Suspension Profiles: Findings From a Cluster Analysis of School Disciplinary Practices

• In Event: 25.010 - Suspension in Context: A Collaborative Study of the School District of Philadelphia

Abstract

Objectives or purposes
This presentation will highlight the results of a cluster analysis that identifies a number of salient “profiles” of Philadelphia public elementary and middle schools based on extensive teacher and principal reports of disciplinary practices. This presentation also explores the relationships between the profiles and school climate. The primary objective is to address the following research question: What are the attributes that define a typology of Philadelphia K-8 schools with regard to student discipline and suspension?

Theoretical Framework
Much of the research base on student suspension using district datasets has examined suspension ratios over a period of time (U.S. Department of Education Office of Civil Rights; 2014; Gregory, Cornell, & Fan, 2011; Mendez, Knoff, & Ferron, 2002). These analyses ignore information about how schools use suspension and other disciplinary practices and may lead to faulty conclusions. Hierarchical cluster analysis uses a large array of attribute data to identify salient types within a population (Raudenbush & Bryk, 2002). This approach enables us to answer important questions in ways that are informed by holistic information about schools’ climates and practices.

Methods
This presentation will describe our process for identifying school profiles using cluster analysis. We use Ward’s minimum variance clustering method, which is regarded as being one of the most stable cluster linkage methods as it maximizes between-group variance (Ward, 1963). We will also discuss the methods used to explore the extent to which the profiles are associated with other school characteristics and climate. These include analysis ANOVA methods and the Kruskal–Wallis one-way analysis of variance that examine differences between profiles across a range of factors such as school leadership and climate variables and student demographics.

Data sources, evidence, objects, or materials
Our analysis uses: 1) descriptive data from the district on school characteristics; and 2) SY 2014-2015 survey data from Philadelphia teachers and administrators. [we should say more about the survey and how we developed it and how awesome it is] The dataset for the typology analysis are based on responses from a random sample of teachers in all 149 district elementary and K-8 schools. Survey items focused on individual beliefs about discipline and suspension, as well as school-level disciplinary practices and cultures.

Results and conclusions
Using cluster analysis with 55 items from teacher surveys, we explicate a typology of school disciplinary practices. The four profiles that emerge are consonant with qualitative findings.

Scientific of scholarly significance
Our strategy of constructing a typology will allow detailed information at the school level to inform a holistic picture of patterns of disciplinary practices that exist within the district and are distinct from one another in meaningful ways. In addition to identifying the combinations of characteristics associated with patterns of disciplinary and suspension activity, this will provide actionable groupings of schools that, depending on strategic goals, can be targeted with different interventions and implementation supports. This analysis therefore contributes both innovative methodological approaches and new policy insights that can be applied to this pressing problem.
 

Authors

• Philip M. Sirinides, University of Pennsylvania

• Abigail Gray, University of Pennsylvania

• Tesla DuBois, University of Pennsylvania