Abstracting is often more effective than de-coupling for these kinds of discussions. The act of changing concrete variables about a "case" requires that one keeps everything else (and their value-laden implications) the same. It is also much easier to mentally manipulate symbols when they are free of moral/emotional weight.

In the example given in the article, the question of IQ heritability can be "safely" abstracted into a population study question without referencing specific groups or populations. Of course, a serious biologist would use higher-resolution concepts like "breeding population" with defensible boundaries.

It seems disingenuous (or maybe naive?)to expect discussions about American racial groupings to remain neutral and freely "de-couplable".

It's like honestly expecting people to adhere to "is !== ought". There is a political implication to any conclusions derived about race, as it is mostly a sociological concept. Discussing it is especially difficult because the people who discuss racial differences (as opposed to other groupings like: family, class, height, religion, etc...) generally employ it in a political context (eugenics, affirmative action, etc...)