Have an idea?

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

Max Diff Sawtooth vs HB/CBC


I have tried running the analysis for my Max Diff in two ways. Firstly running it inside Sawtooth and secondly making a .cho file and running it in the HB/CBC software.

When I get the results, the results from both methods are very similar but not exactly the same and as far as I can tell I am running them on the same settings. Is there a reason why the two results are slightly different? If so, what is the best approach to use?

asked Jun 5, 2019 by DaveT (245 points)

1 Answer

+1 vote
The main difference between them is that they use different random number generators.  Since HB is a highly iterative algorithm and each iteration depends on random draws from a multivariate normal distribution, they will be different but should converge to basically the same answer.  In this case either will work just fine.
answered Jun 5, 2019 by Walter Williams Gold Sawtooth Software, Inc. (18,880 points)