The fidelity of a translated s w c y instrument used to measure attitudes toward mental health was evaluated using statistical methods based on item respotwe t h e w . Data from French and German versions of the attitude surycy w m analyzed, and items that displayed differenrid iremfmctioning (dif) werc identified. Item chcvaclerirtic curvu (ICCs) wcre examined to detamine whetha the source of df d d be attributed to errors in translation or differences in cultural experiences or knowledge. The proposal by Humphreys and Hulin for using ICCs to determine the source of dif is evaluated.Measuring attitudes cross-culturally entails the same basic problem that is associated with cross-cultural administration of any paper-andpencil test or survey. The problem is one of establishing the equivalence of the measurement instrument across the groups surveyed. Unless it can be established that the survey instrument itself is equivalent for all the groups, it cannot be used to compare their respective scores.When the comparison groups speak different languages, an additional problem arises. Is the survey more likely to be equivalent if administered in a single language, or should the instrument be translated into the native language of the subject? Some researchers favor the administration of the survey in only one language to bilingual Requests for reprints should be sent to B.B. Ellis,

The fidelity of a translated survey instrument used to measure attitudes toward mental health was evaluated using statistical methods based on item response theory. Data from French and German versions of the attitude survey wm analyzed, and items that displayed differential item functioning (dif) were identified. Item characteristic curves (ICCs) were examined to determine whether the source of dif could be attributed to errors in translation or differences in cultural experiences or knowledge. The proposal by Humphreys and Hulin for using ICCs to determine the source of dif is evaluated.

Student teachers, working teachers, and retired teachers (N = 595) from four countries (France, Germany, Greece, and the United States) completed a questionnaire containing 186 items in Likert format with instructions to describe the mentally healthy person. Principal component analyses of item responses showed two cross-culturally invariant orthogonal factors, which were interpreted as High versus Low Mental Health and High versus Low Behavior Control. Factor loadings revealed a circumplex structure similar to that repeatedly found by Becker in studies of self-description of personality. Four scales were constructed representing the two main axes and the two diagonals of the circumplex structure. The scales were named Mental Health, Behavior Control, Social Adaptation versus Social Maladaptiveness, and Self-Actualization versus Inhibition. Analyses of variance showed cultural and age differences, which were interpreted as reflecting different degrees of permissiveness, varying self-concepts, and of differences in economic wealth of the countries studied.

No abstract

No abstract