Have an idea?

Visit Sawtooth Software Feedback to share your ideas on how we can improve our products.

HB vs OLS for estimation of individual utility: CVA

Generally, would HB estimation for CBC and OLS for CVA be the most recommended way?

If I use HB for estimating individual utility values of CVA and use zero centered diff utility values as the dependent variable, is it less effective than OLS derived ZCDiff utility values?

I read the relevant technical papers and ran both.

Although they seem to present similar results, the utility values are different and I can't seem to decide which one would be better?

Thank you so much for your advice.
asked Jun 29, 2020 by anonymous

1 Answer

0 votes
Generally HB-MNL estimation (HB, with a multinomial logit formulation of the lower level model fit) is preferred for CBC data.  This is what our Lighthouse Studio software does by default when you ask for HB estimation.

For CVA (traditional ratings-based conjoint), our Lighthouse Studio platform also has built-in capabilities for HB-OLS estimation (HB, with an OLS formulation of the lower level model fit).  And, this is preferred to to a non-HB individual-level OLS approach.

I don't understand your question about zero-centered diff utility values as a dependent variable.  But, I'm guessing that you may be thinking about building a subsequent correlation or regression model where you are using different predictor variables outside the conjoint experiment to predict the preference results of the conjoint analysis.  Zero-centered diffs would be preferred for that, since it normalizes the range of utility scores per respondent to be comparable across respondents.

OLS-derived zero-centered diff utilities probably have more noise and overfitting than HB-OLS derived zero-centered diff utilities.
answered Jun 29, 2020 by Bryan Orme Platinum Sawtooth Software, Inc. (189,140 points)
Dear Bryan, Thank you so much for a clear explanation. One last thing though, when you say HB-OLS derived zcdiff is the best option.
In running the analysis manager, I need to put analysis types as HB and make the recode method as LOGIT to get the HB-OLS, correct?
I really appreciate your wisdom. Thanks again and take care.
No, the choice you make for how to transform or not transform the respondent's answers (the dependent variable) is independent of the choice of whether to run HB-OLS or individual-level OLS.
Dear Bryan, Thanks for such a quick response. Then, simply selecting analysis types as HB stands for HB-OLS and selecting analysis as OLS would run individual-level OLS, correct?
Dear Bryan,

I can't thank you enough for your response.

I wish you all the best and stay well.