A factor of +1 or –1 used to weight predictor variables, where the signs are based on a priori information. These are often equivalent to equal weights. One may need to decide how to scale the variables. Typically, each variable’s observations are transformed to standard normal deviates from the variable’s mean.