Human races do exist in nature

[Hans]

... or at least that's what one individual claimed on a forum.

 

This individual explained that the number of human races is objectively quantified by using "Bayesian K-means clustering and determining the correct K by inferring the second order rate of change in the log probability of data between successive K values. The correct K seems to be 7 which also matches the phylogeographic criteria for race perfectly. Europeans, surprisingly, form a distinct race."

 

Unfortunately I lack qualifications in the field of mathemathics to verify these claims, so I need opinions of others on this.

---

Merged post follows:

Consecutive posts merged

 

I don't know if this helps or distracts, but here's some sort of diagram illustrating that K=7:

 

Edited January 29, 2009 by Hans
Consecutive posts merged.

[CharonY]

The graphic alone is not sufficient data. At the very least one have to include what was used as a marker. Ideally refer to the publication please.

Just to add, what generally is depicted are certain allele distributions or microsatellite studies. While they show the distributions of the markers in question they do not validate a notion like race per se. Also I have to add that K-means clustering essentially groups your data into K groups. You can either estimate K a priori, or by using calculations as indicated in the OP. They means to estimate K are not undebated, though and often depends on the application of the clustering. In this case, for instance it is likely to be sensitive regarding the genetic markers used.

 

Just from the wording I am somewhat sure that it was ripped out of a legitimate study and "race" was subsequently inserted. Maybe search and replace of population with race.

Edited January 29, 2009 by CharonY