Panel Paper: Improve College Readiness through High School Remediation: The Impact of Targeted Interventions in Kentucky

Saturday, November 10, 2018
Lincoln 3 - Exhibit Level (Marriott Wardman Park)

*Names in bold indicate Presenter

Zeyu Xu1, Benjamin Backes1, Amanda Oliveira1 and Dan Goldhaber2, (1)American Institutes for Research, (2)University of Washington

College- and career-readiness for all students has risen to the top of the state and federal education policy agenda. Currently, only a quarter of assessed students meet all four ACT college readiness benchmark scores (ACT, 2014), and more than 1 in every 5 admitted college students are not ready to take credit-bearing college-level courses in their first year of enrollment (Sparks & Malkus, 2013).

Remediation after students enroll in college has generally been ineffective. Bailey, Jeong, and Cho (2010), for example, find that only 11 percent of students referred to developmental math eventually pass introductory college-level math. Consequently, an increasing number of states have attempted to remedy the college proficiency gap at the secondary level (Barnett, Fay, & Pheatt, 2016).

In Kentucky, this type of intervention has been implemented through its Targeted Intervention (TI) program since 2012. The TI program identifies students who are not on track to be college ready when they graduate from high school using a series of tests in Grades 8, 10, and 11. Students who fail to meet test score benchmarks are required by law to receive intervention services or remedial support.

The proposed conference paper—one of a series of studies on the TI program—investigates the impact of 12th grade TI on students’ ability to pass introductory level college courses. TI participation is determined by pre-set test score thresholds (math: 19, reading: 20, and English: 18) on the ACT, administered to all students at the end of the 11th grade. We use regression discontinuity (RD) to estimate the TI impact for students scoring just above and below the cut scores.

Our data include longitudinal student-level administrative records that include high school test scores, student background characteristics, and transcript data from both high school and postsecondary institutions in Kentucky. The data also include student-level intervention records that document intervention type, content area, curriculum, and intensity. The study sample, consisting of the universe of 11th grade students in 2014, has more than 40,000 observations, sufficient to detect effect sizes less than 0.10 standard deviations in a fuzzy RD design.

We find that over half of 11th grade students fail to meet the ACT benchmarks. The discontinuity in the probability of TI participation between students scoring just below and above the threshold scores is about 40 percentage points. Around the cutoffs, we do not detect any discontinuities in terms of student characteristics or prior test scores. There is also no discontinuity of dropout rates in the post-ACT period, which would have led to unbalanced treatment and control samples due to differential attrition. Finally, the distributions of ACT test scores are smooth, indicating no test score manipulation around the cutoff.

Using an RD design, we will estimate the causal impact of TI on students’ propensity to take and pass introductory college-level math and English courses. While results are in progress, they will be ready for presentation at the conference.