Working With a Low-Inference Discourse Observation Tool: What Can We See?

• In Event: Tracking the Development of Classroom Discussion: What Can a Low-Inference Measure Tell Us?

Abstract

Objectives: Some observational tools are created simply to measure a well-understood set of dimensions within a larger, well-understood domain. Such is not the case when studying the development of classroom discourse: it is hoped that the tool will help us develop better understandings of the underlying domain and its changes across time. The purpose of this presentation is to present data on our study of approximately 69 classrooms, in which trained RAs used the LIDO to track whole class discourse in 4th-7th grade classrooms participating in the Word Generation intervention. Each classroom was observed four times: twice in the fall and twice in the spring. We will present three analyses. First we will present results of regression analyses which explore whether the primary coding categories are sensitive to lesson content (mathematics, science, ELA, social studies), grade level, and school district. Second, we present findings concerning associations between LIDO classroom-level values and teacher questionnaires on the conditions for instruction in their classroom and school (e.g., student behavior).

Third we present data from the parts of the observations when whole group discussion took place, in order to focus in on the important markers of discourse development that differentiate classrooms. This analysis is based on several subscales created within the primary coding categories. For example, we created a subscale called “Press for Reasoning” that is the sum of all teacher prompts for (a) evidence, (b) reasoning, (c) clarification, and (d) response to other students’ reasoning. This subscale is highly correlated with our “Student Transactivity” subscale (Spearman’s rho = .626, p< .0001). (“Student Transactivity” is a student subscale that sums low-incidence but informative moves such as “Student states that s/he agrees/disagrees”, “Student states that s/he changed mind”, and “Student asks another student a question.”)

Values on these and other subscales yield a small array of distinct LIDO profiles: these support a rough typology of classrooms according to the discourse environment. Using these subscales we will describe the changes we see across one year in our sample of classrooms, returning to our model of development introduced in Presentation 1, and discussing areas where it supports or does not support our hypotheses.

Theoretical Framework: This first part of the presentation is primarily descriptive statistics and correlations. The second concerns the meaning of various combinations of values on the subscales described. Thus it draws on the theoretical background cited in presentation one, but applies it to the classroom specifics of our sample.

Methods and Data sources: This presentation will rely on data derived from the LIDO coding of four classroom observations per classroom and teacher questionnaires collected from 95% of the teachers.

Conclusions and Significance: This presentation will argue that the LIDO categories subscales are meaningful indicators of longitudinal development of classroom discourse.

Authors

• Catherine O'Connor, Boston University

• Maria D. LaRusso, Harvard University