How can behavioral science be used to improve decision-making in child welfare?


How can behavioral science be used to improve decision making in child welfare?

By Elspeth Kirkman, Senior Director of Health, Education & Communities and Lindsay Moore, Principal Advisor for the Behavioral Insights Team

Behavioral science is evolving, giving new insights into how choices are made and what factors predict or protect against poor judgment. In child welfare, where routine decisions can have life and death consequences, designing moments of choice with this research in mind could have dramatic impact.

A tale of two systems: How decisions are made

Imagine that the brain runs on two systems1: System 1 is fast and automatic, while System 2 is slow and deliberative. To survive, human brains have constructed a division of labor in which System 1 is given as much of the work as possible, acting as a kind of cognitive autopilot that replaces complex decision making with simple guidelines: take the path of least resistance,2.3 only attend to something if it is urgent, watch how others act and follow suit. These rules are allies, serving people incredibly well most of the time. But, like any simplified rules, they also can lead people astray, resulting in poor outcomes for decisions of great importance.

Rule 1: Take the path of least resistance

In a busy courtroom in Israel, judges overseeing a series of cases had to decide whether to grant parole. Researchers interested in judicial decision making noticed a trend: a verdict in favor of parole became less likely the longer the court session ran.4 After a short break, favorable rulings became more likely again, happening around 65 percent of the time. The cases were ordered randomly, so what was the explanation? It turns out, the more cases the judges saw and the longer they went without a break, the more they favored the status quo of no parole. Beyond the courtroom, research conducted in the United Kingdom reveals that the likelihood of a reported case of child maltreatment being screened out changes throughout the week, even when other factors are held constant. This finding suggests that decision-making capacity is at greater risk when the persons making the decision feel fatigued or under time pressure.5

The good news is that processes can be designed to accommodate human nature. The medical profession, for example, introduced the option of delayed antibiotic prescriptions to accommodate the fact that doctors may be overprescribing for very human reasons.6 By giving a script that only can be filled three days after the appointment, doctors can still avoid patient dissatisfaction and repeat visits while doing their part to effectively moderate the use of antibiotics for their patients. The results are striking. In one study, delayed prescribing posed no risk to health outcomes and 69 percent of those who received a delayed prescription did not go on to use the drug.7 To help child welfare professionals make good decisions, systems can design processes and schedule decision points in ways that take into account human conditions like fatigue.

By that point I was so exhausted that I couldn’t decide if I wanted a cup of coffee let alone what to do about the caseload ...

– Hotline and Investigations supervisor

Rule 2: Pay most attention to the things that spring to mind fastest

Humans tend to rely on the things that spring to mind most readily when assessing factors that should go into a decision. Most of the time this shortcut is more helpful than equal consideration of all factors, but there are certain instances in which it fails. Consider the fact that people are intuitively more afraid of shark attacks than rip tides, despite the fact that 10 times more people a year die from being swept up in rip tides than from being attacked by sharks.8 This phenomenon, known as “availability bias,”9 can skew decision making: individuals pay attention to the imaginary fin without heeding the very real undertow.

In the context of child welfare, working against availability bias is a daily challenge: a small number of high profile cases draw attention to the most extreme instances of child abuse, erode trust in the system, and give the public a mental model of abuse that means they are frequently not able to recognize other forms of child maltreatment when they see them. Indeed, even those who work in the system, who know that chronic neglect accounts for three-quarters of child maltreatment cases, may find their frame of reference dominated by the easier recall of shocking acute sexual and physical abuses. For a social worker who spent the day hearing about how a parent beat a child to the point of blindness, it was almost inevitable that a case that would have seemed serious yesterday became trivial today by comparison.

In hospital triage settings, using data on historical cases has proven to be an effective way to improve diagnostic medicine. Indeed, actuarial risk assessment tools have long been a feature of diagnostic medicine and doctors are typically equipped with decision tools such as the Heart Disease Predictive Instrument.

While comprehensive, such tools can also be too complex for fast-paced decision-making environments like the emergency room, meaning doctors might turn to their intuition, overweighing small but salient predictive factors – the ethnicity of the patient, for example – while giving less weight to the less visible signs of cardiac incident. It turns out that introducing a very simple decision aid (shown below) using the top three predictors of cardiac emergency can be enough to improve diagnostic accuracy and patient safety, and reduce over-referral10:

In child welfare, it is unlikely that a three-criteria system will ever be suitable for triage. However, Safety Organized Practice (SOP) offers a powerful antidote to availability bias at every step by using evidence-based tools and methods (including Structured Decision Making) that bring the right information to the fore and ensure consistency in decision making. Apart from structuring the decision process to reduce bias, SOP also helps maintain focus on the overall goal of safety and challenges responses that are driven by emotion, judgment, or compliance.

Such tools – whether for heart attacks or child maltreatment – effectively act as System 2, bringing focus back to the facts and the reminder that a complex calculus exists as an alternative to the simple rules so often relied on. As such, improving the usability and uptake of such tools is already an area of opportunity for many agencies looking to improve decision consistency.

I’d just come out of a four-hour meeting about a child who had been beaten so badly that he was left blind in one eye. Everything else felt ... trivial in comparison.

– Investigations social worker

Rule 3: Follow the crowd wherever possible

People take cues on how to act from others, especially when a situation is uncertain, when disagreement has social consequences, or when they feel as though they are in the minority.11 Being able to detect and follow norms is imperative to survival as social creatures. However, dogged conformity has a downside. In a team environment, especially one with power disparities, people can fall prey to groupthink,12 a phenomenon whereby a collective fails to make a good decision because other factors, such as reaching agreement quickly, override the ability to be critical and realistic. Some child welfare agencies conduct Permanency Teams or Safety Reviews, for example, where the goal is to identify if the child is safe or unsafe through a team process, or to formulate a permanency plan of action. The dynamics in meetings like this are often driven by existing hierarchy or organizational politics, or even by an understanding that “we all know what it takes.” If the whole group discusses the cases and the members jointly decide if the child is safe or unsafe, they may make a worse decision than if each group member independently makes an assessment and cases with disagreement are discussed in depth. Similarly, knowing that others have already made a permanency-related decision can reduce the sense of individual accountability, causing people to act like detached bystanders. For a hotline night shift supervisor tasked with reviewing 50 cases from the day, it may be difficult to see how those judgments could have gone differently since at least one co-worker has already worked on the case and presumably made a sound decision.

We all know what it takes to keep a child safe, so we don’t really disagree on these decisions.

– Social worker at a case review meeting

The good news is that it is possible to guard against some of the pitfalls of groupthink. Even one voice of dissent can be enough to spark change. Group leaders can make all the difference in ensuring the right factors are considered in the right way by explicitly asking for information they do not yet know and will not want to hear, assigning members of the group specific roles in making the decision, and refusing to give their own view until others have given theirs.13 Indeed, these practices are already being used in some child welfare decisions. For example, to determine the right course of action for reports coming into the hotline, the District of Columbia developed a Review, Evaluate, and Direct framework, which triages cases using clear protocols on information gathering, trained facilitators, real-time assessment, and role-play in group discussions.

1 Kahneman, D. (2011). Thinking, Fast and Slow. Macmillan.
2 Kahneman, D., Knetsch, J. L., & Thaler, R. H. (1991). Anomalies: The endowment effect, loss aversion, and status quo bias. Journal of Economic perspectives, 5(1), 193-206.
3 Jachimowicz, J. M., Duncan, S., & Weber, E. U. (2016). Default-Rejection: The Hidden Cost of Defaults. Available at or
4 Danziger, S., Levav, J., & Avnaim-Pesso, L. (2011). Extraneous factors in judicial decisions. Proceedings of the National Academy of Sciences, 108(17), 6889-6892.

5 Danziger, S., Levav, J., & Avnaim-Pesso, L. (2011). Extraneous factors in judicial decisions. Proceedings of the National Academy of Sciences, 108(17), 6889-6892.
6 Little, P. (2005). Delayed prescribing of antibiotics for upper respiratory tract infection: with clear guidance to patients and parents it seems to be safe. BMJ: British Medical Journal, 331(7512), 301.
7 Little, P., Williamson, I., Warner, G., Gould, C., Gantley, M., & Kinmonth, A. L. (1997). Open randomised trial of prescribing strategies in managing sore throat. Bmj, 314(7082), 722.
8 Brander, R., Dominey-Howes, D., Champion, C., Del Vecchio, O., & Brighton, B. (2013). Brief Communication: A new perspective on the Australian rip current hazard. Natural hazards and earth system sciences, 13(6), 1687.
9 Tversky, A., & Kahneman, D. (1974). Judgment under uncertainty: Heuristics and biases. Science, 185(4157), 1124-1131.
10 Green, L., & Mehr, D. R. (1997). What alters physicians’ decisions to admit to the coronary care unit?. Journal of Family Practice, 45(3), 219-226.
11 Cialdini, R. B., & Cialdini, R. B. (2007). Influence: The Psychology of Persuasion (pp. 173-174). New York: Collins.
12 Janis, I. L. (1971). Groupthink. Psychology today, 5(6), 43-46.
13 Sunstein, C. R., & Hastie, R. (2015). Wiser: Getting beyond groupthink to make groups smarter. Harvard Business Press.

This article is the first in a four-part series on decision making and behavioral science in child welfare. The series looks at lessons from other fields and considers their relevance at critical steps in the child welfare system.