Scott Alexander gave a 95% probability that fewer refugees would be admitted in 2017 than in 2016.

Compare the number of refugees admitted to the United States in 2017 with the number in 2016. Calendar years are the primary basis, as the claim names years. US fiscal years (FY2017 vs FY2016) are an acceptable equivalent if calendar-year data are unavailable. TRUE: official admission data show fewer refugees admitted in 2017 than in 2016 on the calendar-year basis, or on both bases. MOSTLY TRUE: the data show fewer admissions on the fiscal-year basis, while calendar-year data are unavailable or the two bases show only a marginal difference. MIXED: the two bases disagree, with one showing a decline and the other not. FALSE: official data show that 2017 admissions equalled or exceeded 2016 admissions.

Set before the research began.

1. E1 Annual Flow Report Refugees and Asylees: 2017, March 2019 DHS Office of Immigration Statistics · March 2019

   “In 2017, the United States admitted 53,691 refugees, a 37 percent decrease from the previous year, due in large part to additional security vetting procedures.”

2. E2 Annual Flow Report Refugees and Asylees: 2017, March 2019 DHS Office of Immigration Statistics · March 2019

   “In this report, years refer to fiscal years (October 1 to September 30).”

3. E3 2017 marks historic drop in refugee resettlement ABC News · 7 July 2018

   “In 2017, the U.S. resettled 33,000 refugees, a 66 percent drop from the 97,000 individuals it resettled in 2016.”

4. E4 Refugees and Asylees in the United States Migration Policy Institute

   “This represents a 58 percent decrease from the 53,716 admitted in 2017”

The DHS Office of Immigration Statistics reports that the US admitted 53,691 refugees in 2017, a 37 percent drop from the year before (E1). That report counts in fiscal years (E2), so the fiscal-year basis shows a decline. ABC News reports 33,000 refugees resettled in 2017 against 97,000 in 2016 (E3), and the Migration Policy Institute gives 53,716 admitted in 2017, close to the DHS figure (E4). Neither the ABC nor the MPI quote says whether its years are calendar or fiscal, so the calendar-year basis is not clearly settled. Every figure points to a decline, but because the calendar-year data are not confirmed, this resolves as mostly true under the criteria.