I have a question regarding the calculation of the average importance. I understand that I take the range of the utility estimates, sum all up and then take the relative contribution to the range sum for every attribute.

How do I calculate this for an attribute with linear coding? E.g. price is linear coded and has a much higher utility level than all the others. Nevertheless, when I look at the average importance it's not that much higher.

When I do the logit analysis with part-worth coding for price, I can reproducethe average importance. But not for linear coding.

What am I doing wrong?

Thanks a lot!

Now, I can reproduce it.