| Abstract: | The purpose of the study was the development of a low-inference observation instrument deductively derived from teacher effectiveness literature that would measure the minimal teaching competencies of teaching-candidates. The instrument was comprised of 62 teaching indicators grouped into six training domains: (1) Instructional Planning; (2) Classroom Management and Organization; (3) Instructional Techniques and Materials; (4) Communication and Presentation Techniques; (5) Motivation of Students; and (6) Academic Assessment of Students. Several procedures were implemented to determine the content validity of the instrument; the criterion-related agreement, interobserver agreement, and the intraobserver agreement of the instrument; and the generalizability of the instrument. To determine the content validity of the instrument, the instrument was sent to a panel of judges. Indicators received high ratings in terms of their clarity and importance to the teaching act. Procedures were developed to train observers to use the instrument. Twenty-seven teaching candidates were recruited from a major southwestern university and agreed to be evaluated four times during the course of their student-teaching semester. Two of the four observations were simultaneously coded by both an observer and the criterion person, the developer of the observation instrument. This was done to determine the ability of the observers to use the instrument in actual classroom occurrences with a high level of interobserver agreement. The teaching-candidates were subdivided into three grade level subdivisions, elementary, junior high, and secondary, and represented a number of subject areas. A videotape of actual classroom instructional sequences was viewed by the observers before, during, and after data collection as a means of determining observer consistency in coding and criterion-related agreement. Data analyses indicated that COAST appears to be a generic observation instrument that can be used across subject areas and grade levels. The instrument has a high degree of content validity. Observer agreement measures were calculated using Scott's (1955) coefficient and analyses demonstrated that observers can be trained to use the instrument with high levels of criterion-related agreement, interobserver agreement, and intraobserver agreement. Agreement measures indicate that observers can be trained to use the instrument with high degrees of agreement and consistency. |
|---|