Hello.
I am planning to conduct an attitudinal segmentation using Sawtooth's CCEA software. Specifically, cluster inputs into CCEA will be a 15-30 item attitude item battery answered on a 5 point scale by appropriately-screened survey panel respondents. Once I've obtained CCEA segments, I will use stepwise discriminant analysis to develop a Typing Tool Algorithm (containing a subset of only the most predictive attitude items used to produce the CCEA segments) to identify the segments in future quantitative/qualitative research with reasonably high (70%+) accuracy.
I will need to then score/tag an internal client database with the CCEA segments BUT customers in this database will not be included in the sample used to produce the CCEA segments - so no database records will have responses for Typing Tool Algorithm attitude items.
In order to score the database with strong accuracy (70%+), I was planning to send a brief survey containing the Typing Tool attitude items to all database records. Database records that complete the survey can then be tagged/scored with CCEA segments - which will likely result in only a small fraction of the database having a CCEA segment (given that only a small fraction will likely respond to the typing tool survey).
Ideally my client would like ALL database records tagged/scored with
the CCEA segments with high accuracy (they are requesting 70%+ accuracy). My client is not interested in using a sample of database respondents in the CCEA segmentation (which would at least provide attitude item data for some database respondents). My client is also not interested in trying to build a Typing Tool algorithm with items outside of the attitude items used as CCEA basis variables - such as demos or other metrics available for their database records - because Typing Tool accuracy is more likely to be below 70%+ if not including CCEA basis variables.
Does Sawtooth have any insight on recommended approach to achieve what's being asked for here (that is, being able to score/tag ALL, or almost all, database records with the CCEA segments at a high 70%+ accuracy)?
Thank you.