Have an idea?

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

Video Options Setup in Lighthouse Studio Software

Thank you very much Bryan.

You have understood the nature of my question, though I am still uncertain of how to set this up in the software. For them to display correctly (just the video), I believe I would have to set this up as one factor, with 8 levels - with each level corresponding to a video. How then do I analyze this as a three-factor solution, each with 2 levels (even though this was set up as a one-factor, 8 level design in the software)? Is there a way to specify this post-study, during the analysis? Or do I need to set this up differently? I am expecting main effects and some first-order interactions to be significant, so I would like a model to take this both into account.

Additionally, I will have too many combinations to create videos for all unique possibilities. For instance, let's assume I have 5^3 possible combinations (125), but would like to limit the number of unique combinations shown to users to (25), due to constraints in making unique videos. Is there a way to specify a design matrix with this constraint using your software? Or, would I need to reduce the number of factors/levels by combining some of them? It seems like some factors/levels would be correlated regardless, and these options might be effectively similar.
related to an answer for: Videos Containing Level Information
asked Nov 13 by anonymous

1 Answer

0 votes
Best answer
Our CBC software can do what's called "conditional display", where you can decide to show a graphic or a video depending on a combination of separate attributes.

Thus, you could set up 3 attributes with 2 levels each, but you create a conditional display association that involves all three attributes.  The conditional display table will open up 8 slots for you to insert what graphic or video to display based on the co-occurrence of the 3 attributes' levels.  

Because you will have set this up as 3 separate attributes, you would be free to model the utilities as main effects, or main effects plus selected first-order interaction effects (interactions between attributes taken 2 at a time).  Note that our software doesn't allow you to model interactions between three attributes taken at a time.

Regarding prohibited combinations in a 5^3 design, this gets tricky.  It usually involves creating just one block (version) of the CBC questionnaire and using some type of experimental design software to see if 25 unique cards could allow you to estimate with efficiency main effects or some selected interaction effects.  This usually takes an expert in design of experiments who would leverage a piece of software such as our traditional conjoint CVA tool or other available experimental design software options on the marketplace.  This effort usually involves collapsing factors that have a great deal of prohibitions between them, while leaving other factors independent that don't involve prohibitions.
answered Nov 13 by Bryan Orme Platinum Sawtooth Software, Inc. (179,015 points)