If you want to use Bayesian statistics to investigate the effect of covariates on utilities, then review the paper: http://www.sawtoothsoftware.com/downloadPDF.php?file=HBCovariates.pdf
If you want to use Frequentist statistics to do this, then you should use the zero-centered diffs normalized utilities. You'll find that our software outputs point estimates (single accumulated set of scores per respondent) for the zero-centered diffs. Use zero-centered diffs, because it is an attempt to get rid of the large scale factor differences that can be observed across respondents. Then, read the zero-centered diffs utilities along with the segment membership info into a standard cross-tab package (or statistical package) and conduct F-tests, T-tests, or ANOVAs between segments.