Stata Impact Evaluation Data Analysis Training Course

Course overview

Development and humanitarian programmes are routinely expected to demonstrate whether an intervention caused measurable change, not simply whether activities were delivered. M&E teams must work with household surveys, beneficiary registries, baseline and endline datasets, administrative records, and panel data while addressing missing values, non-random programme participation, clustered samples, and pressure to report credible results quickly. This course equips practitioners to conduct defensible impact analysis in Stata and explain the assumptions, limitations, and operational implications behind the findings.

Participants learn to build reproducible Stata workflows for impact evaluations, from importing and cleaning raw survey data through constructing indicators, documenting code, and producing publication-ready tables and graphs. The course covers theory of change-linked indicators, descriptive and balance analysis, randomized controlled trial analysis, difference-in-differences, propensity score matching, regression adjustment, fixed effects, cluster-robust standard errors, and treatment-effect interpretation. Participants also learn how to test identifying assumptions, diagnose common data problems, conduct sensitivity checks, and distinguish statistically significant results from decisions that are meaningful for programme management.

Delivery combines instructor-led demonstrations in Stata with guided coding labs based on a realistic development programme dataset. Participants work through an end-to-end impact evaluation case involving a livelihoods or cash-transfer intervention, make analytical choices, review outputs with peers, and receive feedback on code and interpretation. Each participant leaves with an annotated Stata do-file, a cleaned analysis dataset, a reproducible results folder, and a concise impact evaluation results brief containing tables, figures, findings, caveats, and recommendations for decision-makers.

The course is designed for M&E professionals, researchers, programme analysts, and evaluation consultants who already work with quantitative data and need stronger causal-analysis capability for development or humanitarian evidence products.

Course objectives

By the end of this course, participants will be able to:

• Construct a reproducible Stata project using do-files, log files, macros, relative file paths, and structured output folders
• Clean, merge, reshape, label, and validate survey, beneficiary, and panel datasets in Stata
• Create theory of change-linked outcome indicators, treatment variables, baseline covariates, and analysis-ready samples
• Estimate randomized controlled trial effects using regression adjustment, intention-to-treat analysis, and cluster-robust standard errors
• Apply difference-in-differences models and assess parallel-trends evidence for before-and-after programme data
• Implement propensity score matching and regression-based treatment-effect estimation for non-random programme participation
• Diagnose attrition, missing data, baseline imbalance, outliers, and heterogeneous treatment effects using Stata outputs
• Produce an impact evaluation results brief with reproducible tables, coefficient plots, interpretation notes, and decision recommendations

Benefits of attending

Target competencies

Who should attend

• Monitoring and Evaluation Managers — who need to commission, review, and defend credible programme impact analysis
• Impact Evaluation Officers — who analyse baseline, endline, panel, and administrative data for causal evidence
• Development Programme Analysts — who must turn programme datasets into evidence for adaptation and scale-up decisions
• Humanitarian MEAL Specialists — who assess whether assistance models produce outcomes beyond output monitoring
• Research Officers — who require Stata-based methods for experimental and quasi-experimental studies
• Evaluation Consultants — who deliver transparent, reproducible quantitative findings to donors and implementing partners

Requirements and prerequisites

Participants should be comfortable working with quantitative datasets and interpreting basic descriptive statistics, including means, proportions, cross-tabulations, and regression output. Prior exposure to Stata is expected: participants should be able to open datasets, run commands, and save a do-file, although advanced programming is not required. Familiarity with M&E concepts such as indicators, baselines, endlines, sampling, and disaggregation will help. Participants should bring a laptop with Stata 16, 17, or 18 installed. Prior experience with randomized trials, matching, econometrics, or impact evaluation software is not required; these methods are taught from first principles before implementation.

Training methodology

The course alternates short instructor-led method briefings with live Stata demonstrations and supervised coding labs. Participants use a realistic programme evaluation dataset to clean files, construct indicators, estimate treatment effects, test assumptions, and interpret outputs. Case discussions examine choices between experimental and quasi-experimental designs, including what can and cannot be claimed from each. Small groups review model outputs and draft management-facing findings. On the final day, participants assemble their own reproducible analysis workflow and application plan for a current or upcoming evaluation assignment.

Course outline

Day 1: Stata workflows and evaluation-ready data

• Impact evaluation questions, counterfactuals, and causal pathways
• Stata project folders, do-files, log files, and reproducibility conventions
• Importing Excel, CSV, and Stata datasets with variable and value labels
• Data inspection using codebook, describe, summarize, tabulate, and browse
• Cleaning missing values, duplicates, inconsistent identifiers, and invalid ranges
• Merging beneficiary, household, facility, and administrative datasets
• Reshaping repeated-measures data between wide and long formats

Workshop: Participants build a documented Stata project, clean and merge baseline programme files, and produce an analysis-ready dataset with a data-quality log.

Day 2: Indicators, descriptive evidence, and experimental analysis

• Translating a theory of change into outcome, exposure, and covariate variables
• Constructing indices, binary outcomes, standardized scores, and disaggregations
• Baseline balance tables and standardized mean differences
• Exploratory analysis with tabulations, distributions, and subgroup summaries
• Random assignment, intention-to-treat, and treatment-on-the-treated concepts
• Linear and logistic regression for randomized controlled trial analysis
• Clustered assignment, robust standard errors, and confidence interval interpretation

Workshop: Participants create a baseline balance table and estimate intention-to-treat effects for a randomized livelihoods-support programme.

Day 3: Quasi-experimental methods for programme data

• Selecting quasi-experimental methods based on programme rollout and available data
• Difference-in-differences design and two-way fixed-effects models
• Parallel-trends assessment using pre-intervention periods and event-study graphs
• Propensity score estimation using logit and probit models
• Nearest-neighbour matching, calipers, common support, and balance diagnostics
• Regression adjustment and inverse probability weighting concepts
• Comparing matched, weighted, and regression-adjusted treatment-effect estimates

Workshop: Participants estimate and compare difference-in-differences and propensity score matching results for a phased cash-transfer intervention.

Day 4: Diagnostics, robustness, and credible interpretation

• Sample attrition, differential attrition, and analysis-sample tracking
• Missing-data patterns and practical approaches to complete-case and imputed analysis
• Outlier detection, transformations, winsorisation, and influence diagnostics
• Fixed effects for household, community, facility, and survey-wave data
• Heterogeneous treatment effects by gender, location, poverty status, and vulnerability
• Placebo tests, alternative specifications, and sensitivity analysis
• Interpreting effect sizes, statistical uncertainty, practical significance, and limitations

Workshop: Participants conduct a robustness review of a treatment-effect model and prepare a limitations statement supported by diagnostic outputs.

Day 5: Reporting results for programme decisions

• Automating regression tables with estimates store, esttab, and estout
• Creating coefficient plots, trend graphs, and subgroup visuals in Stata
• Writing results narratives that distinguish association, attribution, and contribution
• Applying OECD DAC Evaluation Criteria to interpretation of effectiveness findings
• Documenting analytical decisions, assumptions, code, and dataset versions
• Reviewing impact evaluation reports for methodological and reporting quality
• Developing an evaluation analysis plan for a live programme question

Workshop: Participants produce a reproducible Stata results pack and a two-page impact findings brief for a programme-management audience.

Tools & standards covered

Stata 18, Microsoft Excel, GitHub, OECD DAC Evaluation Criteria

A typical training day

Live online deliveries follow the same structure in the East Africa Time zone, with shorter screen blocks and longer breaks.

What the fee includes

How you can take this course

Certification

Participants who complete the full five days receive the Skillset Development Certificate of Completion, stating the course title, course code, dates and delivery format — suitable for professional-development records and employer reimbursement.

Frequently asked questions