Evidence-Based Decision Making: A Behavioral Economist's Case

Nudge theory proposes improving decisions without restricting freedom. Key principles include: making desired choices the default (90% vs 50% retirement savings), reducing friction (deposits cut restaurant no-shows from 14% to under 3%), and leveraging psychology (the fly in urinals reduces spillage). However, the same principles can be misused—for example, casinos and online gambling exploit these mechanisms to maximize spending. The distinction between good and bad nudges depends on intent and application.

Nudge Theory, developed by Thaler and Sunstein, is based on Kahneman and Tversky's work on fast and slow thinking. It describes 'soft paternalistic nudges' that steer people toward decisions in their best interest without restricting choice. Unlike forceful methods (shouting, counting calories), nudge structures environments to influence behavior. Choice architecture reflects limited rationality—only 5% of daily decisions are truly deliberative, while cognitive biases heavily influenced by context drive most decisions. The environment steers decisions whether we use it or not. Examples include supermarkets placing items at eye level, school canteens normalizing healthy choices, and NHS hand hygiene stations with messaging on door handles.

Nudge theory, developed by Richard Thaler and Cass Sunstein, proposes that policymakers can improve decision-making by redesigning choice environments (choice architecture) to make beneficial options more accessible and attractive without restricting freedom of choice; this approach recognizes that humans are not perfectly rational but have predictable biases such as present bias, inertia, and unrealistic optimism, and that small changes in how options are presented—such as changing defaults from opt-out to opt-in—can significantly improve outcomes in areas ranging from retirement savings to organ donation to climate change mitigation.

Nudge Theory is based on three assumptions: people have limited rationality, limited willpower, and are bound by self-interest. It proposes that small environmental features (nudges) can alter behavior subtly. Choice architecture involves creating choices in environments to make intended outcomes more likely. Examples include placing candy bars near checkout lines to increase sales, pre-slicing apples to increase consumption, and placing fly stickers in urinals to improve accuracy. The theory shows how presenting information differently can influence behavior without restricting choices.

Nudge theory explains how people's choices are often not rational and may not be in their best interest, even when they don't realize it. Choice architecture refers to how situations can be arranged to encourage certain choices over others without restricting options. For example, experiments in schools showed that when healthy foods are placed prominently in cafeterias, students eat more nutritious meals, while less healthy options are consumed less. This demonstrates how subtle environmental changes can guide behavior without forcing people.

This educational session explains how to establish rigorous causal inference in program evaluation using quasi-experimental methods like difference-in-differences, regression discontinuity, and instrumental variable analysis when randomized controlled trials are impractical, emphasizing that good data quality is essential for valid evaluation and that researchers should plan evaluation protocols in advance to ensure meaningful outcomes.

Program evaluation differs fundamentally from research in purpose and approach—while research generates new knowledge, evaluation informs decision-making for stakeholders. The six-step evaluation process includes engaging stakeholders, explaining program scope, selecting appropriate design, collecting data, generating conclusions, and providing recommendations. Two primary evaluation types serve different purposes: formative evaluation occurs during implementation to enable ongoing adjustments and feedback, while summative evaluation assesses final outcomes to determine objective achievement and guide future directions. Three core quantitative evaluation types exist: cost-benefit analysis compares financial costs against monetary benefits to determine if programs generate positive returns; cost-effectiveness analysis evaluates interventions by comparing costs against non-monetary outcomes like lives saved or cases prevented, using natural health units rather than dollars; outcomes assessment determines whether programs achieved intended goals through systematic comparison of actual versus expected results, exploration of contributing factors, stakeholder feedback collection, and application of findings to improve practice.

Organizations can measure program success using numerical and quantitative metrics. The video describes how organizations evaluate cost per conversion, productivity, and youth empowerment across different regions. These metrics enable objective assessment of program effectiveness, allowing for data-driven decision-making about resource allocation and program continuation.

To determine adequate sample size for quantitative program evaluation, researchers must first identify the unit of measurement (individual, classroom, school, or district level) and the sampling unit, then apply either confidence interval calculations for descriptive studies or statistical power analysis for quasi-experimental designs, using tools like the sample size workbook or PowerUp! to calculate the required number of participants based on factors such as effect size, standard deviation, desired confidence level, and acceptable margin of error.

Quantitative methods for alternative evaluation involve: (1) Establishing criteria for each alternative, (2) Assigning coefficients or weights to criteria (since different criteria may have different importance depending on the context), (3) Rating each alternative against each criterion using a scale (high, medium, low), (4) Converting qualitative ratings to numerical values, (5) Multiplying ratings by coefficients to get partial scores, (6) Summing partial scores to get total scores for each alternative, (7) Selecting the alternative with the highest score as the prioritized option. Program and project design elements include: (1) Diagnosis of problems and identification, (2) Baseline data, (3) Target population analysis, (4) Supply study (how services are currently being provided), (5) Demand study (how population is requesting services), (6) Context analysis, (7) Actor identification, (8) Cause and effect analysis (problem tree), (9) Means and ends identification (objective tree), (10) Intervention area analysis for viability and importance. Implementation involves: (1) Formulating solutions based on diagnosis, (2) Selecting alternative solutions, (3) Defining objectives and targets, (4) Describing product objectives at service level, (5) Defining project/program horizon, (6) Defining production targets and plans, (7) Describing technology and production processes, (8) Identifying variables and indicators for management control, (9) Selecting information sources, (10) Characterizing assumptions, (11) Creating a logical framework of evaluated alternatives.

Distinguishing legitimate behavioral interventions from manipulation requires clear criteria. Providing concrete information about medicine side effects, allergen warnings, or appointment reminders is educative, not manipulative. Sludge reduction—simplifying forms or removing unnecessary interviews—is inherently freedom-preserving. Concerns about manipulation are legitimate but most genuinely helpful interventions align with freedom rather than coercion. However, ethical boundaries exist: automatically enrolling people in political parties or assuming religious views would violate rights. A framework for ethical behavioral interventions is needed, recognizing that constitutional and human rights apply equally to behavioral interventions as to traditional policy approaches.

Behavioral interventions should be evaluated through two ethical dimensions: manipulation (considering the level of control and transparency of how an intervention works) and paternalism (considering the consequences, affected parties, and strength of people's preferences). The level of control exists on a spectrum from providing information to eliminating choice, while transparency involves making interventions public before implementation to allow critique and avoid self-serving bias.

Ethical objections to behavioral interventions often arise from concerns about treating people like subjects in animal experiments. However, the fundamental principles of stimulus control and reinforcement can be applied to humans while respecting important differences. Staff attitudes toward behavioral interventions can change significantly through direct experience. Nursing staff initially skeptical of behavioral approaches developed positive attitudes after firsthand exposure to how interventions work in practice. Research comparing nursing staff attitudes found that those on behavior therapy units felt the service was more humane and respectful.

Behavioral intervention programs must consider ethical implications of their methods. Programs that use fear, coercion, or punishment as primary tools may cause psychological harm to participants. Effective programs should prioritize the well-being and dignity of participants while still addressing behavioral issues. The case demonstrates how organizations that use fear-based methods may cause lasting psychological harm to vulnerable individuals, raising important questions about the ethical boundaries of behavioral intervention and the responsibility of those who implement such programs.

This section provides a comprehensive framework for implementing behavioral interventions while addressing ethical concerns. The section outlines three core strategies: automation and default options are highly effective when people receive benefits automatically unless they opt out; reminder strategies help overcome human limitations like limited attention and unrealistic optimism; and making tasks enjoyable represents a new frontier in behavioral science. The section emphasizes that while behavioral science has achieved remarkable results (rated 3-8 out of 10), failures still occur and small-scale testing should precede large-scale implementation. Surveys across 17 nations reveal that people generally support interventions that promote their well-being and align with their values, but are concerned about interventions that conflict with their values or sacrifice their interests. Transparency is essential—people are uncomfortable with manipulation even when outcomes are beneficial. Consent and autonomy are critical; people feel violated when interventions are implemented without their consent. The section introduces libertarian paternalism, which guides behavior toward better outcomes while preserving individual freedom, as exemplified by GPS navigation systems.