I am running HB regression where I only have one row for each respondent. Each respondent was show a subset of features and asked a behavioral (agreement scale) question. The subset of features could differ for each respondent based on previous selections in the survey.
Does the HB reg algorithm 'borrow' information from other respondents in the set to derive the part worths? When I run the regressions, I get results, what is the level of confidence from these results?
Please let me know.