M&E Data Analysis with SPSS and Stata Training Course

Participants can organise M&E datasets and analysis records for secure, auditable use.

• Unique identifiers preserve respondent linkage while protecting personally identifiable information
• File naming conventions separate raw, processed, output and archival materials
• Version control records dataset revisions, syntax changes and analytical decisions
• Informed consent restrictions shape permitted analysis, sharing and disclosure decisions
• Disclosure control suppresses small cells and protects sensitive subgroup findings
• Reproducible workflows retain syntax, logs, source files and output records

Participants can configure SPSS files and import common M&E data sources accurately.

• SPSS Data View and Variable View organise cases and variables
• Variable labels, value labels and missing codes improve output readability
• Importing Excel, CSV and text files preserves field definitions
• Measurement levels in SPSS guide menus, charts and statistical procedures
• Codebooks checked against imported fields identify structural import errors
• SPSS syntax records import settings and variable-definition commands

Participants can establish a documented Stata project and import M&E datasets.

• Stata memory, variables, observations and storage types determine dataset handling
• Project folders organise raw data, do-files, logs, outputs and documentation
• Import excel and import delimited load common monitoring datasets
• Describe, codebook and list commands inspect dataset structure quickly
• Labels and notes document variables, categories and analytical qualifications
• Do-files preserve commands for repeatable data preparation and analysis

Participants can identify, correct and document common errors in M&E datasets.

• Range checks identify impossible ages, dates, quantities and household sizes
• Frequency tables expose invalid codes, duplicate categories and unexpected values
• Missing-data patterns distinguish nonresponse, skip logic and data-entry failures
• Duplicate identifier checks detect repeated interviews and registration records
• Logic checks compare related variables against questionnaire routing rules
• Cleaning logs document corrections, exclusions, assumptions and unresolved anomalies

Participants can create analysis variables and combine M&E files using transparent rules.

• Recode commands group categories while retaining documented original responses
• Computed variables derive rates, indexes, scores and binary outcome measures
• Date functions calculate respondent age, follow-up intervals and reporting periods
• SPSS MATCH FILES merges cases using validated unique identifiers
• Stata merge and append combine longitudinal, household and service datasets
• Post-merge checks confirm matched records, unmatched cases and duplicate keys

Participants can summarise programme data using appropriate descriptive measures and tables.

• Frequencies and percentages report categorical service-use and coverage indicators
• Means, medians and standard deviations summarise continuous outcome variables
• Minimum, maximum and percentile values identify distributional anomalies
• Rates and ratios use correct denominators for programme performance reporting
• Confidence intervals express uncertainty around sample-based indicator estimates
• Table layouts distinguish counts, percentages, denominators and missing observations

Participants can produce clear visual and tabular outputs for M&E reporting.

• Bar charts compare categorical indicators across sites and participant groups
• Line charts track monthly outputs, seasonal patterns and target progress
• Histograms and boxplots reveal skewness, outliers and distribution shape
• Cross-tabulation tables show subgroup differences with stated denominators
• Chart titles state indicator definitions, periods, populations and data sources
• Output formatting removes unnecessary decimals and clarifies statistical annotations

Participants can calculate and verify results-framework indicators from source data.

• Numerator and denominator rules prevent misreporting of coverage indicators
• Indicator calculation sheets trace source variables through final reported values
• Disaggregation rules define mutually exclusive and collectively exhaustive categories
• Target comparisons distinguish annual targets, cumulative targets and planned trajectories
• Output indicators separated from outcome indicators avoid attribution overclaims
• OECD DAC criteria inform reporting questions on relevance and effectiveness

Participants can assess how sample design affects M&E estimates and conclusions.

• Simple random, systematic, stratified and cluster sampling affect precision
• Sampling frames define eligible populations and potential coverage exclusions
• Design effects explain wider uncertainty under clustered survey designs
• Survey weights correct unequal selection probabilities and differential response
• Sample-size logic links expected proportions, precision and confidence levels
• Weighted estimates require documented assumptions about population representation

Participants can assess survey quality risks before drawing M&E conclusions.

• Completion rates distinguish completed, partial, refused and ineligible interviews
• Item nonresponse identifies questions requiring cautious interpretation or follow-up
• Enumerator comparisons detect unusual interview duration and response patterns
• Heaping patterns reveal rounded ages, expenditures and recall-period responses
• Outlier review distinguishes valid extreme observations from recording errors
• Quality dashboards combine timeliness, completeness, consistency and accuracy measures

Participants can test and interpret associations between categorical M&E variables.

• Contingency tables compare outcomes by sex, location and intervention exposure
• Chi-square tests assess associations between categorical variables and groups
• Expected cell counts determine chi-square test suitability and reporting caveats
• Fisher's exact test handles sparse tables with small expected frequencies
• Risk differences and risk ratios express practical subgroup comparisons
• Standardised residuals identify cells contributing most to chi-square findings

Participants can compare continuous and ordinal outcomes using suitable statistical tests.

• Independent-samples t-tests compare average outcomes between two participant groups
• Paired t-tests assess repeated measures for the same respondents
• One-way ANOVA compares mean outcomes across three or more groups
• Levene's test checks variance assumptions before standard mean comparisons
• Mann-Whitney and Kruskal-Wallis tests address ordinal or non-normal outcomes
• Effect sizes report practical magnitude alongside statistical significance values

Participants can estimate and interpret linear regression models for continuous outcomes.

• Pearson and Spearman correlations assess linear and ranked associations
• Scatterplots examine form, outliers and plausible relationships before modelling
• Ordinary least squares regression estimates adjusted associations with continuous outcomes
• Regression coefficients distinguish expected change from causal effect claims
• Residual plots assess linearity, constant variance and influential observations
• Multicollinearity diagnostics identify strongly overlapping explanatory variables

Participants can model binary programme outcomes and communicate results accurately.

• Binary outcomes include service uptake, school attendance and recovery status
• Logistic regression estimates odds of outcomes from participant characteristics
• Odds ratios require reference categories and confidence intervals for interpretation
• Predicted probabilities communicate model findings to non-technical decision-makers
• Classification tables assess sensitivity, specificity and model prediction patterns
• Model specification distinguishes confounders, mediators and plausible programme pathways

Participants can analyse baseline-endline data while stating design limitations clearly.

• Baseline profiles establish pre-intervention values, comparability and priority gaps
• Endline estimates compare indicator changes against stated baseline measures
• Repeated cross-sections differ from panel data with matched respondents
• Paired analysis uses respondent identifiers to assess individual-level change
• Attrition analysis examines whether follow-up losses distort observed results
• Recall periods and seasonal timing affect baseline-endline interpretation

Participants can assess difference-in-differences outputs and their core assumptions.

• Difference-in-differences compares outcome changes between intervention and comparison groups
• Parallel trends assumption requires credible pre-intervention outcome evidence
• Interaction terms estimate the programme-associated change in regression models
• Time and group fixed effects account for common measured differences
• Graphical trend checks reveal possible departures from parallel trajectories
• Interpretation separates design-supported estimates from unsupported attribution claims

Participants can appraise matching approaches and report their limitations for M&E studies.

• Propensity scores estimate participation probabilities from observed baseline characteristics
• Nearest-neighbour matching identifies comparable non-participants using defined distance rules
• Balance diagnostics compare covariate distributions before and after matching
• Common support excludes cases lacking plausible comparison-group counterparts
• Matching addresses observed differences but not unmeasured selection factors
• Sensitivity notes state data limitations, assumptions and residual bias risks

Participants can analyse equity-relevant differences without overstating small subgroup findings.

• Disaggregation plans specify sex, age, disability, geography and displacement status
• Washington Group Short Set variables support disability-disaggregated reporting
• Intersectional tables examine combined characteristics without masking subgroup disparities
• Small-cell suppression protects respondents and prevents unstable percentage reporting
• Absolute and relative gaps compare equitable access and outcome patterns
• Interpretation considers barriers, service availability and differential exposure contexts

Participants can present M&E findings with transparent methods, limitations and recommendations.

• Findings statements answer evaluation questions before presenting supporting statistical detail
• Methods sections document datasets, samples, variables, procedures and assumptions
• Confidence intervals and p-values reported without treating significance as importance
• Limitations distinguish measurement error, bias, uncertainty and design constraints
• Recommendations link findings to feasible programme decisions and accountable owners
• Peer review checklists test traceability from outputs to stated conclusions

Participants can build, review and present a complete reproducible M&E analysis package for a programme decision.

• Select evaluation questions, indicators and disaggregation priorities from a programme case
• Import and clean source files using documented SPSS syntax or Stata do-files
• Construct derived indicators, merge required datasets and complete validation checks
• Produce descriptive tables, charts and statistical tests answering priority questions
• Review assumptions, limitations, disclosure risks and interpretation against quality criteria
• Present findings, recommendations and reproducible documentation for management review