Course Introduction
Structural Equation Modelling for Programme Theory Testing is an advanced professional training course designed to strengthen participants’ ability to evaluate the conceptual and causal logic underpinning development programmes, public policies and social interventions. It combines measurement theory, multivariate statistics and theory-based evaluation to help practitioners investigate whether empirical evidence is consistent with the relationships anticipated in a programme’s theory of change. Participants learn to translate broad programme assumptions into clearly specified, measurable and testable models that support more transparent evaluation conclusions.
Programme theories often propose that resources and activities produce outputs, which influence intermediate outcomes and eventually contribute to sustainable change. However, these relationships may involve multiple pathways, unobserved concepts, contextual influences and competing explanations. Structural equation modelling provides a framework for examining several interrelated relationships simultaneously while distinguishing measurement questions from hypothesised relationships among constructs. The course shows participants how to develop analytical models that reflect programme logic without treating a conceptual diagram as proof that the proposed mechanisms are correct.
A central component of the course is the measurement of latent constructs such as empowerment, institutional capacity, service quality, social inclusion, resilience and beneficiary confidence. These concepts cannot always be represented adequately by a single indicator. Participants learn how to develop measurement models, assess indicator reliability, evaluate construct validity and determine whether observed indicators represent the intended concepts. They also examine how weak measurement can distort conclusions about programme performance, why different constructs must be distinguished empirically, and how to improve survey instruments and monitoring frameworks before undertaking structural analysis.
The programme develops practical competence in testing direct, indirect and mediated relationships within programme theories. Participants learn to assess whether proposed intermediate outcomes help explain associations between intervention exposure and longer-term outcomes, while recognising that statistical mediation does not automatically establish a causal mechanism. They explore confirmatory factor analysis, path analysis, structural regression, model identification, parameter estimation, model fit and competing model comparison. Through applied exercises, participants learn to formulate hypotheses, assess uncertainty, interpret standardised and unstandardised estimates, and communicate findings in terms that are useful to evaluation teams and programme decision-makers.
Participants also learn how to address common analytical challenges in development evaluation, including missing observations, non-normal data, categorical indicators, small samples, measurement error and differences between beneficiary groups. The course introduces appropriate diagnostic procedures and alternative modelling strategies, helping participants understand when conventional maximum-likelihood estimation may be unsuitable and when robust estimators, bootstrap confidence intervals or categorical-data methods may be appropriate. Attention is given to sample-size planning, assumptions, model complexity and sensitivity analysis so that analytical decisions are guided by the research question, data characteristics and substantive programme theory rather than by software defaults.
The final emphasis is on translating statistical results into defensible programme learning and practical recommendations. Participants learn to compare empirical findings with the intended theory of change, identify supported and unsupported pathways, investigate rival explanations and communicate limitations without overstating what the evidence demonstrates. Using applied development scenarios, they prepare model specifications, interpret analytical outputs and develop recommendations for programme redesign, monitoring improvement and further research. The course equips evaluation professionals to use SEM as one component of a broader evidence strategy, complemented where appropriate by experimental, quasi-experimental, qualitative and implementation evidence.
Duration
10 days
Who Should Attend
-
Monitoring, Evaluation and Learning (MEL) specialists responsible for evaluating complex programmes and testing theories of change.
-
Development evaluation consultants undertaking advanced quantitative analysis and evidence synthesis assignments.
-
Research officers and social scientists working with multivariate data, survey instruments and programme outcome indicators.
-
Programme managers seeking stronger evidence about the pathways connecting activities, intermediate outcomes and intended impact.
-
Government policy analysts assessing the performance and underlying assumptions of public programmes and social interventions.
-
Staff of international development agencies and donor organisations reviewing evaluation designs, analytical methods and causal claims.
-
Public health evaluators investigating behavioural, institutional and service-delivery mechanisms associated with health interventions.
-
Education, livelihoods and social protection researchers studying latent constructs and interconnected programme outcomes.
-
Economists and statisticians seeking to integrate SEM into programme evaluation and applied social research.
-
Survey methodologists and questionnaire designers responsible for developing reliable and valid measurement instruments.
-
Academic researchers, postgraduate students and research supervisors conducting theory-driven quantitative studies.
-
Non-governmental organisation monitoring and evaluation teams responsible for generating credible evidence for programme improvement.
-
Institutional learning specialists examining organisational capacity, implementation quality and beneficiary-level change.
-
Evaluation commissioners and technical advisers who need to interpret SEM findings and assess their methodological quality.
Course Objectives
-
Explain the conceptual foundations of structural equation modelling and distinguish its principal applications from regression, path analysis and other multivariate techniques used in programme evaluation.
-
Translate a programme theory or theory of change into explicit, testable hypotheses that define the expected relationships among intervention inputs, outputs, mediators and outcomes.
-
Develop conceptual diagrams and formal model specifications that distinguish latent constructs, observed indicators, measurement relationships, structural pathways and residual associations.
-
Design and assess measurement models using confirmatory factor analysis, evaluating indicator performance, factor loadings, construct reliability and convergent and discriminant validity.
-
Determine whether a proposed structural equation model is identified and estimable, and recognise the consequences of inadequate sample size, excessive complexity and poorly specified parameters.
-
Select appropriate estimators and analytical strategies according to variable measurement levels, distributional characteristics, missing-data patterns and the substantive objectives of the evaluation.
-
Estimate and interpret direct, indirect and total associations, including hypothesised mediation pathways, while distinguishing statistical evidence from stronger claims about causal mechanisms.
-
Evaluate overall and local model fit using multiple diagnostic measures, residual analysis, parameter estimates and theoretically informed assessment rather than reliance on a single threshold.
-
Apply procedures for handling missing observations, non-normality, categorical indicators and potential measurement problems while documenting the assumptions and limitations of each approach.
-
Compare alternative theoretical specifications, assess measurement invariance across relevant groups and examine whether proposed relationships differ across populations or implementation contexts.
-
Use appropriate statistical software to prepare data, specify models, estimate parameters, inspect diagnostics and produce reproducible analytical outputs for programme evaluation.
-
Communicate findings through clear diagrams, accessible reports and evidence-based recommendations that identify supported pathways, unresolved assumptions and priorities for programme improvement.
Comprehensive Course Outline
Module 1: Foundations of Structural Equation Modelling in Programme Evaluation
-
Explain the core principles of SEM and its role in examining interconnected relationships among programme constructs, intermediate outcomes and longer-term results.
-
Distinguish observed variables, latent variables, measurement errors, structural disturbances and residual covariances using practical evaluation examples.
-
Compare SEM with multiple regression, path analysis, factor analysis and other multivariate approaches, identifying the questions each method can address.
-
Examine the strengths and limitations of SEM for theory testing, including the need for appropriate research designs and cautious interpretation of model-based evidence.
Module 2: Programme Theory, Logic Models and Theory of Change
-
Deconstruct programme documentation into inputs, activities, outputs, intermediate outcomes, final outcomes, contextual conditions and long-term intended impacts.
-
Identify the assumptions linking programme components and translate broad narrative statements into specific relationships that can be evaluated empirically.
-
Distinguish theories of action from theories of change and identify how implementation failures may differ from failures in the underlying causal explanation.
-
Develop a theory-testing matrix linking evaluation questions, hypothesised pathways, measurable constructs, indicators, data sources and expected empirical patterns.
Module 3: Translating Programme Theory into SEM Conceptual Models
-
Construct path diagrams that represent hypothesised direct effects, indirect pathways, correlated constructs and theoretically justified contextual influences.
-
Define directional hypotheses and identify the expected signs, relative importance and practical interpretation of relationships proposed by the programme theory.
-
Separate confirmatory hypotheses developed before analysis from exploratory relationships identified during data investigation to reduce unsupported model modification.
-
Develop rival conceptual models that represent plausible alternative explanations and establish transparent criteria for evaluating their relative empirical support.
Module 4: Measurement Theory and Latent Construct Development
-
Define latent constructs such as empowerment, trust, institutional capacity and resilience, and determine whether each requires multiple indicators for adequate representation.
-
Design indicator sets that reflect the conceptual domain of each construct while avoiding excessive overlap, ambiguous wording and redundant survey items.
-
Assess content validity, face validity and the theoretical relevance of indicators using existing evidence, expert review and consultation with programme stakeholders.
-
Investigate response scales, reverse-coded items, measurement error and data-quality problems that may undermine the validity of subsequent structural relationships.
Module 5: Confirmatory Factor Analysis and Construct Validation
-
Specify confirmatory factor analysis models that assign observed indicators to theoretically defined latent constructs and evaluate the plausibility of those assignments.
-
Interpret factor loadings, indicator residuals and latent-factor relationships to identify weak indicators and possible problems with construct specification.
-
Evaluate construct reliability and convergent and discriminant validity using suitable evidence rather than relying exclusively on any single numerical criterion.
-
Examine competing factor structures and document whether measurement findings support, qualify or challenge the conceptual definitions used in the programme theory.
Module 6: Model Identification, Specification and Estimation
-
Explain the information requirements for model identification and assess whether proposed parameters can be estimated uniquely from the available data.
-
Distinguish underidentified, just-identified and overidentified models and explain how identification affects the possibility of evaluating global model fit.
-
Select plausible estimation approaches based on the observed variables, sample characteristics, distributional assumptions and complexity of the proposed model.
-
Diagnose improper solutions, nonconvergence, implausible parameter estimates and unstable standard errors before interpreting substantive evaluation findings.
Module 7: Structural Pathways and Hypothesis Testing
-
Specify structural relationships among latent and observed variables to represent proposed pathways from programme exposure through intermediate outcomes to final results.
-
Interpret regression coefficients, standardised estimates, standard errors, confidence intervals and statistical tests in relation to prespecified hypotheses.
-
Assess direct, indirect and total effects while distinguishing statistical significance, effect magnitude, practical importance and uncertainty in programme-relevant terms.
-
Evaluate whether the estimated relationships align with the expected direction and structure of programme theory and identify pathways requiring further investigation.
Module 8: Mediation, Indirect Effects and Mechanism Testing
-
Develop mediation models to assess whether proposed intermediate constructs statistically account for part of the relationship between programme exposure and downstream outcomes.
-
Estimate indirect effects using appropriate methods and examine confidence intervals, including bootstrap procedures when suitable for the data and model.
-
Distinguish partial, indirect-only and other observed mediation patterns without assuming that statistical mediation proves a real-world causal mechanism.
-
Consider temporal ordering, confounding, competing mechanisms and alternative explanations when interpreting evidence about how an intervention may contribute to change.
Module 9: Model Fit, Residual Diagnostics and Model Comparison
-
Interpret common fit measures, including the chi-square test, comparative fit index, Tucker–Lewis index, root mean square error of approximation and standardised root mean square residual.
-
Examine model residuals, local areas of misfit and parameter estimates to understand which relationships or measurement components may be poorly represented.
-
Compare theoretically defensible alternative models using appropriate fit information, information criteria where applicable and substantive interpretability.
-
Use modification indices and diagnostic suggestions cautiously, requiring substantive justification and suitable validation before adopting model changes.
Module 10: Data Preparation, Sample Size and Missing-Data Management
-
Audit datasets for coding inconsistencies, outliers, duplicate records, impossible values and indicator distributions that could affect model estimation.
-
Assess sample-size adequacy in relation to model complexity, expected effect sizes, estimator choice, missingness and the precision required for key parameters.
-
Examine missing-data mechanisms and select defensible handling strategies, such as full-information maximum likelihood or multiple imputation, where assumptions permit.
-
Evaluate non-normality, influential observations and data limitations, documenting their implications for estimation, uncertainty and the generalisability of findings.
Module 11: Categorical Indicators, Robust Estimation and Advanced Data Issues
-
Distinguish continuous, binary, ordinal and nominal variables and determine how their measurement properties affect model specification and estimation.
-
Examine suitable categorical-data methods and estimators, including robust weighted least squares approaches where appropriate for ordinal or binary indicators.
-
Investigate ceiling effects, sparse response categories, skewed distributions and other characteristics that may compromise parameter estimates or standard errors.
-
Conduct sensitivity checks across plausible analytical choices and explain how estimator selection influences the interpretation and robustness of programme theory tests.
Module 12: Multiple-Group SEM and Measurement Invariance
-
Assess whether measurement instruments represent the same constructs across demographic groups, geographic settings, beneficiary categories or implementation sites.
-
Examine configural, metric and scalar measurement invariance where relevant, understanding the conditions required for meaningful comparisons across groups.
-
Compare structural pathways across groups when the measurement model supports such comparisons, identifying possible differences in how programme mechanisms operate.
-
Interpret group differences in context and investigate whether apparent variation reflects genuine programme heterogeneity, measurement differences or data limitations.
Module 13: Longitudinal SEM and Change Over Time
-
Design longitudinal measurement strategies that track constructs and outcomes across baseline, implementation, follow-up and later assessment periods.
-
Assess longitudinal measurement invariance and examine whether indicators retain comparable meaning across waves before interpreting changes in latent constructs.
-
Explore autoregressive, cross-lagged and latent growth approaches where justified by the research question, study design and available repeated observations.
-
Evaluate attrition, timing, reciprocal relationships and unmeasured confounding when interpreting longitudinal evidence about programme pathways and changing outcomes.
Module 14: Practical SEM Implementation Using R and lavaan
-
Prepare a reproducible analytical environment and organise programme evaluation data, variable definitions and construct documentation for SEM estimation.
-
Specify confirmatory factor analysis, path analysis, mediation and structural regression models using readable and well-documented lavaan syntax.
-
Run estimation procedures, inspect model summaries and diagnostics, extract parameter estimates and generate interpretable outputs for evaluation reporting.
-
Maintain reproducible scripts, document analytical decisions, manage model versions and create an auditable workflow that allows reviewers to verify the reported findings.
Module 15: Integrating SEM Findings with Broader Evaluation Evidence
-
Map estimated relationships back to the original theory of change and identify which proposed pathways receive support, weak support or contradictory evidence.
-
Integrate SEM findings with qualitative interviews, implementation records, administrative data and other evidence to assess plausible mechanisms and contextual explanations.
-
Distinguish implementation weaknesses, measurement limitations and theory weaknesses when results differ from programme expectations or remain inconclusive.
-
Develop transparent triangulation procedures that document convergence, disagreement and evidence gaps without forcing different methods to produce identical conclusions.
Module 16: Reporting, Decision-Making and Applied Capstone
-
Prepare a complete SEM evaluation report covering the conceptual framework, hypotheses, sample, measurement model, estimator, fit assessment and structural results.
-
Present path diagrams, effect estimates, uncertainty intervals and limitations in formats that are accessible to technical reviewers and programme decision-makers.
-
Translate findings into recommendations for programme redesign, improved monitoring indicators, targeted implementation adjustments and additional research where uncertainty remains.
-
Complete an applied capstone by testing a programme theory, comparing alternative explanations, defending analytical decisions and presenting evidence-based conclusions to a review panel.
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.








