This study focused on a comparison of the factor structures of the scores on l8 teacher competencies measured by both low-inference and high-inference observation. The observation inventories used were the Classroom Observations Keyed for Effectiveness Research (COKER) for low-inference observation, and the Teacher Performance Assessment Instruments (TPAI) for high-inference observation. The data were obtained on over 250 student teachers observed in their concluding year of training, in a four-year, competency-based, teacher education program. The factor structures based on the two sets of scores, one from the COKER, the other from the TPAI, were quite different; although variance accounted for was about the same. The TPAI produced a strong, general factor and a second factor concerned with instructional variety. The COKER produced five, consistently strong factors concerned with variety and flexibility, modifying instruction and working with various sized groups, discipline and communication, questioning and discussion, and feedback and correction of cognitive performance. The results indicated little convergent validity between the low-inference and high-inference observation. Measurement in this area appears to be highly instrument dependent. (Author/CM)