STATISTICAL ESTIMATION OF THE BEHAVIORAL IMPACT OF PUBLIC POLICIES IS OFTEN AN IMPORTANT PART OF POLICY ANALYSIS. IN ACTUALITY, HOWEVER, POLICY ANALYSTS OFTEN FIND IT DIFFICULT TO USE THEORY AND STATISTICAL TECHNIQUES PROPERLY, AND POLICYMAKERS FREQUENTLY CANNOT ASSESS STUDIES ESTIMATING THE BEHAVIORAL EFFECT OF VARIOUS POLICIES. IN EHRLICH'S STUDY, ESTIMATES OF COEFFICIENTS WERE GIVEN IN AN EQUATION WHERE THE DEPENDANT VARIABLE WAS A MEASURE OF THE MURDER RATE. INCLUDED AMONG INDEPENDENT VARIABLES WERE THE PROBABILITY OF APPREHENSION, THE CONDITIONAL PROBABILITY OF CONVICTION, GIVEN APPREHENSION, AND THE CONDITIONAL PROBABILITY OF EXECUTION GIVEN CONVICTION. THE FOCUS OF THE STUDY CRITIQUE IS ON CRIMINAL BEHAVIOR AND THE CRIMINAL JUSTICE SYSTEM, MURDER BEHAVIOR AND CRIMINAL JUSTICE SYSTEM BEHAVIOR, AND SPECIFIC MODELS OF CRIMINAL AND CRIMINAL JUSTICE SYSTEM INTERACTION IN THE PROBLEM OF IDENTIFICATION. DEALING WITH THE IDENTIFICATION PROBLEM REQUIRES SPECIFYING APPROPRIATELY BOTH THE BEHAVIORAL THEORY UNDERLYING POLICY ANALYSIS AND A MODEL BASED ON THAT THEORY TO DETERMINE WHETHER IDENTIFICATION IS POSSIBLE. AN APPROPRIATE SAMPLE OF BEHAVIOR SHOULD BE EMPLOYED IF IDENTIFICATION IS THEORETICALLY POSSIBLE. IN THE PROPER USE OF THEORY AND MODELING, IT IS NECESSARY TO NOTE ALL EQUATIONS THAT CONTRIBUTE TO DETERMINING ENDOGENOUS VARIABLES IN A PARTICULAR EQUATION TO BE ESTIMATED. EQUATIONS FOR SOCIETY'S RESPONSE TO MURDER BEHAVIOR ARE SPECIFIED, INCLUDING EQUATIONS FOR PRODUCTION BEHAVIOR WITHIN THE CRIMINAL JUSTICE SYSTEM, THAT DETERMINE ENDOGENOUS VARIABLES IN EHRLICH'S MURDER SUPPLY FUNCTION. SUPPORTING DATA ARE PROVIDED, AND APPENDIXES CONTAIN A MODEL OF CRIME AND CRIME CONTROL AND A MODEL OF THE CRIMINAL JUSTICE SYSTEM. (DEP)