Recent studies find that intensive college advising models can significantly improve college choice and later success among lower-income students (e.g. Barr and Castleman, 2021; Bettinger and Evans, 2019; Castleman, Deustchlander, and Lohner, 2020). However, the impacts are more modest for remote advising programs like the CollegePoint initiative (Sullivan et al, 2021). In an attempt to further improve college choice and success, other programs have included a financial incentive component; the evaluations of these programs show that its the combination of the financial incentives and intensive support services that drive the large impacts (Angrist, Lang, and Oreopolous, 2009, Carrell and Sacerdote, 2017).
In other education contexts, research suggests that incentivizing student inputs (e.g. time spent in a learning module) can be more effective and efficient than incentivizing student outputs (e.g. text scores) (Clark et al, 2020; Fryer, 2011, Hirshleifer, 2021). This pattern of findings is likely due to students having a high degree of control over task-oriented inputs, while facing greater uncertainty about outputs that occur further in the future and not possessing the necessary skills to effectively improve the outcomes on which they are assessed on their own.
In this paper, we test the addition of input-based incentives to an existing remote advising program targeted toward high-achieving, low- to moderate-income high school students. This population is of particular interest due to the high degree of undermatch based on college-quality (Hoxby and Avery, 2013) and the relationship between college quality and social mobility (Chetty et al, 2017).