions adopted by
recorded vote from 1946 to 2025, from the UN Digital
Library's voting data. Resolutions adopted without a vote, most of the Assembly's output, are not
included: they say nothing about who differs from whom. The source lists adopted resolutions only. The
previous version of this tool used Erik Voeten's roll-call data, which has more votes in the early
decades (249 against 72 in the 1940s), so results before the 1980s can differ from the old viewer.
States over time. Predecessor seats are merged into their successor: the Soviet Union into Russia,
West Germany into Germany, Yugoslavia and Serbia and Montenegro into Serbia, Tanganyika into Tanzania.
China's seat until 25 October 1971 belonged to the Republic of China and is shown as Taiwan. East
Germany, Czechoslovakia, South Yemen and Zanzibar appear in the ranking and the network but have no shape
on today's map; in the network, Czechoslovakia flies the flag the Czech Republic kept, Serbia the
Yugoslav flag for selections ending before 2003, and the others a labelled circle.
Topics. Each resolution is assigned to topics by rules that read its title, agenda item and UN
subject headings, with a few corrected by hand; what no rule catches is Other. Topics overlap,
and the selection counts each resolution once. Checked against Voeten's hand-coded issue areas, the
rules find 90–98% of his Israel–Palestine, nuclear and arms-control resolutions. Economic
development is the weakest topic: it finds 55% of the resolutions he codes as economic development
(67% counting UN budget votes, which have their own topic here), so for economic questions, check the
titles or add a search.
Groups of states. The Alliances view covers the five UN regional groups, nine regional
organisations and eleven alliances and blocs, each with dated memberships: a pair of members counts
only the votes both cast while both belonged, so the EU of 1980 is nine states and Croatia adds nothing
before 2013. A group's cohesion is the mean score over those pairs; a member's alignment
is its mean score with the other members. Membership is formal: suspensions do not end it (Egypt in
the Arab League 1979–89, Syria 2011–23), but two exclusions that were never reversed do (Cuba in the
OAS from 1962, Venezuela in Mercosur from 2016). The UN regional groups use today's composition, except
that Israel joined WEOG in 2000 and Kiribati the Asia-Pacific Group in 2024; the United States, an
observer that counts as a member for elections, is in WEOG.
Sources: UN Digital Library, General Assembly voting data (2026_02_06_ga_voting.csv). Issue-area benchmark:
Voeten, Strezhnev & Bailey, United Nations General Assembly Voting Data, Harvard Dataverse.
Häge, F. M. (2011), “Choice or Circumstance? Adjusting Measures of Foreign Policy Similarity for Chance
Agreement”, Political Analysis 19(3). Default partners in the time series: the ten largest economies by
nominal GDP, IMF World Economic Outlook, April 2026 (2025 values).