Panel Paper:
How Student Misinterpretation of College Admissions Data Reduces College Aspirations
*Names in bold indicate Presenter
A main feature of Naviance is its scattergrams. These are scatterplots through which a high school student can see, for any college, the GPA and SAT scores of prior applicants from their high school to that college as well as the admissions decision each of these applicants received. The scattergrams contain a circle marking the current student’s SAT and GPA as well as lines indicating the average GPA and SAT scores for previously accepted students (from one’s high school) at the college. Roughly half of accepted students have GPA and SAT scores below these “typical acceptee” lines.
I examine how students’ college choices respond to the information provided by Naviance scattergrams using college application, admissions, and enrollment data from a medium-sized school district in a mid-Atlantic state. I use a regression discontinuity design to assess how students’ decisions are impacted by their positions on scattergrams relative to the typical acceptee. The typical acceptee varies across high schools and over time so I can compare how students with similar academic achievement vary in their choices based on what they observe. The exact SAT and GPA of the typical acceptee has no direct impact on a student’s admissions probability since they represent average scores rather than minimums or thresholds.
However, I find that students react strongly to this discontinuity. Students are 10% more likely to apply to a college if their GPA is just above the typical acceptee’s GPA than if it is just below it. The effects are largest for elite and out-of-state colleges which previously received few applications from the district, and students react more to the typical acceptee’s GPA than SAT. This impact on applications translates into an effect on where students attend college. Being above the typical acceptee’s GPA increases the probability that one attends an elite college by 2.2 percentage points and increases the average annual earnings associated with the college attended by $1,814.
These results indicate that students update their beliefs about admissions probabilities based on the experiences of their peers and the information Naviance provides. This information is particularly impactful in settings where students appear to have less information about a college or their admissions probability. However, students are not using this information in a fully rational way because they are reacting strongly to the scattergrams’ reference points. This is concerning given the widespread adoption of Naviance and other information interventions which may be incorrectly interpreted.