Have an idea?

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

Reformat CVA data into CBC to use CBC model explorer

Keith, thank you so much for the detailed answers.

It really helped me lots.

To provide more information, I am using CVA for my dissertation to reveal the preferences, not much focus on predictability.

This being said, although it may provide very little diagnostic value, I would need to pick one of the ways: adding at least three holdout fixed sets or reformat CVA to CBC data.

I looked for postings on the forum and technical papers in reformating CVA data into CBC for potential use in CBC model explorer, but couldn't find one.

My 15 CVA scenarios are 9 point RATING scale with paired concepts.

Would you please lead me a way to reformat the data?

If not, would adding 3 holdout fixed scenarios (assuming 3 is the enough number) at the beginning of the survey before the actual CVA scenarios be the only option?

Once again, thank you so much for any input or feedback.
related to an answer for: Validity without holdout tasks for CVA
asked Jun 13 by anonymous

1 Answer

0 votes
If you're only looking to show that your model fits the data well and that it predicts holdouts well, then I think 3 will be fine for your needs.  I would add more if you needed to use the holdouts to test alternative model formulations.

I would not show them all three up front, in case there are learning effects and the utilities change a little during the course of the interview.  I would intersperse them throughout your interview, one toward the beginning, one in the middle and one toward the end.
answered Jun 13 by Keith Chrzan Platinum Sawtooth Software, Inc. (95,775 points)