Have an idea?

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

T-ratio and standard error not computed


I am able to see utility values but "*****" is displayed in place of the standard error and t-ratio for all but one level in my conjoint output(I'm using CBC logit). Do you know why this could be happening?
asked Jul 10, 2018 by anonymous

1 Answer

0 votes
This can happen with aggregate logit if there is a problem with the utility estimation and the solution has failed to converge.  The problem can be caused by one or more factors:

1.  Not enough respondents, so some attribute levels were either picked 0% of the time, or 100% of the time.

2.  Deficient (poor) design, caused by too many prohibitions.

3.  Trying to estimate interaction effects among attributes involved in a prohibition.

If you continue to have questions or problems on this, you could send your project files to technical support (support@sawtoothsoftware.com) and ask them to look at it.
answered Jul 10, 2018 by Bryan Orme Platinum Sawtooth Software, Inc. (189,140 points)
Thanks for the help!
I see that the output file states that the model converged after 0.11 seconds.
I generated a pair of profiles using the same pool of attributes/levels so don't see the 0% issue either.
While I did not impose any prohibitions, can you think of anything else I can try? I want to make sure there is nothing wrong with my design.

I have written to the support ID, waiting to see if they can help.