Oddly enough, the IAT itself has major problems with the primacy effect - reversing the mappings causes confusion, producing a bias in the data that needs to be carefully excised. So, using the IAT to test the strength of the primacy effect in regards to something else is going to be extra-tricky.
"First is Best"
While this effect has been known about for many years, these researchers added an interesting component, an "Implicit Association Test (IAT)":
and
I was surprised to find there is no reference to "recency", "primacy" or "serial position" on the LessWrong Wiki. A search on LessWrong.com for "recency effect" turns up 8 posts that mention it but don't give it a thorough discussion as far as I can tell; "primacy effect" turns up 1 post about Rationality & Criminal Law; and "serial position" turns up nothing. Is there another name for this effect that I'm missing?
Wikipedia has some discussion of the serial position effect here, although from a quick skim it doesn't appear that they talk about preference at all.