Have an idea?

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


Hi everyone,

I have conducted an ACBC with summed price. I have used HB to estimate importance of attributes and PWU of attribute levels.

Now, I also want to look at WTP. If I understood the Sawtooth resources right, I have two options:

1.    Computing Dollar Equivalents from Utilities

This is not recommended in your resources. However, I can see that it is applied in some peer-reviewed papers.

I have one question:

I have used a summed price approach for the ACBC. For the HB estimation, I used piecewise price with a total of six levels (lowest + highest + 4 in between). Is it acceptable to still use the formula

(pm,i - pn,i )/ (v(pm,i) - v(pn,i))   ?

(where pm,i represents the lowest price point in a linear price function and pn,i displays the highest price point for each respondent, v(pm,i) represents the utility of pm,i and v(pn,i) displays the utility of pn,i.)

2.    Using market simulations
a.    SOS approach

When I use the SOS approach, I get the error:

“In scenario “My WTP Scenario”: Could not find a competitive set that satisfies the prohibitions.

I have set 2 prohibitions in my ACBC. How can I solve this problem?

b.    Fixed WTP product

I define my product according to https://sawtoothsoftware.com/help/lighthouse-studio/manual/index.html?understanding-willingness-to-p.html .

My question is: what about the price? What price level do I put? For the competition products, I set realistic prices as they don’t change (however, I am restricted to my 6 price levels). For my own product, do I put the lowest price level (as the default levels are the cheapest ones)?

Also: with my current settings, my results show no problems with extrapolation or WTP not converged. However, my surcharges for enhanced attribute levels seem quite high. Calculating the surcharges via (pm,i - pn,i )/ (v(pm,i) - v(pn,i)) render much lower (and realistic) values. What could be my mistake? (I configured 5 competition products and a none option).

Thank you in advance for any guidance!

asked Feb 10 by MichaelFu (220 points)

1 Answer

0 votes
First, an approximation to the linear approach is to take the endpoints from the piecewise price function.  But, to stay more true to the linear approach it would be better to rerun the ACBC HB estimation with just the two endpoints specified in the price grid and to specify using the Linear estimation approach rather than piecewise.

Regarding your second question, I don't know why with only two prohibitions the SOS approach breaks out without ability to find non-prohibited solutions.  Perhaps that's a good issue to package up your files and send to our technical support team.

On the third issue, it's probably unrealistic to ask our software to use the lowest price point as the base starting point to estimate WTP.  Probably something more toward the realistic price of the product given an average configuration of enhancements would be more appropriate.  Don't know if that's why you're seeing large WTP values.
answered Feb 10 by Bryan Orme Platinum Sawtooth Software, Inc. (199,115 points)
Dear Brian,

thank you for your answer.

I have one more question regarding Computing Dollar Equivalents from Utilities:

When I use the average part-worth utilities of my lowest and highest price point shown in the summary section after HB estimation for the total sample and insert them to the formula (pm,i – pn,i) / v(pm,i) − v(pn,i), I get a different Dollar Equivalent than when I use individual respondent part-worth utilities of the lowest and highest price point and then look at the average of the resulting individual Dollar Equivalents.

Where is my mistake? When I average individual part-worth utilities for the lowest and highest price point, I get the same values as indicated in the summary section. Also, the lowest and highest price point are always the same, so both results should be equal?!

Thank you in advance for any guidance!