Have an idea?

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

Clarification on formula

Thanks for providing us with this information Brian.  I had never thought to use SOP in our simulators before!  To clarify:

Pa = the summation of utility values describing Product A
Pb = the summation of utility values describing Product B
Pn = the summation of utility values describing Product N

Share of Preference is:

=EXP(Pa) / (EXP(Pa) + EXP(Pb) + ... +EXP(Pn) )
=EXP(Pb) / (EXP(Pa) + EXP(Pb) + ... +EXP(Pn) )
=EXP(Pn) / (EXP(Pa) + EXP(Pb) + ... +EXP(Pn) )

Generally, what is the SOP for including a "None of these" option in this calculation?  My assumption would be that we would take an EXP() function of a single None value as opposed to the EXP() of the summation of values for a given product.  Is this correct?
asked Apr 10, 2015 by Nick Rowland

1 Answer

0 votes

Very clear, indeed.  

You are correct.  You would add the exp(none) to the formula - all by itself it is the total value of the none alternative.  

When including it in a simulation it behaves just like the other products:  it goes into every product's denominator and also in the numerator for the none alternative.
answered Apr 11, 2015 by Keith Chrzan Platinum Sawtooth Software, Inc. (92,075 points)
Keith, one minor clarification.  Is it recommended that we use ZC Diffs or Raw utility values to do this transformation?
Raw utilities!  ZCD are NOT scaled to work with the MNL choice rule.