This paper reports a test of the presence of embedding effects in a health care contingent valuation study. A within-subject, mixed qualitative–quantitative approach was used to identify and explain the presence of embedding in estimates of willingness to pay for vaccinations. Embedding effects persisted despite controlling for known causes and did so even among respondents who perceived the effect to be anomalous. Results from the qualitative interviews suggest that embedding effects arise for varied reasons but might be indicative of incomplete preferences. It is questionable, however, whether survey techniques can be better designed to encourage values clarification.