The Unfinished Path: Discontinuous Growth Modeling of Relative Wage Trajectories for High School Graduates with Some College Experience

Background/Purpose: Postsecondary education is a common pathway for pursuing upward social mobility; however not all students who attend college earn a diploma. Extant research is limited by both data availability and statistical methodology, typically using a regression framework to predict future annual wages. Applying this approach limits researchers’ understanding of the multiple status changes that occur as high school (HS) graduates transition into and out of postsecondary and the workforce over time. Consequently, the relative returns of postsecondary education remain a matter of debate; especially for students who attend some college without graduating.

Recent federal investments in Statewide Longitudinal Data Systems have resulted in large multi-sector linked-administrative data systems, prompting opportunities to apply advanced methodologies to understanding the relationship between postsecondary degree pathways and wage trajectories. In this presentation we illustrate an application of discontinuous growth modeling (DGM; Bleise & Lang, 2016; Singer & Willett, 2003) to answer the question:

Among HS graduates, what are the effects of leaving college without a degree on immediate workforce earnings and earnings trajectories over time?

Study design: Data and Sample: We use population-level longitudinal linked-administrative data, including statewide public PreK-12, postsecondary, and workforce records. We followed an 8th grade cohort for 12 years including 35 quarters of wage data (N= 40,243; observations=604,857).

Measures: Dependent variable. Inflation-adjusted log-transformed quarterly wages. Independent variables. College enrollment (2- and 4-Year), college completion (associates, bachelor’s) and leaving college without a degree (some college). Time was measured in financial quarters. Covariates. Student characteristics, jurisdiction, and employment sector.

Analytic/statistical approach: We used DGM to investigate the relationship between log-transformed wages and educational attainment over 35 financial quarters post-HS graduation. Discontinuities (intercepts and slopes) were introduced for enrollment in college, college degree attainment, and some college.

Results: Wage growth followed a quadratic pattern (e.g., initial increase, followed by a deceleration). Students’ wages dropped and wage growth slowed upon postsecondary enrollment, relative to those with no postsecondary enrollment (HS Only). Earning a Bachelor’s, but not an associates, resulted in an immediate wage increase and greater wage growth compared to the HS only group.

There was no significant difference in the initial wages earned by students with some college relative to the HS Only group–meaning that some college students did not fall behind their HS Only counterparts in terms of wages while attending postsecondary. Students with some college experienced greater wage growth post-exit compared to the HS Only group, though the slope for 4-year students was steeper than for 2-Year students.

Conclusions/Implications: Using DGM we modeled the complex educational and labor pathways taken by a statewide cohort of HS graduates. We found that postsecondary attendance resulted in a wage advantage upon exit, even for those who did not graduate. By including intercept and slope terms for each postsecondary transition, we captured both the immediate and ongoing effects of each discontinuity on wages. A discussion will focus on the strengths and limitations of DGM for social work research and will provide practical guidance on issues such as modeling time, coding discontinuities, and interpretation.