Obviously it's based on the samples they tested. So 1/3 of that sample showed white americans were part african
It explained clearly how the one drop works:
"
Whites’ Skin Tone as Function of Admixture
The combination of narrow phenotype variation (skin tone) along with a wide range of inter-population (Afro-European) admixture variation among White Americans shows that a selection process has taken place. Few human populations display such a clear mark of selection. Narrow phenotype variation alone does not necessarily indicate selection. Northern Europeans display little skin tone variation, but they lack a wide range of African admixture. Broad genotype variation alone does not necessarily indicate selection. Puerto Ricans average 50-50 Afro-European admixture, but they also display a wide range of skin tones. The late Stephen J. Gould, Harvard biology professor and columnist for Natural History magazine, used to explain this principle with a baseball analogy. Plot a scatter diagram of the batting averages of a thousand amateur or minor-league players and you will find a very large range of variation. A few such players are very bad, a few are very good, and most spread across the entire range of batting averages in-between. Now plot the batting averages of professional athletes in the major leagues. All are very good indeed. More importantly, the range of batting averages among them is tiny. The difference between an outstanding star of the game and a rookie is a matter of mere hundredths of a percentage point. The reason, of course, is because you cannot get into the majors unless you are very good at it. Similarly, wild cows vary greatly in the amount of milk that they produce. The cows in a dairy farm produce more milk on average but, more importantly, their milk production varies very little among themselves (compared to wild cows). The reason? Those cows who do not make the cut become hamburger.
And so, why do few if any White Americans display a strongly African appearance (have a high melanin index) despite having detectable African admixture? Because those Americans who “look Black” are assigned involuntarily to the Black endogamous group, whatever their genetic admixture. The scatter diagrams of the two endogamous U.S. groups are not symmetrical because the selection process acts only upon the White group. As revealed in court records, discussed elsewhere, a person of mixed ancestry who “looks European” (like Dr. Shriver or his maternal grandfather) in practice has the option of either adopting a White self-identity, thus joining the White endogamous group or a Black self-identity, thus joining the other group. But a person of mixed ancestry who “looks African” lacks such a choice. U.S. society assigns such a person to membership in the Black endogamous group, like it or not.25
In conclusion, U.S. society has unwittingly applied selection pressure to the color line. The only American families accepted into the White endogamous group have been those whose African admixture just happened not to include the half-dozen alleles for dark skin (or the other physical traits associated with “race”). Since those particular alleles were sifted out of the portion of the White population that originated in biracial families, the relative percentage of the remaining, invisible, African alleles in this population cannot affect skin color. That skin-color does not vary with African genetic admixture among American Whites, despite their measureably recent African admixture, demonstrates and confirms that physical appearance has been an important endogamous group membership criterion throughout U.S. history. It has resulted in genetic selection of the White U.S. population for a European “racial” appearance, regardless of their underlying continent-of-ancestry admixture ratio."
A mulatto is a person of BLACK and WHITE ancestry. Two mulattoes getting together and producing a kid, the kid is still mulatto.
then you get proved wrong and it was all in your mind 
