Predicting Freshman Grade‐Point Average from Test Scores: Effects of Variation Within and Between High Schools

Abstract

Most studies predicting college performance from high‐school grade point average (HSGPA) and college admissions test scores use single‐level regression models that conflate relationships within and between high schools. Because grading standards vary among high schools, these relationships are likely to differ within and between schools. We used two‐level regression models to predict freshman grade point average from HSGPA and scores on both college admissions and state tests. When HSGPA and scores are considered together, HSGPA predicts more strongly within high schools than between, as expected in the light of variations in grading standards. In contrast, test scores, particularly mathematics scores, predict more strongly between schools than within. Within‐school variation in mathematics scores has no net predictive value, but between‐school variation is substantially predictive. Whereas other studies have shown that adding test scores to HSGPA yields only a minor improvement in aggregate prediction, our findings suggest that a potentially more important effect of admissions tests is statistical moderation, that is, partially offsetting differences in grading standards across high schools.

High School GPAs and ACT Scores as Predictors of College Completion: Examining Assumptions About Consistency Across High Schools

“…For judging college readiness, school-average ACT scores would provide a stronger prediction than students’ individual scores. This is consistent with the findings and recommendations in Koretz and Langi (2018) and Bowen et al (2009). The same pattern is observed with school-average poverty levels (in models that do not control for average ACT scores), which echoes Rothstein’s (2004) findings.…”

“…The similarity in results that are available from studies of schools in other places provides some indication of their generalizability. Studies that used data from samples that include 21 prestigious flagship universities from across the country and all public universities in four states (Bowen et al, 2009; Koretz & Langi, 2018; Rothstein, 2004) all showed that HSGPAs are strongly related to either college graduation or to college freshman GPA and that students’ individual ACT or SAT scores add only modestly to the prediction beyond HSGPA, if at all, in models that include high school fixed effects and compare similar colleges. The graduation rates presented by Bowen et al (2009) for specific HSGPAs are also similar to the graduation rates found here and shown in Figure 1.…”

“…It is possible that the overall achievement level in a student’s school—information that is publicly available—might be just as useful, or more useful, than individual students’ test scores. Two prior studies noted that the size of the relationship between HSGPA and college outcomes (graduation or college GPAs) is larger among students within the same high school (i.e., when high school fixed effects are used in a model) than across schools (Bowen et al, 2009; Koretz & Langi, 2018). They interpreted this as meaning that HSGPA represents a higher level of achievement at some schools than others, which would be consistent with the frogpond effects discussed previously, and suggested adjusting for these differences with information on school average achievement.…”

Is There a Real Need for the Preparatory Years in Higher Education? An Educational Data Analysis for College and Future Career Readiness

“…However, the comparison made in this study between the performance of students coming from high school and those who passed the preparatory year shows that direct admission from high school to colleges is less convincing in assessing students' readiness to enroll in colleges. These results have been confirmed by Koretz and Langi (2018), who found that high school grades are not sufficient to predict students' performance at university because grading standards vary significantly.…”

“…Using a component analysis approach to stem a decision tree, the study found that the data need to be applied as a whole for a successful prediction of grades, particularly to predict failure. Koretz and Langi (2018) found that grading standards in high schools vary significantly; therefore, they are not sufficient to predict performance at university. As a result, the researchers used regression models to predict the college performance of freshmen based on scores in high school and admission and state tests.…”