The reliability and validity of in-training evaluation reports in obstetrics and gynecology.

In a specific Obstetrics and Gynecological program, the program and certifying ITERs were evaluated for their measurement qualities. The internal consistency of the ITERs is supported. The tendency for high inter-item correlations suggest overall judgment of candidates may be influencing individual item rankings--particularly on the Program ITER. Unfamiliarity of faculty with appropriate behaviors may be one of the reasons for this effect based on the faculty's inability to select correct behavior for each item. Very limited inter-form consistency is noted and random associations of items often correlate higher than parallel items. The stability of the Program ITER is supported, but there is little support for criterion validity based on the criterion variables available. It is concluded that more clearly defined behaviors must be identified for each ITER item and faculty must be trained in their use. The use of the same ITER for all specialities may be a major reason for this inconsistency. Lastly, more studies of validity are advised.

Transfer of student learning in medical education.

Transfer of learning among the cognitive, psychomotor, and affective domains and among three clinical disciplines--medicine, pediatrics, and surgery--was examined in the final year of a medical student clerkship program. Rather than following the traditional psychological approach, the authors used a model based on ethnographic analysis followed by performance measurement. The general assumption that what students learn in one subject area is transferred to another area was questioned. A total of 120 medical students were assessed in knowledge, skills, and attitudes prior to and after three clinical rotations. The results indicated that learning was discipline-specific as well as specific to the learning domain studied. Therefore, no generalizations can be made from the study of one subject area or one specific aspect of student learning to another. The use of an alternative methodology to study complex behavior rather than the traditional paradigm is suggested.

[Computer models of personality: implications for measurement].

Computer models of human behavior have long existed in the world of science fiction; however, in reality, progress has been slow. Research has concentrated, in recent years, on the simulation of memory and cognition. Computer models of personality, although fascinating and potentially useful, have been neglected. This paper reviewed the research done to date, categorizing it under five headings: (a) models of belief systems; (b) models of interpersonal behavior; (c) models of decision-making processes; (d) prediction models; and (e) theory-based simulations of specific processes. One practical application was then explored in depth, that of using the computer models in personality measurement. Although some practical development of the working simulations would have to precede their application to personality measurement, it was felt that such an application would be feasible and useful.