Current Research:

Social Capital

What's In a Relationship?: Examining Race, Class, and Contact Time as Determinants of Social Capital in Mentoring Relationships

S. Michael Gaddis

Abstract: Over the past twenty-five years, social capital has emerged as an important yet highly debated concept in social science research. The current literature leaves researchers to question what characteristics of a relationship are important in producing beneficial outcomes. Conflicting theories highlight the importance of differing relationship characteristics, including the amount of time spent between individuals, social class difference and racial similarity. I argue that the answer is unclear due to the measurement of substantially different concepts as well as a persistent endogeneity problem. I examine data from a quasi-experimental design on mentoring relationships and run estimations using propensity score weighting to address selection bias in estimating causal effects of social capital. The results indicate that the amount of time spent in a relationship has a significant effect on both academic and deviant behavioral outcomes for youths, but this finding is moderated through the racial match of a relationship. Additionally, counter to what some theory suggests, social class difference between individuals has no significant effect on any of the examined outcomes.

School Poverty

Contagion, Institutions, or Selection?: The Effect of Peer Poverty on Student Test Score Growth

Douglas L. Lauen and S. Michael Gaddis

Abstract: That impoverished contexts have harmful effects is a widely held belief among social scientists and policymakers. Disentangling the influence of the effects of individual and family background from the effects of context is conceptually and methodologically complex, however, so the hypothesized existence of contextual effects of peer groups, schools, neighborhoods, organizations, and occupations remains open to scholarly inquiry. Of particular importance is evidence that context affects changes in outcomes rather than simply the level of outcomes at one point in time. Also important is evidence that the effects of context persist once the factors predicting selection of context are statistically controlled. This study seeks to contribute to contextual effects research by carefully specifying and accounting for bias from omitted and mismeasured student, family background, and school characteristics. We use a multiple cohort panel design to estimate the effect of peer poverty rates on elementary school student test score initial status and growth. To estimate effects we use interval metric and vertically equated test score outcomes and variation across time in peer poverty from the universe of children in grades three through eight in the state of North Carolina from 2001 to 2006. The study first estimates multilevel growth models, which produce a composite of the within- and between-student peer poverty effect. Consistent with prior research, we find a negative effect of attending school with high poverty peers on student 3rd grade scores. We find no evidence, however, of peer poverty on test score growth over time. To address selection bias, we then use a two-way fixed model to control for time-invariant student and school unobservables. Results from these models fail to reproduce the negative effect of high poverty peers on initial status and also fail to produce evidence that peer poverty affects test score growth. These results raise important doubts about whether impoverished contexts harm either initial status or learning growth in young and adolescent children and provide a set of analytical approaches for scholars in other sub disciplines to consider when estimating contextual effects.

School Accountability under NCLB

Shining a Light or Fumbling in the Dark?: The Effects of NCLB's Subgroup-Specific Accountability Pressure on Disadvantaged Student Performance

Douglas L. Lauen and S. Michael Gaddis