Introduction to Data Science for Public Policy

Lesson 1 • The Policy Decision-Making Cycle

Maps how evidence enters agenda-setting, design, implementation, and evaluation stages. Shows where data science tools add the most value.

Lesson 2 • Types of Public-Sector Data

Surveys administrative, survey, geospatial, and real-time data sources used in government. Prepares students to identify appropriate data for a given policy question.

Lesson 3 • Ethical Foundations of Policy Data Work

Introduces fairness, privacy, and accountability as non-negotiable constraints. Establishes an ethical lens applied throughout the entire course.

Lesson 1 • Data Documentation and Metadata

Teaches codebook creation, data dictionaries, and provenance tracking. Proper documentation ensures reproducibility and auditability in policy contexts.

Lesson 2 • Finding and Accessing Public Datasets

Covers open data portals, freedom-of-information requests, and API access. Students practice locating authoritative sources for common policy domains.

Lesson 3 • Data Governance and Compliance

Addresses data-sharing agreements, access controls, and regulatory compliance requirements. Students learn to operate within institutional data governance frameworks.

Lesson 1 • Diagnosing Data Quality Issues

Introduces completeness, consistency, accuracy, and timeliness as quality dimensions. Students learn to profile a dataset before any cleaning begins.

Lesson 2 • Standardizing and Transforming Variables

Teaches recoding, normalization, and unit harmonization across heterogeneous government sources. Prepares data for valid cross-agency comparisons.

Lesson 3 • Reproducible Data Preparation Workflows

Introduces scripted pipelines and version control to make cleaning steps auditable. Reproducibility is framed as a professional standard in public-sector analysis.

Lesson 4 • Handling Missing and Inconsistent Data

Covers deletion, imputation, and flagging strategies for missing values. Connects each strategy to its effect on downstream policy conclusions.

Lesson 1 • Visualization Design Principles

Applies principles of clarity, accuracy, and accessibility to chart design. Students learn to avoid misleading visuals common in public reporting.

Lesson 2 • Communicating Findings to Policy Audiences

Adapts exploratory outputs for executive briefings, public dashboards, and legislative reports. Emphasizes plain-language narrative alongside visuals.

Lesson 3 • Univariate and Bivariate Exploration

Teaches histograms, box plots, scatter plots, and correlation analysis. Students identify relationships worth investigating before formal modeling.

Lesson 1 • Common Tests for Policy Data

Applies t-tests, chi-square tests, and ANOVA to typical government datasets. Students select the appropriate test based on variable type and research design.

Lesson 2 • Probability and Sampling Foundations

Reviews probability rules, sampling distributions, and the central limit theorem. Provides the theoretical base for all inferential methods that follow.

Lesson 3 • Hypothesis Testing in Policy Contexts

Covers null hypothesis logic, p-values, and Type I and II errors with policy examples. Students learn to frame testable policy questions rigorously.

Lesson 4 • Multiple Testing and Replication Issues

Addresses p-hacking, multiple comparisons corrections, and replication standards. Frames these as integrity issues specific to public-sector evidence use.

Lesson 1 • Regression with Categorical Variables

Covers dummy coding, interaction terms, and subgroup analysis. Enables comparison across demographic or geographic groups in policy data.

Lesson 2 • Simple Linear Regression

Introduces the regression equation, ordinary least squares estimation, and goodness-of-fit. Students interpret coefficients in units meaningful to policy stakeholders.

Lesson 3 • Logistic Regression for Binary Outcomes

Applies logistic regression to binary policy outcomes such as program participation or recidivism. Students interpret odds ratios and predicted probabilities.

Lesson 4 • Multiple Regression and Confounding

Extends to multiple predictors and addresses omitted variable bias. Students learn to control for confounders when estimating policy-relevant relationships.

Lesson 1 • Regression Discontinuity Design

Exploits eligibility cutoffs to estimate local treatment effects near thresholds. Students apply RDD to programs with score-based eligibility rules.

Lesson 2 • Difference-in-Differences Analysis

Teaches the parallel trends assumption and DiD estimation using panel data. Applied to before-and-after policy rollouts with comparison groups.

Lesson 3 • Randomized Controlled Trials in Government

Covers RCT design, randomization mechanics, and ethical constraints in public programs. Students assess when RCTs are feasible and when alternatives are needed.

Lesson 4 • Instrumental Variables and Matching

Introduces IV estimation and propensity score matching as alternatives when randomization is impossible. Covers assumptions and limitations of each approach.

Lesson 1 • Unsupervised Learning and Segmentation

Uses clustering and dimensionality reduction to identify population segments in administrative data. Supports needs assessment and resource targeting in policy programs.

Lesson 2 • Supervised Learning for Policy Outcomes

Applies decision trees, random forests, and gradient boosting to classification and regression tasks. Students evaluate models using accuracy, precision, recall, and AUC.

Lesson 3 • Deploying and Monitoring Policy Models

Covers model deployment pipelines, performance monitoring, and model refresh cycles. Addresses governance requirements for algorithmic decision-making in government.

Lesson 4 • Fairness and Bias in Predictive Models

Examines how predictive models can encode and amplify historical inequities in public services. Students apply fairness metrics and mitigation strategies to model outputs.