Have an idea?

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

Raw vs. ZC Diffs utilities


I want to predict houldout tasks of my CBC within excel.
Which values do I have to sum up? The individiual utilities (raw) or the individiual utilities (ZC diffs)?
How can I include a linear model (price)? Until now I calculated as the example shows below:

Concept 1: Attr1(level 3 - zc diff utility) + Attr2(level 2 - zc diff utility) + Attr3(price - linear - zc diff utility)

Is this right?

Do I just compare to the None option (utiliy ZC diffs) and pick the maximum?

I tried to calculate the zc diffs by my own and compared them to the ssi web output "hb report". I couldn't get the same values. Is there a difference by using a linear variable?

I took the formula:

"For each respondent...

    1. Within each attribute, compute the mean utility.  Within each attribute, subtract the mean utility from each utility (this zero-centers the utilities within each attribute...which often doesn't have to be done since they are often already zero-centered in their raw form).

    2. Then, for each attribute compute the difference between best and worst utilities.  Sum those across attributes.

    3.  Take 100 x #attributes and divide it by the sum achieved in step 2.  This is a single multiplier that you use in step 4.

    4.  Multiply all utilities from step 1 by the multiplier.  Now, the average difference between best and worst utilities per attribute is 100 utility points."

Thanks for your help!
asked Feb 4, 2016 by anonymous

1 Answer

0 votes
If you are doing a first choice simulation (highest utility wins), then it doesn't matter if you are using raw or rescaled utilities.  However, if you are doing a vote splitting method such as the share of preference math, then you want to use raw utilities since they have the correct scale.

I would recommend that you run a nice, simple part-worth model first and make sure you feel comfortable in rescaling and running the predictions.  After that you could switch to linear and take the linear coefficient * the value of the level you wish to simulate and you should be able to replicate things correctly.

If you are on a subscription license you can download the new choice simulator which can export an template/starting point for an Excel simulator and save yourself some time.
answered Feb 4, 2016 by Brian McEwan Gold Sawtooth Software, Inc. (46,970 points)