INTERVIEWS WITH 300 SUBJECTS PROVIDED DATA FOR ANALYSIS OF JUROR PROPENSITY TO CONVICT OR ACQUIT AND OF THE DIFFERENCES IN THOSE PROPENSITIES ACROSS TYPES OF JURORS AND CASES. THE RESULTS INDICATE THAT POTENTIAL JURORS ARE WILLING TO CONVICT WHEN THEY PERCEIVE DEFENDANT'S PROBABILITY OF GUILT TO BE ONLY .55. SUCH FINDINGS ARE CONSISTENT WITH OTHER STUDIES, REVEALING THAT JURORS TEND TO APPLY A 'MERE PREPONDERANCE OF THE EVIDENCE' STANDARD EVEN IN CRIMINAL CASES, RATHER THAN THE STRICTER STANDARD OF 'BEYOND A REASONABLE DOUBT.' SOMETIMES WHEN THE FACTUAL SITUATIONS WERE VARIED, THE RESPONDENTS SHOWED HIGHER THRESHOLD PROBABILITIES, BUT WITH THE ADDITION OF INFORMATION ON THE HARM DONE TO VICTIMS, THE JURORS AGAIN SHOWED A LOWER STANDARD FOR CONVICTION. PERSONAL EXPERIENCES OR FEARS OF VICTIMIZATION MAY CONTRIBUTE TO THE WILLINGNESS OF POTENTIAL JURORS TO CONVICT. WOULD-BE JURORS FROM DIFFERENT BACKGROUNDS AND SEXES SHOW ONLY SMALL DIFFERENCES IN THEIR AVERAGE THRESHOLD PROBABILITIES. INVESTIGATION OF THE RATIONALIZATION PHENOMENON REVEALS THAT CONVICTION-PRONE JURORS DECIDE TO CONVICT BEFORE HEARING THE EVIDENCE AND THEN PERCEIVE THE FACTS SELECTIVELY TO JUSTIFY THEIR DECISION. IMPROVED JURY INSTRUCTIONS ARE OFFERED TO ASSURE JURIES, USE OF THE 'BEYOND A REASONABLE DOUBT' STANDARD. TABULAR DATA AND REFERENCES ARE PROVIDED. (TWK)