Associations Between School Connection and Non-Cognitive Skills: Moderation By Family Income
*Names in bold indicate Presenter
While a broad literature suggests that school connection positively influences SE skills, much of the extant research relies on cross-sectional samples of unrelated youth, resulting in several threats to causal inference, including reverse causality and omitted variable bias. This project, therefore, examines the relationship between school connection and SE outcomes in a longitudinal, national dataset using various estimation strategies to address causal threats. Specifically, it uses a lagged dependent variable (DV) model, a sibling fixed effects model, and a first difference model to explore whether the association between school connection and delinquency, depressive symptoms, and self-worth is robust to rigorous specifications, and whether these associations vary by family income.
Data are drawn from the Maternal and Child Supplement to the National Longitudinal Survey of Youth, the children of female respondents of a national sample of youth aged 14-21 in 1979. The sample consists of a repeated cross section of children aged 10-15 during 1988-2012. School connection is measured with 9 items assessing youths’ subjective feelings of belongingness in school, including whether teachers are responsive to students’ personal problems and academic interests. Socio-emotional outcomes include delinquent behavior, a sum of 5 items assessing theft, violence, and truancy; depressive symptoms, 9 items assessing sadness, listlessness, and melancholy; and self-esteem. A three-level income variable is drawn from mother’s report on family income across childhood. A robust set of child and family covariates are included in all models.
Preliminary findings suggest that there is a moderate, beneficial association between school connection and all SE outcomes tested. Effect sizes range from 0.20 SDs in lagged DV and family fixed effects models to 0.10-0.15 SDs in first difference models. This relationship was largely equal among income groups, with the exception of some evidence for a slightly more beneficial relationship between connection and depressive symptoms for upper-income youth. Importantly, however, low-income youth in these data report levels of school connection that are 0.25 SDs lower than their high-income peers (90/10 gap, e.g. Reardon, 2011). To the extent that school connection may be an important avenue for raising students’ SE skills– thus influencing long term educational, labor market, and health outcomes– educational policies and school practices must consider this important construct, particularly policies designed to narrow income-based gaps in academic achievement. To continue to assess this relationship, next steps include replicating these analyses in the National Longitudinal Study of Adolescent Health, and exploring long-run outcomes.