THE SEARCH FOR A SCREENING DEVICE BEGAN WITH THE ADMINISTRATION OF TWO PRELIMINARY QUESTIONNAIRES TO 30 KNOWN CHILD ABUSERS AND 30 NONABUSERS. QUALITATIVE AND QUANTITATIVE RESULTS THEN WERE USED TO DESIGN A SELF-ADMINISTERED INSTRUMENT WHICH WAS TESTED ON 30 ABUSERS AND 30 NONABUSERS MATCHED FOR AGE, EDUCATION, SOCIOECONOMIC STATUS, AND FAMILY STRUCTURE. THE INSTRUMENT CONSISTED OF 82 SHORT STATEMENTS WITH WHICH RESPONDENTS WERE ASKED TO INDICATE THEIR DEGREE OF AGREEMENT. THIRTY-TWO OF THE ITEMS SHOWED NO POTENTIAL FOR DISCRIMINATING BETWEEN ABUSERS AND NONABUSERS AND WERE EXCLUDED FROM FURTHER ANALYSIS. THE REMAINING 50 ITEMS WERE SUBJECTED TO CLUSTER ANALYSIS, A STATISTICAL PROCEDURE FOR GROUPING QUESTIONS WITH SIMILAR PATTERNS OF CORRELATIONS. THE FOLLOWING FOUR CLUSTERS--AREAS IN WHICH ATTITUDES OF ABUSERS COULD BE DIFFERENTIATED FROM THOSE OF NONABUSERS--WERE FOUND: FEELINGS OF LONELINESS AND ISOLATION; EXPECTATIONS FOR PERFORMANCE OF CHILDREN; RELATIONSHIP WITH OWN PARENTS AND SPOUSE; AND FEELINGS OF ANXIETY IN RESPONSE TO CHILDREN'S BEHAVIOR. IT APPEARS THAT, RATHER THAN DEFINING A POTENTIAL ABUSER BY HIS OR HER OVERALL SCORE ON THE QUESTIONNAIRE, THE MOST SUCCESSFUL APPROACH PROBABLY WILL LIE IN DEFINING A NORMAL PERSON AS HAVING SCORES THAT LIE WITHIN A CERTAIN RANGE FOR EACH OF THE CRITICAL ATTITUDE AREAS. ONCE A NORMAL RANGE IS SPECIFIED, A RESPONDENT'S SCORE CAN BE PLOTTED AGAINST THE NORM, AND THE NUMBER OF DEVIATIONS FROM NORMAL LIMITS NOTED. A THIRD STUDY, USING A REFINED VERSION OF THE 82-ITEM QUESTIONNARIE, WAS BEGUN IN 1971. ATTITUDE AREAS INCLUDED IN THE REVISED VERSION ARE FEELINGS OF ISOLATION, REACTION TO CRITICISM, FEELINGS TOWARD SPOUSE, PARENTAL TREATMENT, EXPECTATIONS CONCERNING CHILDREN, ATTITUDES TOWARD PUNISHMENT, AND FEELINGS OF BEING NERVOUS, UPSET, AND POTENTIALLY LOSING CONTROL OF BEHAVIOR. A LIST OF REFERENCES AND SAMPLE QUESTIONNAIRE ITEMS ARE INCLUDED. (LKM)