High School Academic Rigor and Georgia Tests as Predictors of Post-Secondary Success

May 16, 2012

In spring 2012, the Governor’s Office of Student Achievement contracted with Dr. Ben Scafidi and Dr. Chris Clark to conduct a study comparing End of Course Tests (EOCTs) and Georgia High School Graduation Tests (HSGTs) as predictors of post-secondary success.  The following education update summarizes their findings.

Overview of HSGTs and EOCTs

The goal of HSGTs is to identify students who need additional instruction in the concepts and skills required for a high school diploma. Georgia students seeking a high school diploma who entered high school prior to July 2011 must pass the four content area HSGTs—English Language Arts (ELA), Mathematics, Science, and Social Studies—as well as the Georgia High School Writing Test.  Students who entered 9th grade in 2011-12 or after no longer take the HSGT. Students take all four sections of the HSGT for the first time in spring of their 11th grade year. Students who fail on their first attempt have multiple opportunities to retest, or they can pass an EOCT in the same subject to still graduate on-time.

The EOCTs, on the other hand, are curriculum-based assessments that students take at the end of core courses.  All students enrolled in and/or receiving credit for an EOCT course, regardless of grade level, must take the EOCT. Additionally, after July 2011, the students’ scores must count as 20 percent of their final course grade.  The GaDOE offers EOCTs in 10 courses across mathematics, social studies, science, and ELA. Schools administer EOCTS three times a year upon completion of one of the above courses. Beginning in 2011-2012, the EOCT became Georgia’s high school accountability assessment.

Study Overview

The Clark and Scafidi study examines two issues:

 (1) The impact of academic rigor in high school on success in college, and

 (2) Whether EOCTs are a better predictor of college success than HSGTs.

To examine these questions, Clark and Scafidi used student-level data on all Georgia high school students from 2006 to 2008 who matriculated to a two- or four- year University System of Georgia institution immediately upon graduation.

High School Academic Rigor and College Success

With the above data, Clark and Scafidi constructed a HSGT Rigor Index[1] and an EOCT Rigor Index[2] for each student using the following formulas to estimate the academic rigor of a student’s high school experience.

As an example, students with higher HSGT or EOCT scores have higher index scores than other students with lower test scores but the same high school GPA. In other words, these students had to learn more content knowledge and skills as measured by the HSGTs or EOCTs to earn the same grades as lower scoring students with similar grades. As such, students with higher test scores experienced higher academic rigor than the lower scoring students with the same GPA. Thus, the HSGT or EOCT Rigor indices capture the rigor of a student’s high school coursework.

Using these indices, Clark and Scafidi examined the relationship between high school academic rigor and two measures of post-secondary success:  First Year Grade Point Average (FGPA) and HOPE Scholarship eligibility at the end of the first year in college.

To isolate the relationship between high school rigor and college performance, they used regression analysis controlling for student- and school-level characteristics.[3]  They found that academic rigor in high school has a positive and statistically significant effect on success in college, as seen in the table below.  For example, students with a HSGT Rigor Index in the 90th percentile, meaning they faced high academic rigor, had a college FGPA that was 0.13 points higher than students with a Rigor Index in the 10th percentile even though they had the same high school GPA. On the EOCT Rigor index, the FGPA for 90th percentile students was 0.19 points higher than similar 10th percentile students. For both indices, 90th percentile students were 7 percentage points more likely to retain HOPE eligibility at the end of the first year than 10th percentile students. Therefore, students who faced higher academic rigor as measured by the indices were more successful in the first year of college.

Increase in High School Rigor from 10th to 90th percentile
 
Freshman GPA
HOPE Eligible after First Year
HSGT Rigor Index
0.13
7 percentage points*
EOCT Rigor Index
0.19
7 percentage points*
*This increase translates to an 18.5% increase in freshman HOPE eligibility.
Predictors of College Success

To compare high school GPA, EOCT scores, and HSGT scores, Clark and Scafidi regressed FGPA and HOPE eligibility on student- and school-level characteristics in the same way as the above analysis but included high school GPA, total EOCT score, and total HSGT score independently in the regression. To compare the magnitude of each variable on college success, they used “marginal effects” to estimate the impact resulting from a one-standard-deviation increase in one of the variables. More specifically, it allowed them to predict the difference in college success between two students who were the same in every way except one variable of which they were one standard deviation apart.

As seen in the table below, they found that high school GPA is the best predictor of FGPA and HOPE eligibility. A one standard deviation increase in high school GPA is associated with a 0.535 point increase in FGPA and a 26.5 percentage point increase in the likelihood of being HOPE eligible. Both findings are statistically significant.

Marginal Effects of HS GPA and Test Scores on College Success
 
Freshman GPA
HOPE Eligible after First Year
HS GPA 0.535** 26.5 percentage points**
EOCT 0.116** 3.6 percentage points**
HSGT
0.002
0.6 percentage points*
**Indicates statistical significance at the 5% level. *Indicates statistical significance at the 10% level.
 

EOCT scores are also strong predictors of FGPA and HOPE eligibility. A one standard deviation increase in EOCT score is associated with a 0.116 increase in FGPA and a 3.6 percentage point increase in the likelihood of being HOPE eligible after the first year. Both findings are statistically significant.

While HSGT scores still have a positive association with FGPA and HOPE eligibility, the magnitude is much smaller than EOCT and high school GPA. A one standard deviation increase in HSGT score is associated with a 0.002 point higher FGPA and 1.6 percent increase in the likelihood of HOPE eligibility. However, the FGPA coefficient is not statistically significant and the HOPE coefficient is only significant at the 10 percent level.

In sum, high school GPA is a much stronger predictor of college success than either EOCTs or HSGTs. Additionally, students with high EOCT scores perform better in the first year of college, all else equal. However, students with high HSGT scores do not perform significantly better in the first year of college. As such, EOCT scores are stronger predictors of first year college success than HSGT scores.

Conclusion

In sum, students who experienced higher academic rigor in high school as measured by the HSGT and EOCT Rigor indices earn higher grades during their first year of college and are more likely to retain HOPE eligibility at the end of the first year. In addition, high school GPA and EOCT scores predict college success better than HSGT scores. Since Georgia now exclusively uses EOCTs, this report indicates that the state actually not has more information about student learning prior to college than when HSGTs were in place.

 

[1] Total GHSGT = Total Score on All Four Parts of GHSGT

[2] Total EOCT= U.S. History Scale Score + Economics Scale Score + American Literature and Composition Scale Score. It only includes these three courses to maximize sample size since other EOCTs are offered early a students’ high school career. Since other EOCTs were first given in 2004, they were not widely available for students who graduated between 2006 and 2008.

[3] Student-level characteristics include gender, race, poverty, disability, limited English proficient, graduation class, and SAT score. School-level characteristics include the average student-teacher ratio, percent of teachers with an advanced degree, teacher experience, percent economically disadvantaged, and percent of students by race. The models also included college and high school fixed effects to hold constant any unmeasured differences between schools.