Course Introduction
The Propensity Score Methods for Programme Evaluation Training Course provides a comprehensive framework for evaluating programmes and interventions when randomized experiments are unavailable, impractical, or ethically unsuitable. Participants learn how propensity scores can help construct comparable treatment and comparison groups and strengthen observational programme evaluation through systematic adjustment for observed differences between participants and non-participants.
The course develops a strong conceptual foundation in causal inference, counterfactual reasoning, treatment effects, selection bias, confounding, and programme participation mechanisms. Participants examine why individuals or organizations receiving an intervention may differ systematically from those who do not, and how these differences can distort impact estimates. The training demonstrates how propensity score techniques can address observed sources of selection bias when their underlying assumptions are reasonably satisfied.
Participants gain practical knowledge of major propensity score approaches, including propensity score matching, inverse probability weighting, stratification, subclassification, and covariate adjustment. The course explains how treatment assignment models are developed, how propensity scores are estimated, and how different methods affect sample composition, balance, efficiency, and interpretation. Practical exercises reinforce the process of moving from programme evaluation questions to defensible statistical analysis.
A major emphasis is placed on assessing covariate balance and diagnosing the quality of the evaluation design before estimating programme effects. Participants learn to examine standardized mean differences, graphical balance diagnostics, overlap and common-support conditions, extreme propensity scores, model misspecification, and sensitivity to alternative specifications. The training also addresses practical challenges such as missing data, limited overlap, extreme weights, treatment heterogeneity, clustering, and differences between analytical populations.
The programme also explores emerging developments in propensity score methodology and modern impact evaluation. Topics include doubly robust estimation, targeted estimation approaches, machine learning for propensity score estimation, high-dimensional covariates, generalized propensity scores, continuous and multiple treatments, longitudinal treatment settings, and integration with other quasi-experimental methods. Participants examine how modern computational tools can improve estimation and diagnostics while recognizing the importance of transparent assumptions and methodological validation.
By the end of the course, participants will be able to design and implement propensity score-based programme evaluations using appropriate research questions, data structures, estimation strategies, balance assessments, and robustness checks. They will be better equipped to interpret treatment effects responsibly, communicate methodological limitations, assess the credibility of observational evidence, and produce transparent evaluation reports that support evidence-informed programme improvement and policy decisions.
Duration
5 days
Who Should Attend
-
Monitoring and evaluation professionals responsible for assessing programme effectiveness and intervention outcomes.
-
Policy analysts conducting quantitative evaluations of government programmes and public-sector interventions.
-
Economists and econometricians seeking practical expertise in observational impact evaluation and causal inference.
-
Researchers working with survey, administrative, longitudinal, programme monitoring, and beneficiary datasets.
-
Data analysts and statisticians involved in estimating treatment effects from non-randomized programme data.
-
Development practitioners evaluating social protection, education, health, agriculture, employment, and poverty-reduction programmes.
-
Academic researchers and postgraduate students conducting empirical programme evaluation and applied research.
-
Consultants supporting NGOs, governments, development agencies, and institutions with quantitative impact assessments.
-
Programme managers seeking to understand how statistical evidence can be used to assess programme effectiveness.
-
Professionals working with observational datasets where randomized controlled trials cannot readily be implemented.
-
Researchers interested in matching, inverse probability weighting, doubly robust estimation, and modern causal inference techniques.
-
Evaluation specialists using R, Stata, Python, or other statistical software to conduct advanced quantitative programme analysis.
Course Objectives
-
Develop a comprehensive understanding of propensity score methods and their application to causal inference and non-randomized programme evaluation.
-
Explain selection bias, confounding, counterfactual outcomes, treatment effects, and the assumptions required for credible observational impact evaluation.
-
Estimate propensity scores using appropriate statistical models while considering treatment assignment mechanisms, covariates, specification choices, and model adequacy.
-
Apply propensity score matching techniques to construct comparable treatment and comparison groups and estimate programme effects from observational data.
-
Implement inverse probability weighting, stratification, subclassification, and covariate adjustment methods while understanding their distinct analytical properties.
-
Assess covariate balance, common support, overlap, extreme propensity scores, and other diagnostic indicators before drawing conclusions about programme impacts.
-
Identify methodological challenges involving missing data, limited overlap, extreme weights, treatment heterogeneity, clustering, model misspecification, and unmeasured confounding.
-
Apply advanced propensity score approaches including doubly robust estimators, generalized propensity scores, machine learning, and multiple-treatment evaluation methods.
-
Conduct sensitivity, robustness, placebo, and alternative-specification analyses to determine how dependent programme impact estimates are on methodological assumptions.
-
Communicate propensity score evaluation findings clearly through statistical reports, visualizations, policy briefs, and presentations that acknowledge uncertainty and limitations.
Comprehensive Course Outline
Module 1: Foundations of Programme Evaluation and Causal Inference
-
Understanding causal inference, counterfactual outcomes, treatment effects, and the fundamental challenges of evaluating non-randomized programmes.
-
Examining selection bias, confounding, programme participation mechanisms, and systematic differences between treated and untreated populations.
-
Comparing randomized experiments with observational and quasi-experimental evaluation approaches and identifying appropriate applications for each design.
-
Translating programme objectives, implementation mechanisms, beneficiary characteristics, and measurable outcomes into credible evaluation research questions.
Module 2: Propensity Scores and Treatment Assignment Models
-
Understanding the propensity score as a conditional probability of treatment assignment based on observed participant and contextual characteristics.
-
Identifying appropriate covariates for propensity score estimation while distinguishing pre-treatment confounders from mediators and post-treatment variables.
-
Estimating treatment probabilities using logistic regression and other suitable statistical models for binary programme participation.
-
Examining model specification, nonlinear relationships, interactions, omitted variables, and alternative treatment assignment models affecting propensity scores.
Module 3: Propensity Score Matching Methods
-
Understanding nearest-neighbor, caliper, radius, kernel, stratification, and other matching approaches for constructing comparable evaluation groups.
-
Selecting matching algorithms and tuning parameters based on sample characteristics, treatment prevalence, overlap, and programme evaluation objectives.
-
Examining matched and unmatched observations, replacement strategies, matching ratios, discarded observations, and implications for the target population.
-
Estimating average treatment effects using matched samples while accounting for uncertainty, dependence, weighting, and the structure of the evaluation design.
Module 4: Covariate Balance and Common Support Diagnostics
-
Assessing balance between treatment and comparison groups using standardized mean differences and other quantitative diagnostic measures.
-
Creating graphical diagnostics such as propensity score distributions, love plots, density comparisons, and balance plots to evaluate matching quality.
-
Understanding common support and overlap requirements and identifying observations whose propensity scores provide insufficient comparison information.
-
Diagnosing poor balance, extreme propensity scores, model misspecification, and inadequate overlap before interpreting programme impact estimates.
Module 5: Inverse Probability Weighting and Alternative Methods
-
Understanding inverse probability of treatment weighting and how estimated treatment probabilities can create a pseudo-population with improved covariate comparability.
-
Applying stabilized and alternative weighting approaches while examining the effects of extreme propensity scores and influential observations.
-
Comparing matching, weighting, subclassification, and covariate adjustment methods in terms of assumptions, efficiency, interpretation, and practical implementation.
-
Selecting appropriate propensity score methods based on the research question, treatment structure, available data, overlap conditions, and evaluation population.
Module 6: Advanced Propensity Score Estimation and Robustness
-
Applying doubly robust estimation strategies that combine treatment and outcome models to provide additional protection against certain forms of model misspecification.
-
Conducting sensitivity analyses for unmeasured confounding and examining how hidden bias could alter estimated programme treatment effects.
-
Addressing missing observations, extreme weights, limited overlap, influential cases, treatment misclassification, and measurement problems in evaluation datasets.
-
Evaluating alternative propensity score specifications and documenting analytical decisions to support transparent and reproducible programme evaluation.
Module 7: Complex Treatment Settings and Heterogeneous Effects
-
Extending propensity score methods to multiple treatment groups and examining differences in programme impacts across alternative intervention options.
-
Applying generalized propensity score approaches when treatment intensity or exposure varies continuously rather than being limited to participation versus non-participation.
-
Investigating treatment effect heterogeneity across demographic, socioeconomic, geographic, institutional, and baseline outcome characteristics.
-
Examining longitudinal and time-varying treatment settings where programme participation, covariates, and outcomes change across multiple observation periods.
Module 8: Emerging Technologies and Modern Propensity Score Methods
-
Exploring machine learning approaches for estimating propensity scores in high-dimensional datasets while addressing prediction, balance, regularization, and interpretability.
-
Examining modern causal inference approaches including targeted estimation, flexible outcome modelling, and doubly robust machine learning techniques.
-
Applying computational workflows for automated balance diagnostics, sensitivity analysis, reproducible estimation, visualization, and large-scale administrative programme data.
-
Evaluating responsible applications of artificial intelligence in programme evaluation while maintaining transparent assumptions, human oversight, methodological validation, and reproducibility.
Module 9: Applied Programme Impact Evaluation
-
Developing complete propensity score evaluation strategies for programmes in education, healthcare, employment, agriculture, social protection, and economic development.
-
Preparing observational datasets for impact evaluation through data cleaning, variable construction, treatment definition, outcome selection, and appropriate sample restrictions.
-
Comparing estimated programme effects across alternative methods and interpreting differences in results in relation to assumptions, balance, overlap, and target populations.
-
Integrating propensity score methods with complementary approaches such as regression adjustment, Difference-in-Differences, instrumental variables, and sensitivity analysis.
Module 10: Interpretation, Reporting, and Evaluation Practice
-
Interpreting average treatment effects, average treatment effects on the treated, heterogeneous effects, confidence intervals, and practical programme significance.
-
Developing evaluation reports that clearly document treatment assignment, covariate selection, estimation procedures, diagnostics, assumptions, and robustness findings.
-
Communicating technical results through policy briefs, dashboards, visualizations, presentations, and stakeholder reports designed for different levels of analytical expertise.
-
Developing transparent conclusions that distinguish statistical evidence from causal assumptions while clearly documenting limitations and implications for programme improvement.
Training Approach
This course will be delivered by our skilled trainers who have vast knowledge and experience as expert professionals in the fields. The course is taught in English and through a mix of theory, practical activities, group discussion and case studies. Course manuals and additional training materials will be provided to the participants upon completion of the training.
Tailor-Made Course
This course can also be tailor-made to meet organization requirement. For further inquiries, please contact us on: Email: training@upskilldevelopment.com Tel: +254 721 331 808
Training Venue
The training will be held at our Upskill Training Centre. We also offer training for a group (at a discount of 10% to 50%) at requested location all over the world. The Onsite course fee covers the course tuition, training materials, two break refreshments, buffet lunch, airport transfers, Upskill gift package, and guided tour.
Visa application, travel expenses, dinners, accommodation, insurance, and other personal expenses are catered by the participant
Certification
Participants will be issued with Upskill certificate upon completion of this course.
Airport Pickup and Accommodation
Airport pickup and accommodation is arranged upon request. For booking contact our Training Coordinator through Email: training@upskilldevelopment.com, +254 721 331 808
Terms of Payment:
Unless otherwise agreed between the two parties’ payment of the course fee should be done 3 working days before commencement of the training so as to enable us to prepare better.
