Studying Equity Issues through Secondary Data Analysis: Lessons Learned using Datasets Large and Small

• In Event: 2-010 - Utilizing Multiple Data Sets

Abstract

A wide range of publicly available data are available to developmentally-oriented education researchers. This paper highlights three studies, each focused on equitable developmental outcomes for youth, that draw on different types of data for secondary data analysis. Our overarching aim is to share lessons learned in using such datasets, large and small. Study 1, a recently completed project, utilizes a community sample (n = 261) to examine how marginalized youths’ participation in social action supports career development. Using longitudinal SEM across four waves of data (collected when participants were about 17, 19, 21, and 29 years old), this study explores how social action and career expectancies in adolescence shape occupational attainment in adulthood (see Figure 1). Given the role of occupational attainment in social mobility, results suggest an important mechanism by which societal inequities may be narrowed. Study 2, another recently completed project, draws on nationally-representative data (ECLS-K; n = 8,500) to assess the longitudinal effect of teacher expectations on student achievement in literacy. The study tests an autoregressive cross-lagged model with five time points (grades K-8) and uses multi-group modeling to examine differences in teacher expectancy effects on achievement, based on gender and ethnicity (see Table 1). Results indicate that effects of teacher expectations on future student achievement increase after kindergarten and remain relatively stable, regardless of gender or ethnicity. Yet, patterns of association between teacher expectations and student achievement within a school year vary significantly between White students and students of color. Study 3, a project just initiated, capitalizes on the affordances of rich, population-based data in the newly-developed Stanford Education Data Archive (SEDA). Coupling SEDA’s data with data from other state-based agencies (e.g., State Departments of Education), this study assesses the effects of income inequality, poverty, teacher turnover, and teacher match on student achievement. It also examines how effects differ by grade (grades 3-8), subject area (ELA-math), and locale (i.e., city, suburban, town, and rural). Path analyses, a special case of SEM, are being used for analyses. Broadening the conversation beyond the specific contents of these three exemplar studies—and beyond the theme of equity, which ties them together—this paper will present opportunities and challenges faced by authors using secondary data to address substantively meaningful research questions. Specific issues addressed will include missing data (Study 1), merging data sets across waves (Study 1 and 2), and combining data sets and sources (Study 3), among others.

Authors

• Luke J. Rapa, Clemson University
  Presenting Author

• Faiza M. Jamil, Clemson University