What is UNGAInsight?
UNGAInsight (UNGA-I) provides new data on the politics of voting at the UN General Assembly. Historically, data on UNGA resolution content were coded for six broad categories: colonialism, arms control and disarmament, economic development, human rights, Israel/Palestine, and nuclear weapons and material, capturing important components of the US-led international order. Yet, some of the most consequential recent votes (such as votes related to Ukraine) fall outside of these categories. UNGA-I’s 32 categories allow scholars to explore critical junctures that define the contours of the order itself.
UNGA-I also helps scholars understand the targets of UNGA action. Member states have used the UNGA’s rhetorical authority to censure numerous countries, adopting 29 resolutions condemning South Africa alone for apartheid and nuclear proliferation. The 79th session (2024–2025) included resolutions calling attention to human rights concerns in Syria, Russian-occupied areas of Ukraine, Iran, Myanmar, and North Korea. Identifying instances where the General Assembly makes specific demands of a particular country provides insight into high-stakes political debates and facilitates comparisons between UNGA and the UNSC.
UNGA-I further helps scholars understand the political dynamics undergirding absences, which constitute nearly 10% of possible roll-call votes. Bailey et al. (2017, p. 432) note that absences are often sequential, and call for future extensions that might include, “more nuanced models of absences” (436). UNGAInsight follows through on this recommendation, providing important information on the broader geopolitical context and internal dynamics of many UNGA voting absences.
The most commonly used data on UNGA roll-call voting come from Bailey et al. (2017) (hereafter, BSV), who used individual country votes (yes, no, abstain) on UNGA resolutions to infer latent policy preferences, proxying for alignment with the US-led liberal order. Historically, the BSV dataset included several key variables that provided insight into the geopolitics of voting, including the total numbers of yes, no, and abstain votes, a binary indicator of whether a vote was important to the United States, and six non-mutually exclusive subject matter categories. From these data, the authors assembled a dynamic ideal point model capturing “the position of states vis-a-vis a US-led international order” (Bailey et al., 2017, p. 431). These ideal points provide a crucial foundation for scholars interested in understanding preference alignment across states and across time.
Recently, FHK introduced a new dataset (UNGA-DM) to address various gaps in roll-call voting data. Noting deficiencies in coding, reliability, and scope within previous datasets, the UNGA-DM project provides the most comprehensive dataset to date of UNGA decision-making. Not only does the dataset include resolutions adopted via roll-call vote, it also includes consensus resolutions, failed resolution drafts, amendments, motions, and separate recorded and non-recorded votes. With UNGA-DM’s impressive scope, BSV now exclusively publishes data related to ideal points, using UNGA-DM’s data to generate estimates and omitting other previously included variables such as subject-matter categories. Thus, while the underlying data for UN roll-call voting have improved in coverage and reliability, scholars now have less information than ever on resolution content via BSV.
UNGA-I also complements recent textual data on resolutions. Using machine learning techniques to analyze all UNGA and UNSC resolutions from 1946–2018, Arias (2025) offers insightful information about references to other resolutions, textual alignment, and the mix of topics discussed in each resolution. Each resolution is coded based on its proportion of 50 topics, an approach that captures rhetorical progression across time. However, because the data reflect the mix of topics within a given resolution, it may sometimes miss the core substance. UNGA-I can be usefully combined with Arias’s data, allowing scholars to explore how the content of a specific area (such as Science or Nuclear Disarmament) changes across time.