Have an idea?

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

Input File CCEA

Hello Sawtooth-Community,

we have a question regarding Sawtooth's CCEA module. We want to run a consensus clustering on individual-level HB part-worths obtained from a CBC analysis (zero-centered diffs have been applied during the HB estimation).
Should the input file contain the respondents in rows and the attribute levels in columns? Do we also need the dual none-option as a separate column?

Furthermore, we do not understand how the so-called grand means are calculated. Is it correct, e.g., that we have to calculate the mean of the means over all groups for attribute level a?

Thank you very much for your answer.

Kind regards,

asked Mar 13, 2014 by Steve

1 Answer

0 votes
Respondents are rows.  Utilities are columns.  

If you want to include the None utility as a driver of clusters, then use the None utility as well as a column.

Grand mean is the average across all respondents.  It will only be equal to the mean of the means over all groups if the group sizes are exactly equal.
answered Mar 13, 2014 by Bryan Orme Platinum Sawtooth Software, Inc. (174,740 points)