THE METHODOLOGICAL PROBLEMS DISCUSSED ARE AS FOLLOWS: (1) LACK OF CONTROL GROUPS, (2) OVERRELIANCE ON IDENTIFIED CASES OF CHILD ABUSE WITH LITTLE REGARD TO UNIDENTIFIED CASES, (3) OVERRELIANCE ON A MEDICAL MODEL FOCUSING ON ONE OR TWO 'CAUSES', (4) SEARCHING FOR THE PERFECT ASSOCIATION RATHER THAN CONSIDERING A COMBINATION OF FACTORS AND LESS-THAN-PERFECT ASSOCIATIONS, (5) MISTAKING CONCURRENT TRENDS FOR CAUSE-AND-EFFECT RELATIONSHIPS, (6) IDENTIFYING PSYCHOLOGICAL CONDITIONS WHICH ARE PRESENT AFTER THE ABUSE AS A CAUSE OF THE ABUSE (SUCH AS FINDING THAT ABUSERS ARE PARANOID AND DEPRESSED, CONDITIONS WHICH COULD WELL BE RESULTS OF THE INCIDENT), (7) ATTEMPTING TO DEVELOP CAUSAL RELATIONSHIPS FROM AFTER-THE-FACT EXAMINATIONS OF DATA; AND (8) INAPPROPRIATE METHODOLOGY IN THE PRESENTATION OF TABULAR DATA. FOLLOWING THIS DISCUSSION, A NUMBER OF CHILD ABUSE STUDIES ARE BRIEFLY SURVEYED AND THE IDENTIFIED ASSOCIATIONS ARE SUMMARIZED AS FIVE PRINCIPLES: (1) VIOLENCE TENDS TO BE A RESPONSE TO PARTICULAR STRUCTURAL AND SITUATIONAL STIMULI; (2) STRESS IN AMERICAN SOCIETY IS DIFFERENTIALLY DISTRIBUTED IN THE SOCIAL STRUCTURE; (3) EXPOSURE TO AND EXPERIENCE WITH VIOLENCE AS A CHILD TENDS TO TEACH THE CHILD THAT VIOLENCE IS AN ACCEPTABLE RESPONSE TO STRUCTURAL AND SITUATIONAL STRESS; (4) INDIVIDUALS IN DIFFERENT SOCIAL POSITIONS ARE DIFFERENTIALLY EXPOSED TO BOTH LEARNING SITUATIONS WHICH EMPHASIZE VIOLENCE AND STIMULI WHICH MAY PROVOKE VIOLENCE; AND (5) INDIVIDUALS USE VIOLENCE IN THE FAMILY AS AS RESULT OF BOTH LEARNING EXPERIENCES AND STRUCTURAL CAUSAL FACTORS. RESEARCH OF BETTER QUALITY TO STUDY SPECIFIC TYPES OF ABUSE IS URGED. REFERENCES ARE APPENDED. (GLR)