Have an idea?

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

for HB results how to do two-way analysis of variance

It is easy to do one-way analysis of variance with covariates for e.g. gender and university. However, how can I calculate if the interactions of gender and university  are significant.
asked Jan 3, 2017 by merjah (180 points)

1 Answer

0 votes

ANOVA is something you'll need to run in your general purpose stat testing software (SAS, SPSS, etc.).
answered Jan 3, 2017 by Keith Chrzan Platinum Sawtooth Software, Inc. (110,575 points)
How do I make the different respondents utilities to be on the same scale. To make them to be of same length?
You can use the ZCD transformed utilities instead of the raw utilities.  

If you need to do this calculation, it is described here:  https://sawtoothsoftware.com/forum/6140/is-there-a-formula-for-calculating-the-zero-centered-diffs
Thank you! I am doing the two way analysis for a scientific journal article. I always before used covariates for one way analysis. In your forum that is suggested. How  "correct" is it to use traditional statistics for this?
Oh, if you're using the analysis for a journal article, I would think you would want avoid traditional frequentest statistics, if that's what you've done in the past.  I'm not clear, though, on how you ran the analysis before with covariates.
I have used covariates to do ANOVA, one-way analysis of variance following the technical paper of Sawtooth by Orme and Howell. The problem is that I have two categorical variables gender and university and I need to check also if there are significant interaction effects of gender and university. Absolutely routine in traditional statistics but not here . I wrote your support and they advised me to put a question here and possibly call Kenneth.
I see the problem but I do not have a quick answer for you.  Contacting Kenneth is a good idea but I believe he is out of the office today.
I would be very grateful for ideas how to do this. Thank you for this discussion. I have referred to a technical paper by you and Orme many times in articles.