| Classroom questioning has a long history as an instructional process. Investigation of classroom questioning, is a very labor-intensive endeavor. This study has developed the software tool, QANALYZE, an expert system shell, to aid in examination of classroom questioning. The importance of QANALYZE to the educational research community is its ability to classify questions at least as well as human experts, its capacity for gathering large amounts of data, and its speed. Furthermore it is sufficiently flexible to be used with differing questioning systems. These capabilities reduce the need for human labor formerly employed in classroom questioning investigation.;Both theoretical validity, practical validity, and robustness have been established for QANALYZE in this study. Methodology for testing the theoretical validity involved selecting a panel of five human experts trained to classify questions. This panel classified 358 questions taken from 15 transcripts of questioning sessions. Satisfactory interrater reliability was indicated by use of the chi square procedure. An expected agreement level found in an evaluation of a similar expert system. A Pearson r correlation procedure was performed to test the classification ability of QANALYZE against results produced by human experts. The correlation coefficients suggested that QANALYZE could classify questions in a superior manner to the expert panel. Testing the practical validity of the expert system involved a comparison of the time required by human experts and the time required by QANALYZE for the same task. QANALYZE completed the classification project in less than one-tenth of the time. This performance demonstrated the practical validity of the expert system. Robustness was demonstrated by installing and running QANALYZE on five different computers.;Finally, the study used QANALYZE to examine 36 additional transcripts to produce data that could be used in future investigations of classroom questions. Five potential research questions were proposed using the data provided by the expert system. Other suggestions were made that could improve the software. |
