Association for Public Policy Analysis & Management

Making choices about college is difficult. Students may not have accurate information about the costs and benefits of higher education. Student beliefs are typically found to be noisy or biased (Dominitz & Manski, 1996; Avery & Kane, 2004; Wiswall & Zafar, 2015; Huntington-Klein 2015). The College Scorecard is a website that attempts to address this information deficiency, with the intent, in the words of Secretary of Education Arne Duncan, to “help families make smart decisions about where to enroll.” The Scorecard provides information about thousands of different United States colleges, with an emphasis on graduation rate, tuition, and median earnings ten years after graduation among Pell Grant recipients.

There are reasons to doubt, however, that the free access to information from the College Scorecard will have much of an effect on student behavior. Not all students are likely to seek out the information. Also, a growing number of research interventions attempt to change student beliefs about the financial costs and benefits of college as a means of affecting behavior (Jensen, 2010; Oreopoulos & Dunn, 2013; Fryer, 2013; Hoxby & Turner, 2013; Wiswall & Zafar, 2015; Hastings, Neilson, & Zimmerman, 2015; Kerr, Pekkarinen, Sarvimaki, & Uusitalo, 2015). Even in these studies, where every student is known to be exposed to information, nearly all find that information about earnings alone has relatively small effects on behavior among students in developed nations.

This paper studies the effects of the College Scorecard launch in September 2015 on student interest in different colleges. Student interest is measured by the volume of Google searches conducted about each school. Search volume does not capture actual matriculation, but it allows for the observation of interest even among students who would not be accepted to a given college, and allows for an estimate of effectiveness years before application data is available. Data consists of weekly Google Trends indices recording the popularity of search terms related to each college from March 2013 to March 2016 (see Stephens-Davidowitz & Varian, 2014). Google Trends data is linked to College Scorecard information about each college.

The effect of the College Scorecard is estimated using a college fixed effects model with trend controls and interactions between college characteristics, such as tuition or median graduate earnings, and an indicator for the post-Scorecard period.

The College Scorecard is found to have statistically significant effects on search behavior. Colleges with higher median graduate earnings and colleges with lower tuition are both searched for more often relative to other colleges after the launch of the Scorecard. However, the effect size is small. A $1,000 increase in median graduate earnings or decrease in tuition increases search popularity by less than 1% of a standard deviation.

The Scorecard has a negligible effect on overall search, but this is an intent-to-treat estimate. The study also provides estimates of the treatment-on-the-treated under differing assumptions about the use of the Scorecard.