Analytic Strategies for Sequentially Coded Classroom Observation Data.

Sequentially coded classroom observation data present considerable problems for data analysis. The typical approach of analyzing only category frequencies loses the potentially important information about classroom interaction which might be gained by looking at sequences of events. However, the analysis of sequences poses data management problems which tend to increase as the sequence length or the number of categories increases. The problems are, paradoxically, those of too much data and too little data--too much data in the sense that even an hour's worth of coding produces a long string of events to be interpreted; too little data in the sense that, as the number of categories or the length of the sequences increases, the number of potential sequences, or cells in the transition matrix, tends to increase beyond the capacity of the collected data to fill all the cells. This study proposes an analytic strategy which reduces the complexity of this data analysis task via a four-stage process: (1)The original categories are collapsed into a set of "macrocategories" based on a primary facet, or dimension, of the category system. This primary facet is determined by the research questions of interest and by the nature of the classroom activities being observed. (2)The initial search for consistent sequences is carried out at the macrocategory level, reducing the number of categories to be managed. The use of the lag sequential analysis procedures developed by Sackett further reduces the complexity in the search for sequences by looking only at two-step transitions at progressively longer lags rather than at the complete contingency table for the entire length of the sequences. (3)Once sequences of the primary facets have been identified, the macrocategories are converted back to the original categories to allow for interpretation of the sequences. Key dimensions of the sequences are identified, and the sequences are recoded on these dimensions. (4)The sequence level data are analyzed both within and across dimensions. The entire four-stage process may be repeated here in order to explore sequences of sequences. This strategy is demonstrated using data from the Massialas, Sprague, and Sweeney (1970) study, Structure and Process of Inquiry into Social Issues in Secondary Schools. The analysis provides information concerning variations in the structure of classroom discussions which was not available from simple category frequency counts. The analytic strategy described in this study is applicable for developing new category sytems or for enhancing the information from established ones. It may be of particular use in making comparisons among different classroom settings or groups at multiple levels of classroom "events." The primary advantage of the strategy is its ability to reduce the complexity of the analytic task, while retaining the potential information from the full category system.PhDEducational psychologyUniversity of Michiganhttp://deepblue.lib.umich.edu/bitstream/2027.42/158349/1/8116337.pd