Neighborhood Effects on Teachers and Children: A Causal Study Using Violent Crime

• In Event: 2-179 - Beyond the Census: Understanding How Neighborhoods Matter for Children by Incorporating Novel Data Sources

Integrative Statement

Neighborhood and community contexts of development matter for children. Although older children and adolescents may directly experience neighborhood effects as they encounter community members and navigate the physical terrain and institutions of the neighborhood, very young children largely experience neighborhoods through the lenses of their adult caregivers (Chase-Lansdale, Gordon, Brooks-Gunn, & Klebanov, 1997). Therefore, in early childhood, the way that adults such as teachers experience and respond to neighborhood contexts may play an important role in the effects that neighborhoods have on children.

The fact that teachers and students are not randomly distributed among schools, but instead select into schools based on observable and unobservable characteristics makes drawing causal claims about the effects of school neighborhoods on teachers and students difficult using observational data. However, recent methodological work by McCoy, Raver and Sharkey (2015) allows for causal conclusions to be made from observational data. In this study, we use one particularly salient aspect of neighborhoods, violent crime, to extend a mediation analysis of observational data by making causal inferences about the effects of neighborhood crime on teacher quality, teacher psychosocial factors, and child outcomes.

We draw on data from multiple sources. Crime data for this study are from a Washington Post database of geocoded homicides in the United States’ 50 largest cities. Schools vary from 0 to 86 homicides within a mile over the course of the study (M=15.47, SD=17.86). Child and teacher measures come from the National Center for Research on Early Childhood Education-Professional Development Study, a randomized controlled evaluation of a professional development program for early childhood teachers across eight U.S. cities. For the teacher participants, CLASS™ data on various dimensions of teaching quality (see Table 1), along with measures of state anger and anxiety, were collected. Students from each class were also assessed on vocabulary, literacy skills, and self-regulation.

Using this data, we first use structural equation modeling to test whether the relationship between violent crime and child outcomes can be explained through teacher factors (see Figure 1). We then use a fixed effects method similar to McCoy, Raver, and Sharkey (2015) to draw causal inferences by using the timing of homicides as an exogeneous shock. The radial fixed effects method makes comparisons among teachers and children who experience a local homicide shortly before they are assessed to those who experience a local homicide at the same distance either shortly after or long before they are assessed.

Preliminary correlational findings (Table 1) indicate that crime is mildly correlated with teacher state anger and two subdimensions of emotional support, negative climate and regard for student perspectives, as well as child self-regulation. These preliminary results reveal the importance of testing the causal effects of violent crime on teacher factors and child outcomes and identify potential pathways that will be tested in the final analyses. The final paper will provide updated results on the pathways shown in Figure 1 as well as policy and practice implications for reducing crime around schools and supporting teachers and students.

Authors

• Andrea Kinghorn Busby, Northwestern University
  Presenting Author

• Terri Sabol, Northwestern University
  Non-Presenting Author

• Dana McCoy, Harvard Graduate School of Education
  Non-Presenting Author