Panel Paper: Experimental Impacts of Customized Information on College Aspirations

Friday, November 9, 2018
8222 - Lobby Level (Marriott Wardman Park)

*Names in bold indicate Presenter

Albert Cheng and Paul E. Peterson, Harvard University


In this paper, we conduct an experiment embedded in a survey to test whether the provision of customized information about the costs and returns to completing a post-secondary education affects parents’ post-secondary aspirations for their children. In all prior experiments that provide information, parents are given a national average of net costs and returns. This approach masks local idiosyncrasies in college costs and returns. National level cost figures also mask variation in net price across families with different household income levels, given the variety of financial aid packages that are available to them. Few experiments also provide this information separately for two- and four-year post-secondary options.

To improve on the limitations of prior research, we experimentally provide returns information at the census tract level and college net costs at the state level. We additionally adjust net-cost information by the respondents' household income. Respondents are randomly-assigned one of four treatment conditions in which they receive (1) no additional information, (2) customized information about net costs, (3) customized information about returns, or (4) customized information about net costs and returns. We then ask whether respondents would prefer that their child enroll in a two-year program, four-year program, or neither.

More specifically, respondents receiving net cost information are told: “According to recent estimates, it costs a student with a household income of ___ in your state ___ per year to complete a four-year bachelor’s degree at a public university, while it costs ___ per year to complete a two-year associate’s degree at a public community college in your state. These are average costs (including tuition, fees, and room and board) after deducting the amount that students typically receive in scholarships and grants.”

Respondents receiving returns information are told: “Workers in your local area who have a four-year bachelor’s degree typically earn ___each year over the course of their working lives. Those who have a two-year associate’s degree typically earn ___. Those without either a four-year bachelor’s degree or a two-year associate’s degree typically earn ___.”

We additionally test potential mechanisms behind treatment effects. Prior to the experiment, we ask respondents to estimate local net costs and returns for families with similar household incomes. With this information, we test (1) the accuracy of respondents’ perceptions of net costs and returns to postsecondary education and (2) whether treatment effects are attributable to families updating these perceptions.

Data for the experiment come from the Education Next poll, which will be administered to a nationally-representative sample of at least 1,500 U.S. parents in the summer of 2018.