Have an idea?

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

CBC price attribute - Continuous Price


We are looking at a CBC design with 10 SKUs and 5 price levels which we will show as conditional pricing (very high / high / medium / low / very low).

What we would like to do is offer the client the possibility to simulate  any price he desires. For instance, if sku A had the price levels set at $5 / $6 / $7 / $8 / $9 we would like to be able to run simulation with prices such as $6.49 or $7.35.

I read Bryan Orme's paper "Three Ways to Treat Overall Price in Conjoint Analysis" and it seemed to indicate that it is possible through the continuous price approach, however it's not fully clear how to use it.

So my questions are:
1. Is this possible?
2. If yes, how? Should we edit the .cho file and if so in which way?

Thank you for your help inadvance.
asked May 13, 2013 by anonymous

1 Answer

0 votes
Standard CBC designs where you've shown respondents discrete price points (e.g. $5, $6, $7, $8, and $9) for SKU A will still allow you to simulate any possible price between $5 and $9, even if you don't use "summed price".  Our simulators support interpolation between points shown to respondents in the study.  Straight-line interpolation is applied, with straightline between the two nearest estimated price points.

So, if this is the main goal, you can save yourself a world of effort by just doing CBC in the standard way (not continuous price) but using the interpolation capability within our simulators.

answered May 13, 2013 by Bryan Orme Platinum Sawtooth Software, Inc. (152,955 points)
So I should use "Assign level values" in SMRT, correct?

If so, how would I deal with conditional pricing - sku A has prices set at $5 /6 / 7 / 8 / 9, but sku B has prices set at $6 / 8 / 10 / 12 / 14, so I can't use the same values for all level, can I?
When you use conditional price, usually you make the difference between the levels equal for all ranges. Therefore, the utilities are interpreted as "a difference of $2 from base price" or "a difference of 15% from base price" instead of "a price of $6".

If this assumption holds true for your study, all you'd need to do is load the conditional table into the simulator (as I've never used the sawtooth simulator, I'm afraid I can't promise you this feature exists but I'm pretty sure it does). This table will then be used to determine which price range should be used for interpolation for a certain product. That way, SKU A will get $5/6/7/8/9 assigned to the price interpolation and SKU B will get $6/8/10/12/14 assigned to the price interpolation.
I don't see any way to load the price matrix into SMRT; obviously I could be wrong, but if I am correct this means that I need to set up a simulator of my own.
No, but it does mean that you'll have to do the interpolation yourself. SMRT does support the use of linear interpolation but from what I remember (sorry, I've been using custom Excel simulators so I'll have to rely on memory here) you can enter level values like 1.45.
Imagine we want to simulate a price of $5.50 from the price range of SKU A, the value you'd input into SMRT would be 1.5 cos $5.50 should take 50% from level 1 ($5) and 50% from level 2 ($6). This trick works for sure. Just remember to check your scenario before interpreting the SMRT results to make sure you associated the interpolated values with the right product profile!

You could check the manual to see if you could also use prices. I think you'd need to re-create the SSI-Web study in SMRT (CBC used to be SMRT based!) and use the associated data files to run the simulation. But please remember that this last paragraph is from memory so I might be completely off here!