"Trees and People"—A research design for evaluating the outcomes of neighborhood and nonprofit urban forestry: Does planting trees improve neighborhoods?

In this document, we will propose a research design to answer the following questions: 1. Do the institutional designs of urban nonprofit tree-planting programs and neighborhood tree-planting projects affect planted-tree success? 2. Does participation in a tree-planting project have social effects on neighborhoods and individuals? Using data from nonprofit urban tree-planting programs in 5 eastern U.S. cities, we propose to evaluate both ecological and social outcomes of these programs at the neighborhood, individual, and tree level. Outcomes of interest are tree success and whether or how tree planting increases community capacity. Practically, nonprofits hope that their trees survive and grow, and that their tree-planting programs strengthen familiarity and trust among neighbors; increase community capacity to be resilient in the face of external shocks to the community; improve understanding of the benefits of urban trees and awareness of ecological surroundings; and initiate future instances of community collective actions to improve social, public health or environmental conditions in the neighborhood. In short, we propose to examine how people influence trees and how trees influence people. Our research is informed by the Model of Urban Forest Sustainability (Clark et al. 1997) and the social-ecological systems (SES) framework (e.g., Ostrom 2009). To evaluate tree outcomes, we propose a post-test only with stratified random selection of tree-planting neighborhoods and stratified systematic random sampling of trees within neighborhoods. Within selected neighborhoods, we sample fifty percent of planted trees and gather data for each sample tree according to the Planted Tree Re-Inventory Protocol. To evaluate social outcomes, we propose a post-test only with non-random treatment and comparison groups and with stratified random sampling and estimate outcomes using propensity score matching and instrumental variables techniques. We use the same sample of neighborhoods as the tree design and match these neighborhoods to comparison neighborhoods using a suite of covariates to create a similar-looking comparison group. Within neighborhoods, we select a random sample of residents and over-sample participants. We include several mechanisms to reduce selection bias including propensity score matching and two-stage least squares. We will begin pilot research this summer and the full project with all cities in Spring 2014.