IN ESTIMATING THE SIGNIFICANCE OF THE DETERRENT EFFECT, RESEARCHERS COMMONLY ANALYZE NATURAL VARIATIONS IN CRIME RATES AND SANCTION LEVELS ACROSS DIFFERENT UNITS OF OBSERVATION. STUDIES CONSISTENTLY REPORT A NEGATIVE ASSOCIATION BETWEEN CRIME RATES AND RISK OF PUNISHMENT. THE VALIDITY OF RESEARCH WHICH PROVES THAT SANCTIONS DETER CRIME IS COMPROMISED BY THREE FACTORS: ERRORS IN REPORTING CRIME BY CITIZENS OR POLICE, THE FACT THAT IMPRISONED OFFENDERS WILL REDUCE CRIME BY INCAPACITATION AS WELL AS DETERRENCE, AND SIMULTANEOUS RELATIONSHIPS BETWEEN CRIME AND SANCTIONS, SUCH AS JURISDICTIONS WHICH IMPOSE LOWER PENALTIES BECAUSE THEIR SYSTEM IS OVERBURDENED BY A HIGH CRIME RATE. QUASI-EXPERIMENTAL STUDIES ARE FREQUENTLY DONE WHEN LAWS CHANGE OR NEW ENFORCEMENT TECHNIQUES ARE APPLIED, BUT RESULTS ARE USUALLY LIMITED TO A SPECIFIC LOCALE, TYPE OF CRIME, OR TACTIC AND CANNOT BE APPLIED TO GENERAL THEORIES ON DETERRENCE. EFFORTS TO ASSESS THE EFFECT OF CAPITAL PUNISHMENT ON HOMICIDE USUALLY FAIL TO ACCOUNT FOR DEMOGRAPHIC, SOCIOECONOMIC, AND CULTURAL FACTORS, AND NO CONCLUSIVE EVIDENCE HAS BEEN PRODUCED TO PROVE CAPITAL PUNISHMENT IS A DETERRENT. SINCE HIGH CRIME AREAS TYPICALLY HAVE A LOW RATE-OF-TIME-SERVED PER CRIME, THEY WOULD REQUIRE THE LARGEST INCREASE IN PRISON POPULATIONS. RECOMMENDATIONS FOR RESEARCH INCLUDE EXAMINATION OF THE EFFECTS OF WORKLOADS ON CASE DISPOSITIONS, USE OF TIME-SERIES TECHNIQUES, DECRIMINALIZATION OF CERTAIN OFFENSES, AND PATTERNS OF INDIVIDUAL CRIMINAL ACTIVITY. COMMON DATA BASES ON CRIME, JUDICIAL PROCESSES, AND CRIMINAL HISTORIES ARE NEEDED. (MJM)