Have an idea?

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

Statistical test for revenue simulation and WTP

Dear all,

I conducted cluster analysis and I want to proof that offering more customized cluster-specific products results in higher revenues than only offering one generic product. I was wondering how I could conduct a t-test on the different results since I lack individual WTPs and standard errors or confidence intervals. I use ACBC and for revenue calculation RFC and exhaustive search.
Thank you in advance!

asked Feb 11 by Lisa

1 Answer

0 votes
Best answer
You don't need to do a WTP analysis to show what you'd like to demonstrate.  You just need to show that shares x prices (indicating revenues) for the products sold by the firm are greater with a line extension focused on meeting needs of clusters rather than the alternative.

In Sawtooth Software's market simulator, there is a setting you can click to report individual-level results (shares of preference).  So, for each respondent you can calculate their expected revenue captured by the firm by multiplying the shares of preference by the price for each of the firm's products and sum them.

You run a market simulation that is your base case (prior to the clustering-line extension offerings) and calculate the revenue to the firm for each respondent.  Then, you do the same with the new simulation scenario involving the line extension.

This then becomes a matched-pair T-test that you can calculate in Excel.  You create a new variable which is the difference between the revenue to the firm from each respondent for the new simulation scenario vs. the base case.  Take the mean of that difference variable.  Calculate its standard error by taking the standard deviation of that difference variable across respondents, then divide that by the square root of your sample size.

The t-value is given by taking the mean difference in revenue divided by its standard error.  A t-value of 1.96 or greater indicates 95% or better confidence.

When creating your base case scenario, it would be important to have a rich variety of the expected competitors in the simulation scenario, plus the None.
answered Feb 16 by Bryan Orme Platinum Sawtooth Software, Inc. (199,115 points)
Hi there,

one more question: I use a market size of 90000, which results in great revenue. Should I change it for this calculation to 1 or to the number of actual respondents (1165)?