We conduct a two-phase laboratory experiment, separated by several weeks. In the first phase, we conduct urn games intended to measure ambiguity aversion on a representative population of undergraduate students. In the second phase, we invite the students back with four different solicitation treatments, varying in the ambiguity of information regarding the task and the payout of the laboratory experiment. We find that those who return do not differ from the overall pool with respect to their ambiguity aversion. However, no solicitation treatment generates a representative sample. The ambiguous task treatment drives away the ambiguity averse disproportionally and the detailed task treatment draws in the ambiguity averse disproportionally. Finally, the standard laboratory recruitment e-mail disproportionately draws in those who are not ambiguity averse. (JEL A12, C81, C90)