Good point, this is exactly how it was decided:

The 1000 Genomes Project studied a total of 2,504 samples, about 500 samples from each of five continental ancestry groups, with generally five populations for each group. For the samples with ancestry from Europe, East Asia, and South Asia, populations across the geographic range had about 1% FST. The amount of divergence among African populations is much higher; it would require sampling many more populations to capture rare variation adequately. With a limited budget, the project decided to focus on populations related to the Yoruba, but not to attempt to be comprehensive within Africa. Other projects, such as the African Genome Variation Project2 and the Human Health and Heredity in Africa (H3Africa) Project3 are studying variation in additional African populations. For populations in the Americas, known admixture among groups with ancestry from Europe, Africa, and the Americas led the project to adopt a different strategy. Two populations had primarily African and European ancestry (ASW from the U.S. and ACB from the Caribbean) and four populations (MXL, CLM, PUR, PEL) with a wide range of European, African, and indigenous American ancestry chosen to represent the wide variation in ancestry proportions observed in North, Central, and South America.

The project ignores many populations in Asia, America and especially north Africa which is another niche for SNPs totally different from te rest. My guess on the focus in Yoruba is the human ancestor origin.

Taking in consideration the strategy, the project was not limited to only Caucasian populations and cannot be taken as the only source for SNPs.

The project was published in Nature: A global reference for human genetic variation | Nature