Have an idea?

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

CBC Paper-and-Pencil: How to analyze data in Lighthouse without design file

Hi Everyone,

I have a complicated thing to do in Lighthouse and I hope somebody is able to help me.

I have received data from 590 respondents of a Paper-and-Pencil CBC in the field of health economics. Here comes the problem: Respondents answered questions according a balanced overlap design which was constructed using SSI Web 8.x. Unfortunately, (1) the design file is not available anymore to me. And (2) our field partner forgot to insert the design number to the individualized questionnaire versions.
Therefore, I have one question: Is it possible to analyze the CBC data in Lighthouse (HB et.) when I solely can provide the choices of respondents and the seen choice tasks? If yes, what is the easiest way to conduct such an analysis?
I understood that it is very easy if I would have the design in Lighthouse. Then I could use the Data Template File functionalities. However, this seems not possible in the present case.

Many thanks in advance!
asked Feb 22, 2016 by Marcel (290 points)
retagged Feb 22, 2016 by Walter Williams

1 Answer

+1 vote
If you know the respondent answers and you have recorded which attribute levels each respondent saw for each attribute in each concept of each task, then I can immediately think of two options:

1) Using your own data processing skills, create a .CSV file of the format used by our standalone CBC/HB or CBC/Latent Class systems (if you have a license for CBC within Lighthouse, then you have free access to these standalone tools and may download them from our website).  The format for the .CSV file is provided in section 4.2 of the CBC/HB Manual: http://www.sawtoothsoftware.com/download/techpap/CBCHB_Manual.pdf

Then, estimate your model using CBC/HB or standalone Latent Class software.  Then, read the .HBU file that is created into the standalone market simulator (same simulator as used within Lighthouse, but this time running outside of Lighthouse).

2) It's quite probable that the person who created the design for your study just used default settings within the software so you might be able to recreate it by asking Lighthouse to generate it again.  Then, check your .CSV file that you created in Step 2 above to see that indeed each respondent is getting the same designs as you've generated again...then you will also have the ability to match each respondent with whatever design they received.  This will give you the ability to read the respondent answers and their design numbers through the paper-and-pencil import.  Then, you are back to working with your data within Lighthouse.
answered Feb 22, 2016 by Bryan Orme Platinum Sawtooth Software, Inc. (195,590 points)
Dear Bryan,

Many thanks for your suggested solutions!

I also found out another way to analyze the data within the new and wonderful Lighthouse Software. I like to share it here in the hope, other researchers will benefit from it.

The problem was, that I had collected 590 interviews in a Paper-and-Pencil CBC. Unfortunately, I had no choice design when it came to the analysis stage. Here the workaround.
(1)    I firstly create a CBC design in Lighthouse. Then I exported the design CSV file just to get an impression of how the format looks.
(2)    In the design Tab of a new CBC study I then reconfigured the CBC study and set the number of questionnaire version to be 590.
(3)    Now, I extracted which respondent saw which profiles from the conducted Paper-and-Pencil CBC.
(4)    In a next step, I placed the realized design in the exported CSV file
(5)    Then, I import the design to the reconfigured CBC in Lighthouse
(6)    At the final stage I provided the realized choices by using the standard template created by Lighthouse.

Best Regards

Very nice job!  Glad you figured out another way to get where you wanted to be.

Note for others: Lighthouse is limited to 999 versions, so this trick can only work for up to 999 unique versions (also known as blocks).