The term "bias" is unhelpfully vague, as is the term "unconscious", so "unconscious bias" is a dialectical disaster zone. The IAT tests measure association (th

The term "bias" is unhelpfully vague, as is the term "unconscious", so "unconscious bias" is a dialectical disaster zone. The IAT tests measure association (that's what the "A" stands for), not "bias". Probably I associate "Norwegian" with "tall", but that doesn't necessarily mean I'll overestimate the height of a Norwegian person who is standing in front of me. Perhaps I will, but that's an entirely different question, and the bias might go in the other direction from the association: if you're expecting someone to be tall then they might look short when they're in fact average.