Ariel Kalil, University of Chicago
John List and co-authors offer a framework for thinking about investments in early childhood, particularly as a means to close gaps in achievement and behavior between economically advantaged children and their less advantaged peers. The authors point out two relevant facts: first, that gaps in development between rich and poor children appear long before children enter formal schooling, and second, that achievement-promoting interventions delivered early in life appear to yield the greatest impact. These points support an argument for targeting parents and other caregivers for intervention during the earliest years of children’s development.
The authors argue that the application of a behavioral lens to current parental intervention efforts for low-income families could reduce attrition and improve parental engagement with the programs. This could presumably better help to close skills gaps between rich and poor children. Because most large-scale programs suffer from extraordinarily high rates of attrition and lack of parental engagement, and because behavioral tools are generally low cost and light touch, there is on its face little to argue against here. However, the assertion that behavioral tools will solve problems of attrition and lack of parental engagement requires some assumptions. Chief among these assumptions is that cognitive biases, as opposed to real structural issues, inhibit parents’ participation and engagement. It could also be the case that the programs themselves are not adequate because they make the wrong assumptions about parent’s motivations. If this were the case, then program practitioners, instead of or in addition to parents, need the behavioral tools.
The authors describe two different parent-focused interventions in Chicago, each of which included multiple behavioral “nudges” coordinated with an information-focused parenting curriculum. It makes sense to think about behavioral tools in the context of these and related interventions in a public health perspective as the authors describe. Indeed, prior research has described the importance of cognitive biases to parent engagement (see, e.g., Gennetian, Darling, & Aber, 2016; Kalil, 2014; Mayer, Kalil, Oreopoulos, & Gallegos, 2015). Prior research has also suggested a role for behavioral tools to overcome these biases (see, e.g., Gennetian et al., 2016; Kalil, 2014; Mayer et al. 2015; Robinson, Lee, Dearing, & Rogers, 2017; Rogers & Feller, 2016). But I would add another reason to push for the application of behavioral tools to early childhood field experiments. In my view, one of the most exciting possibilities that could arise from this integration is the opportunity to advance theory about parental behavior and the production function of child development.
More specifically, the integration of behavioral tools into early childhood field experiments has the potential to yield new insights into why parents make (or fail to make) decisions to spend time, money, attention, or affection in the service of promoting their children’s development. The field of developmental psychology does not typically characterize parenting in a decision framework, preferring instead to characterize parenting as a multi-dimensional “style” that aggregates parental behavior across a wide number of dimensions. This has greatly impeded our understanding of the decision-making process that leads to specific, measurable behaviors such as reading to young children on a daily basis, enrolling children in early education programs, taking children to school on time every day, or disciplining children with reason as opposed to physical aggression. These behaviors vary widely across rich and poor parents and have been the targets of intervention for many years.
I agree with the authors of this paper on the putative merits of using a suite of coordinated behavioral tools to increase the efficiency and cost-effectiveness of population-level parenting programs. But my push for theoretical understanding as a goal takes a different approach to combining behavioral tools and field experiments in early childhood development than either of the projects detailed in their paper.
Rather, I argue that we should leverage the integration of behavioral tools into parenting interventions during early childhood to identify the specific cognitive biases that impedes specific parental behaviors. As an example, consider the Parents and Children Together (PACT) Study, a field experiment conducted by Susan Mayer and colleagues at the Behavioral Insights and Parenting Lab at University of Chicago (Mayer et al., 2015). This experiment was designed to understand why low-income parents read so infrequently to their children and to test a behaviorally-informed intervention to increase parent-child reading time among low income parents. Extensive survey work conducted as part of this study revealed that parents in this population (a) understood the importance of reading to children (in part this may be due to their attending Head Start, which strongly emphasizes parental reading to children); (b) had access to reading materials (again, possibly due to Head Start making books available to borrow); (c) understood the “production function” of their time investment (i.e., they reported on surveys that the more time they invested reading to their children the greater would be the likelihood of their children’s success in kindergarten); and (d) reported that it was just as much if not more their responsibility (as opposed to the teacher’s responsibility) to stimulate their child’s development. Other work, moreover, shows that low-income parents enjoy spending time in developmentally-relevant activities with their children to the same extent as do high-income parents and that low-income parents do not report feeling particularly stressed or unhappy engaging in learning and teaching activities with their children (Kalil et al., 2017).
What then, is the cognitive bias that leads to infrequent book reading among low income families? The PACT study hypothesized that a very specific cognitive bias – present bias – may be a key culprit and the PACT intervention was designed specifically to overcome this bias with a specific set of behavioral tools (goal setting, feedback, timely reminders, and social rewards) designed to “bring the future to the present” and help parents form a habit of regular book reading. On average, the PACT intervention had a very large treatment impact (~ 1 SD) on the amount of time parents spent reading with their children (the study measured time use objectively using digital tools). But even more important was the study’s finding that the intervention was substantially more effective for those parents who suffered from present bias (the study used standard survey techniques to assess present bias among all of the parents prior to the intervention). In short, parents who suffer from present bias are the very ones who benefit from an intervention designed to overcome it. Those parents who were not present biased were already reading at higher levels to their children and the intervention had little impact on them.
It takes time to collect the relevant data, in multiple forms and from multiple sources, to develop a theory and hypothesis-driven approach to integrating behavioral tools into parenting interventions. Because many behavioral tools are inexpensive, easy to implement, and scaleable, it is tempting to think that more will be merrier in any given intervention. But one should be cautious in doing so. For one, we do not yet know which parental decisions are driven by structural barriers as opposed to cognitive biases. We know little about spillovers, or whether subjecting parents to multiple behavioral tools to drive a single behavior will have synergistic or antagonistic effects. Most importantly, we do not yet know how to make a habit of new behaviors so that cognitive biases are overcome in the long run. A theory-driven approach to integrating behavioral tools into parenting interventions has the potential to answer these questions, thereby advancing science at the same time it improves public health.
The article “Combining behavioral economics and field experiments to reimagine early childhood education” by John List, Anya Samek and Dana Suskind is available free in the May 2018 issue of Behavioural Public Policy
Gennetian, Lisa; Darling, Matthew; and Aber, J Lawrence (2016) “Behavioral Economics and Developmental Science: A New Framework to Support Early Childhood Interventions,” Journal of Applied Research on Children: Informing Policy for Children at Risk: Vol. 7 : Iss. 2 , Article 2. http://digitalcommons.library.tmc.edu/childrenatrisk/vol7/iss2/2
Kalil, A. (2014). Addressing the parenting divide and children’s life chances. The Hamilton Project Discussion Paper (Chapter 2 in The Hamilton Project’s “Policies to Address Poverty in America.”). Washington, D.C.: The Brookings Institution.
Kalil, A., Mayer, S., Delgado, W., & Gennetian, L. (2017). The education gradient in parental time use: Investment or enjoyment? Manuscript under review.
Mayer, S., Kalil, A., Oreopoulos, P. & Gallegos, S. (2015) Using behavioral insights to increase parental engagement: The parents and children together (PACT) intervention (No. w21602). National Bureau of Economic Research. http://www.nber.org/papers/w21602
Robinson, Carly D., Monica G. Lee, Eric Dearing, and Todd Rogers. “Reducing Student Absenteeism in the Early Grades by Targeting Parental Beliefs.” HKS Faculty Research Working Paper Series RWP17-011, March 2017. https://www.hks.harvard.edu/publications/reducing-student-absenteeism-early-grades-targeting-parental-beliefs
Rogers, Todd and Avi Feller. Intervening through influential third parties: Reducing student absences at scale through parents. Goldman School of Public Policy Working Paper, 2016. Available at: http://scholar.harvard.edu/files/todd_rogers/files/reducing_student_absences_at_scale.pdf