This paper presents some considerations of data validity and consistency for conjoint measurement implemented on the Internet. The results from a large scale ‘mega-study’, with 3715 respondents across 21 studies (Crave It! ), reveals that the vast majority of modeled results from each of the studies are statistically significant, with high R2 values. The use of the R2 statistic to assess data quality (viz., respondent consistency) provides a way to better assess the quality of results on an individual respondent basis, after the study has been run.