I'm wondering if it can be done by assuming that this variable is a latent variable and setting the regression weight (via parameter constraint menu option) to specific value.
Anyway it would be very useful to be able to insert unmeasured concepts into my model based on literature/prior knowledge of correlations. . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
