# Calculating importances for continuous attributes?

Suppose I have both part-worth and continuous attributes (estimated as user-specified in CBC/HB) in my model.  How would I calculate the importance of all my attributes, including the continuous ones?  For part-worth attributes, we use the range of its levels, but there's only 1 parameter for a continuous attribute so I don't know the best way to come up with a range to include in the importance calculation.  Thanks!
retagged Sep 13, 2012

The way we do it is to take the difference in utility for that continuous attribute that would be seen by computing the utility at the lowest level and highest level shown to respondents.  In other words, it's the difference between the highest and lowest levels of that attribute, multiplied by the coefficient for that attribute.
answered Feb 10, 2012 by Platinum (198,315 points)
We can take a range for the attribute with partworth utilities but in case of continuous attribute where there is only one parameter estimated, I guess we need to take the value as is? Am I right Bryan?
What we do with continuous attributes (if we estimate linear terms rather than partworth) is to compute the utility difference between the best and worst levels.  So, that's the difference in the X value between the best and worst levels of that continuous attributes times the slope coefficient.