depends on n and m...
what's the population for europe+america?
Even China has a smaller std, but because of large population the distribution (cdf) F(max(x1...xn)<=z)=(G1(X<=z))^n maybe smaller than the cdf F(max(y1...ym)<=z)=(G2(Y<=z))^m--even if G1(X<=z)>G2(Y<=z) for some large z, i.e. the right tail of the distribution of European and Americans are longer, but the distribution of the best of a much larger chinese population would have longer right tail.