Have an idea?

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

Average importance in Summary by attribute level

I would like to know if there is a way to get this result, I will be grateful if any colleague can help me learn.

My project is CBC with 7 attributes and among them the Brand.
I estimated the utilities with CBC HB, so I got the general summary with Importance Average.

my question is how can I get this same summary for each level of the attribute tag.
can you help me?

Best Regards
asked Sep 13, 2017 by Walas (160 points)

1 Answer

0 votes
Attribute importances are only computed at the attribute level, not for levels within attributes.  Levels within attributes are reported as "utilities" (the utility report) and those utilities are typically zero-centered (the average of the utilities within each attribute is zero).
answered Sep 13, 2017 by Bryan Orme Platinum Sawtooth Software, Inc. (173,090 points)
Bryan, thanks for your quick answer,
Is not there a way to get this when importing utilities into smrt?
I thought if I could clean the utilities of some brands and run the market simulator, I'm afraid I should not do that.
I think I'm not understanding what you are looking for.  When you import utilities into SMRT, you can go to the market simulator and specify any product scenario you want.  When you run that market scenario, the software automatically gives you the utility report and the importances report.  The utility and importances reports only look at the utilities.  Anything you specify as competing products in the market simulator does not affect the utilities and importances report.

Next, you talk about "clean the utilities of some brands".  Are you trying to remove some brands from your experiment, even though you asked respondents to evaluate them?  That would be a strange thing to do.  Perhaps you can explain yourself further?
Let me see if I can improve my own doubt, I'll just talk about smrt.
If I compute the market simulator without any scenarios, it will give me the importance of all attributes without any scenario interference.

I would like to see the importance of all factors but considering only one brand.

I have 7 attributes, A1L4, A2L3, A3L3, A4L2, A5L6, A6L7 and A7L13
I can not visualize how to solve this by creating the scenarios by brand.

Let me know if you can get it
Yes, you are right that you don't have to enter a simulation scenario to get SMRT to report the importances and the utilities.  Just click "compute".

Standard utility estimation for conjoint analysis is to estimate the independent effect (main effects) of all attributes.  Thus, you are getting utility estimation for all attributes considering the average brand (brand held constant).

If you want to see what the importance score is for respondents who prefer a specific brand, then I would suggest creating a new segmentation variable based on HB utility scores.  First, create a new .CSV file that contains the respondent number and a new variable.  For each respondent, create a new variable that is a "1" if the respondent prefers that brand or a "2" otherwise.  Then, use the File + Merge Variables to merge that .CSV file into SMRT.  

Now use the new variable as a filter and ask SMRT to report the utilities and importance scores for you only for the respondents who prefer the brand you are interested in isolating.
Hi Bryan, you're fantastic.
Thanks for all the help and time with my question, thank you!