Since this topic has come up quite a bit on Twitter, I thought it prudent to settle the question of what the transracial and international adoption studies actually show.
The typical sets of studies (on black individuals) cited in these ‘debates’ are the Minnesota Transracial Adoption Study (Scarr & Weinberg 1976; 1977; Weinberg et. al 1992; 1994), the German studies (Moore 1986), and an adoptive admixture study (Tizard et. al 1972). There are also a select few data on East Asian IQs, but these are the least relevant (and worst done). These data have been fought over time and time again, so I suggest reading the recent Thomas (2016) paper on the topic, which gives a very clear exposition of how they are either ambiguous, uninterpretable, underdetermined or do not support the hereditarian hypothesis.
There are two other transracial adoption studies that I have read about. The first is Buchanan et. al (2009), who reported that “East Asian participants did not score significantly differently than Caucasian participants and participants of other ethnic groups on any of the outcome measures of interest, with the exception of the internalizing cluster”, where one of their outcomes was IQ. The second is Kirkegaard et. al (2019) , who report that the IQs of black-Asian, white-Asian and Asian-Asian parent pairings as roughly equal.
What’s interesting in the recent aspects of this ‘debate’ is the resurrection and ‘discovery’ of a number of international adoption studies, where country of origin can be used as a proxy for race. Common examples cited are Dalen et. al (2008), Lindblad et. al (2010), Odenstad et. al (2008), which all use the same corpus of data (Swedish conscription data) to estimate racial gaps in academic performance. These three studies shown the ‘typical’ ordering of Koreans scoring better than native Swedes, who score better than adoptees from non-Korean countries.
There are several issues to consider here in interpreting these data. The first is that adoptive parents are not representative of the general population, as they are known to be more encouraging of students in school (Dalen 1995, 2001, 2005), have higher socioeconomic status (Hjern, Lindblad & Vinnerljung 2002; Juffer & van IJzendoorn 2005; Lindblad, Hjern & Vinnerljung 2003; Verhulst, Althaus & Versluis-den Bieman 1992; Rutter et. al 1998; Verhulst et. al 1990), etc. As a result, we already have a prior expectation that adoptees as a class should have higher cognitive-related outcomes. This, in addition to the unique factors spurring Korean adoption in the first place (Kim & Jefferson 2009, Selman 2000), provide an ample explanation for their outcomes. In addition, it has been documented that racial stereotyping of East Asians individuals has lead to their placement in particular schooling environments (Lee & Zhou 2015, p. 18-19). This could result in higher school retention rates and increased educational attainment (Booij et. al 2017), both boosting cognitive outcomes in the longitudinal studies (Ritchie et. al 2018, Ceci 1991, Brinch 2012) . Because of the unique circumstances of Korean adoption programs, adoptive parents of Korean adoptees seem to have much greater knowledge of early childhood experiences and environment (Welsh et. al 2008, p. 189-190). Finally, Koreans adopted from Korea and Koreans born in places like America have been shown to differ on variables related to discrimination, school belonging (Seol, Lee 2016, p. 299), indicating that adoptees are a selective sample.
The other thing that needs an explanation here (in these studies) is the observation that non-Korean adoptees had lower outcomes. Despite the lower outcomes, one notes that there is a 60% reduction in the gap between PISA-expected scores and observed outcomes for these adoptees  (Caplan 2017). This provides a succinct demonstration that this portion of the gap is environment, and the result is up to explain. The first factor we should consider is the prenatal environment, which has been shown to have immense influences on all sorts of behavioral outcomes. This undoubtedly differs between different countries, especially with regard to the nutrition that gestators receive during gestation, beginning the process of differential epigenetic imprinting (Carpenter et. al 2018, Devlin et. al 1997, Harris & Seckl 2011) and brain growth (Welberg et. al 2001, Toledo-Rodriguez 2010),. In addition, Korea and non-Korean countries (which were reported in the Swedish paper to range from India, Thailand, Ethiopia, and Sri Lanka) differ in their early environments, which have been shown to have profound influences on the trajectory of development (Beam et. al 2015, Smythe et. al 1994), meaning that the Korean and non-Korean adoptees differ in other important aspects that can explain their difference.
Finally, there are several studies that come from international adoptee data that do not support the racial ranking interpretation that some have forced out of these papers. For instance, Lindblad, Dalen, Rasmussen, Vinnerljung & Hjern (2009) reports that the (white) siblings of adoptees had higher scores than the Asian adoptees, meaning that the ranking would then be white > Asian > non-Asian adoptee. Similarly, Dalen et. al (2008) (p. 1215 Table 2 + 3 & p. 1216 Table 4) and Vinnerljung et. al (2010) reported larger means for siblings of international adoptees than Korean adoptees. Even more, Kristine et. al (2014) reports in table 3 that adoptees from Korea/China had about equal motor scores to the Colombia/Other group, while the Colombia/Other group had slightly higher values on the communication task (Kristine et. al 2014, p. 34).
 Kirkegaard. et al (2019) claim that early environment effects do not persist into later life, which might be true for the general population, but seems to be particularly false for adoptees (Beckett et. al 2006). They also incorrectly claim that Korean adoptees are selected to be lower than their countries genotypic IQ, which does not comport with the literature’s understanding of the unique role of Korea in international adoption (Ja Sook Bergquist et. al 2015). They also fail to consider that this is an ad hoc interpretation. Finally, they appeal to cutoff scores in an attempt to dismiss the results of Eyferth and their own scores; a dismissal that can be rectified by referring to Flynn (1980).
[1.5] Kirkegaard et. al also seem to deny the fairly-consistent result that age of adoption is related to outcomes (Bakersman-Kranenberg 2008, Lien et. al 1977, Julian 2013). Though some studies do not detect it, this is to be expected given that age of adoption is supposed to be a proxy for early environment, which is itself very heterogeneous.
 We should note that a small difference in environment can lead to substantial differences in phenotype through phenotype-environment feedback loops (Dickens & Flynn 2001; more research reviewed here). Differential ethnic developmental pathways based on the differences in environments can cause these outcomes.
 This is actually an underestimate of the closing of the gap, as we are comparing observed PISA scores for the entirety of African countries, which includes all class strata, to the outcomes of the selected adoptees, which are likely to come from worse environments and genetic backgrounds. Their counterfactual scores would then be even lower, and subsequently a greater closure in the gap.