Sid wrote: > Hi, > > I am investigating a way to determine whether the value of a > particular variable is either a 'positive' or 'negative' influence > wrt to a particular model and by how much. > > To illustrate, say we are modelling the type of man a women goes for, > and we are trying to predict the length of time they will date. We > suspect income is a relevant variable, and we are looking to provide > a measure of preference/aversion for a particular value. We have a > set of historical data of her previous relationships. > > So, when a new man turns up, we know his income. I would like to know > by how much this income is a positive or negative factor wrt the > length of time she will go out with him, based on the historical data > of her past relationships. > > Does this make sense? > > Probably not :) > > Thanks for any help.
Does not multiple regression give you what you want ? David Jones . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
