Have an idea?

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

Is it possible to randomize component prices rather than total price in ACBC?

We are thinking about using ACBC for a pricing piece for which it seems well suited given the number of attributes and the existence of non-compensatory decision rules. However, the client wants to test pricing variation at the attribute rather than the total level because the aim of the project is to estimate the impact of changing the prices of individual attributes on share of preference.

I am on the fence about whether this necessitates varying prices of individual components, after all traditional conjoint can be used for this purpose quite readily, but in the relevant market consumers are used to seeing component prices and as such it would be preferable to be able to display them individually (as well as a total) for each offer. Since the total price needs to match the combined price of the components, varying component prices and totaling them up might be the best solution.

Aside from the need to display the prices for individual attributes, the impact on utility might be different depending on how the random change of the overall price is distributed across attribute levels. This would be a psychological effect because rationally it obviously doesn't matter which element of an offer makes it more or less expensive but I think it's likely to exist (given that components are explicitly priced individually).

Does anybody have experience with this problem?

I understand continuous price can be introduced to CBC by manipulating the .CHO file and I could see this working in a similar way but I am not sure whether this is also possible for ACBC and what effect this would have on the estimation. Is it be feasible to estimate utilities for different prices on a per-attribute basis?

Many thanks in advance
asked Apr 11, 2012 by anonymous
retagged Sep 13, 2012 by Walter Williams

1 Answer

0 votes
If buyers in the real world see the components of the product priced separately, and can select among those components to construct their preferred product (a Build-Your-Own approach), then probably the best thing to do is to create a conjoint analysis survey that mimics this buying process: MBC.  Within MBC, you can vary the prices of the components, as well as the base price of the entire product.

But, if buyers just see the total product price, and face a choose/no choose decision, rather than a configure the product (feature-by-feature) to match your needs decision, then ACBC or CBC is the right approach.

Beware that MBC requires larger sample sizes than ACBC to stabilize utility estimates.
answered Apr 12, 2012 by Bryan Orme Platinum Sawtooth Software, Inc. (198,515 points)