Have an idea?

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

Substantial increase in the price elasticity with introduction of SKU x Price interaction is significant or not?

In all the previous study, we were including Brand x Price interaction for generating utilities. Here we have assumed that price is linear. Recently in some studies the sensitivity of the model is coming out to be very low, say, about 0.2-0.3. On introducing SKU x Price interaction in utility generation the model becomes highly sensitive, the elasticity of level had increased to 0.9-1.0.

We have also examined the impact of introducing SKU x Price interaction in the other studies as well. Elasticity numbers had gone up significantly for the studies with higher number of SKU levels. Now we need to understand whether the rise in elasticity is significant not.

Thanks in advance for the help.
asked Aug 8, 2016 by sanmit1988 (260 points)

1 Answer

0 votes
First thing is that you should pay attention to the simulated shares of preference for the SKUs as you change the price (holding all prices constant for competitive brands).  That is because different HB-MNL models with different numbers of parameters estimated can have large shifts in scale for the magnitude of the parameters which may be more due to fit rather than to actual changes in price sensitivity.

The test that occurs to me is the test of whether the interaction terms are statistically significant.  Refer to the alpha draws file (history of population mean estimates).  Count for what percent of draws (after convergence) the interaction term is either positive or negative.  If 95% or more of the draws share the same sign, then you are that % confident that the interaction term is significant.
answered Aug 8, 2016 by Bryan Orme Platinum Sawtooth Software, Inc. (132,290 points)
...