Here is one I remember:
CHYMPS ABD giving talk at well-ranked dept recently presented a slide w/ regression regression results that included a binary x continuous interaction term.
Didn't discuss it when presenting results. Was asked about it by a full...didn't get it...then by an AP who tried to give her a big hint to be nice...then by an ABD at the talk who did the same. Still had no idea how this worked nor could she explain why she chose not to include one of the constitutive terms in the model.
That was pretty painful...