is available) and listed their ancestry as Nepali or their language spoken at home as Nepali, Dotyali, or Dangaura Tharu.46 To ensure that its estimates are as accurate as possible, CMS considered only non-naturalized, foreign-born persons; that is, persons without US citizenship. The final estimates represent aggregated estimates of individuals matching the CMS profiles, weighted to population level, using standard Census Bureau-established person weights that take into account the ACS’s sampling design and the aggregation of 1-year samples into the 5-year microdata sample.47 Using these weights, CMS is able to produce estimates of the number of individuals matching each profile above that resided in the United States in 2017. The ACS does not permit researchers to determine definitively whether individuals are stateless or at risk of statelessness. Its questionnaires do not include explicit questions about nationality or statelessness. It also does not ask questions about non-US citizenship and, therefore, it is not possible to gauge whether ACS respondents possess nationalities other than US citizenship. As stated, individuals included in CMS’s estimates may possess the nationality of their country of origin or a third country. Thus, CMS’s estimates should be considered provisional estimates of persons who are potentially stateless persons or potentially at risk of statelessness. As noted above, also, the incidence and risk of statelessness is likely to vary significantly across profiles. Some small groups included in CMS’s profiles, notably former Yugoslavians from North Macedonia, ethnic Somalis from Kenya, ethnic Eritreans from Ethiopia, ethnic Ethiopians from Eritrea, ethnic Tamils from Malaysia, Palestinians from Jordan, and Serbians from Croatia may experience statelessness at lower rates than other profiled groups. Without additional data and individual screening, it is impossible to estimate the incidence of statelessness for each profile. The 5-Year ACS aggregates some countries of birth into larger geographic regions.48 As a result, the ACS precludes the identification of individuals born in some key countries and territories set forth in the CMS profiles, including Estonia, the Western Sahara, Mauritania, Côte d’Ivoire, Brunei, Madagascar, Turkmenistan, Tajikistan, and South Sudan. Likewise, ACS data does not capture speakers of several languages, including Rohingya, Tedaga (Tebu Libyan), Dinka, Nuer, and several minority languages of Kenya, Ethiopia, Côte d’Ivoire, and Thailand. The ACS aggregates these into larger language families, making it difficult to identify speakers who are potentially stateless or potentially at risk of statelessness with precision. ACS data also does not capture certain ethnicities from CMS’s profiles, including Sahrawi, Lebanese Bedouin, Dinka, Nuer, Burkinabé, Guinean, Tanzanian, Mozambican, and Zimbabwean individuals and members of some Thai minority groups. As in the case of languages not captured by the ACS, these ancestries are aggregated into larger categories. In addition, some ethnicities, such as Afghan Jogi and Bakassi are missing entirely from the ACS. Furthermore, ACS data does not permit the identification of Iraqi Feyli Kurds born in Iran and Chinese Cambodians born in Vietnam, as 46 The ACS permits respondents to list up to two ethnicities. Both ethnicities were taken into account in CMS calculations. 47 ACS weights incorporate Census-derived population data produced by the Census Bureau’s Population Estimates Program (PEP), including data on respondents’ sex, age, race, and Hispanic origin, to increase the precision of the estimates. Person weights also correct for differential coverage by geography and demographic detail (US Census Bureau 2014, 135-167). 48 Individuals born in the Côte d’Ivoire or Mauritania, for example, are recorded only as having been born in Western Africa. 42