Have an idea?

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

How to capture regional variations in base price for ACBC with summed pricing

What is the best way to capture regional variations in a base price in an ACBC that uses summed pricing (base + component prices)?  

Would you recommend something similar to the multiplier approach described in the 2008 home purchasing study*, where prices shown to the respondent were multiplied by regional multipliers/deflators?  If so, how do you go about applying such a multiplier when creating the summed pricing attribute?  I understand how constructed lists can be used to populate levels for non-price attributes (such as brand) based on earlier questions in the survey, but I'm not clear on how to do I am describing with price.

*Orme, Bryan and Rich Johnson (2008), “Testing Adaptive CBC: Shorter Questionnaires and BYO vs.‘Most Likelies”
asked Nov 20, 2019 by dpatek (265 points)
I think I may have figured it out, but I want to double-check to be sure.   For traditional CBC studies, I sometimes use PERL script to pipe in the appropriate foreign currency for price based on a currency question earlier in the survey.  My guess is that I would do something similar here, with the difference that the actual new converted price would be used in the analysis and not just a discrete price attribute level number 1, 2, 3 or 4.

1 Answer

+1 vote
You're correct that with summed pricing, there is no list of levels, so constructed list building won't work.  You also can't just use Perl to fill in a price, because again you don't have level 1, level 2, etc. in the design.

You can put in math with our scripting tags or Perl into the pricing boxes where you set up all the price adjustments.  The main downside to this is that the actual value of the price is stored in the data.  As an example, I'm programming a survey in English and defaulting to dollars and put feature X at $100.  When I make a copy of my survey in Turkish and multiply all the prices by about 5.5 to turn it into Lira (550 lira), there isn't any way for me to undo the 4x multiplication in the raw data.  The software will think that my English respondent saw feature X at $100 and my Turkish respondents saw it at $550.

I haven't personally done a project like this, but if you keep everyone together, you probably want to include currency as a segmentation variable so you can include it as a covariate and segment your market simulations and keep your dollars and lira separate. I've heard anecdotally from a few people that seemed to work out alright.
answered Nov 20, 2019 by Brian McEwan Gold Sawtooth Software, Inc. (45,995 points)
Thanks, Brian.  The regional variations in this particular study are all within the US, so I won't need to do currency conversions. I was thinking perhaps I could use something like Perl in the "base price" box that will pipe in a higher price for high-cost regions and a lower cost for low-cost regions.  
I essentially want to replicate what you did for the 2008 home study, where you applied multipliers and deflators to arrive at a base home price (ignoring the price adjustment for size of house done in that study).
Yes, the trick is to recognize that the price fields in ACBC's Summed Pricing tab can accept fixed values as well as script.  Since your project is all in the US with dollars, then you don't have to worry about currency differences.  The script involved in setting the base price and the component prices different per region don't need to involve perl.  Perhaps Sawtooth Script (enclosed in [% %] brackets) would work.  Or, even Javascript.   Obviously, make sure to test thoroughly and check the raw data file to see if the prices shown to respondents are being stored, etc.
Thanks, Bryan!  And sorry for misspelling your name the first time.
You're actually dealing with two Brian's here who have posted responses to your query, so you were right in your spelling!  (Brian McEwan and Bryan Orme).  I was just late to the party.