Module 1: STATA Programming and Automation (Do-Files)

• • Mastering Do-file structure for reproducibility and project management
  • Using local and global macros for dynamic command creation
  • Implementing loops (foreach, forvalues, while) for repetitive tasks
  • Creating and calling custom programs (program define) for complex routines
  • Advanced debugging techniques and flow control (capture, if)
  • Practical session: Automating the data cleaning and descriptive statistics generation for a large dataset using macros and loops within a master Do-file.

Module 2: Advanced Data Management and Manipulation

• • Restructuring data between long and wide formats (reshape)
  • Merging and appending datasets (merge, append) with one-to-one, one-to-many, and many-to-many links
  • Handling missing data efficiently using mvdecode, egen, and imputation basics
  • Working with date and time variables and time series operators (L., F., D.)
  • Generating complex variables using extended functions (egen) and conditional logic
  • Practical session: Converting a cross-sectional dataset of repeated measurements into a panel structure, filling in missing values using imputation, and verifying unique identifiers.

Module 3: Introduction to Linear Regression and Diagnostics

• • Review of the Ordinary Least Squares (OLS) assumptions and interpretation
  • Post-estimation commands for coefficient testing (test, lincom) and marginal effects
  • Detection and impact of multicollinearity (VIF) and outliers (predict with influence statistics)
  • Plotting regression results using margins and marginsplot
  • Introduction to interaction effects and interpreting interacted coefficients
  • Practical session: Running a multivariate OLS regression, performing all standard diagnostic checks, and visualizing the effect of a key interaction term.

Module 4: Advanced OLS and Heteroscedasticity/Autocorrelation

• • Correcting standard errors for heteroscedasticity using robust standard errors (vce(robust))
  • Addressing autocorrelation in time series/panel data using clustered standard errors (vce(cluster))
  • Implementing Generalized Least Squares (GLS) for efficiency when residuals are non-spherical
  • Modeling non-linear relationships using splines, polynomials, and logarithmic transformations
  • Utilizing bootstrapping for robust standard error estimation in small samples
  • Practical session: Running a cross-sectional regression, testing for heteroscedasticity (e.g., Breusch-Pagan test), and comparing standard, robust, and bootstrapped standard errors.

Module 5: Instrumental Variables (IV) and Two-Stage Least Squares (2SLS)

• • Understanding the problem of endogeneity and conditions for IV validity (relevance and exclusion)
  • Implementing 2SLS using the ivregress command
  • Conducting weak instrument tests (e.g., Cragg-Donald F-statistic, Kleibergen-Paap rk Wald F-statistic)
  • Overidentification tests (Sargan-Hansen test) to check instrument validity
  • Handling multiple endogenous variables and multiple instruments
  • Practical session: Applying 2SLS to an endogenous model using a provided dataset, performing the necessary instrument strength and overidentification tests, and comparing results to OLS.

Module 6: Panel Data I: Pooled OLS, Fixed Effects (FE), and Random Effects (RE)

• • Setting up panel data using the xtset command and understanding data structure requirements
  • Implementing Pooled OLS and discussing its limitations
  • Estimating Fixed Effects (FE) models (xtreg, fe) to control for unobserved heterogeneity
  • Estimating Random Effects (RE) models (xtreg, re) and its assumptions
  • Conducting the Hausman test to select between FE and RE models
  • Practical session: Analyzing a panel dataset, comparing the results of Pooled OLS, FE, and RE models, and using the Hausman test to justify the final choice of model.

Module 7: Panel Data II: Dynamic Panel Data (GMM)

• • Introduction to dynamic panel models and the issue of lagged dependent variables
  • Understanding the Bias issue in FE for dynamic models (Nickell bias)
  • Implementing the Arellano-Bond Difference GMM (D-GMM) estimator
  • Implementing the Arellano-Bover/Blundell-Bond System GMM (S-GMM) estimator
  • Performing crucial GMM diagnostic tests (Sargan/Hansen test and Arellano-Bond autocorrelation tests)
  • Practical session: Estimating a dynamic model of economic growth using System GMM and ensuring the validity of the instruments through diagnostic testing.

Module 8: Limited Dependent Variables I: Binary Outcomes (Logit/Probit)

• • Theoretical foundation for modeling binary outcomes (probability vs. odds)
  • Running and interpreting the Logit and Probit models (logit, probit)
  • Calculating and interpreting marginal effects for easier policy implications (margins, mfx)
  • Assessing model fit using sensitivity, specificity, and ROC curves
  • Using Predicted Probabilities and thresholds for classification
  • Practical session: Modeling the likelihood of a binary event (e.g., loan default) using Logit, interpreting the coefficients, and calculating marginal effects for key predictors.

Module 9: Limited Dependent Variables II: Ordered and Multinomial Outcomes

• • Modeling categorical outcomes with inherent ordering using Ordered Logit/Probit (ologit, oprobit)
  • Interpreting the threshold parameters and proportional odds assumption
  • Modeling categorical outcomes without inherent ordering using Multinomial Logit (mlogit)
  • Calculating and interpreting the predicted probabilities for each outcome category
  • Implementing Marginal Effects for Ordered and Multinomial models
  • Practical session: Analyzing survey data on satisfaction levels (low, medium, high) using Ordered Logit and interpreting the effect of income on the probability of reaching each level.

Module 10: Time Series Analysis I: Stationarity and ARIMA Models

• • Defining and testing for stationarity using graphical methods and the Augmented Dickey-Fuller (ADF) test
  • Applying difference operators to achieve stationarity
  • Identifying appropriate ARIMA model parameters (p, d, q) using ACF and PACF plots
  • Estimating ARIMA and Seasonal ARIMA (SARIMA) models in STATA
  • Forecasting future values and confidence intervals using predict
  • Practical session: Analyzing a time series dataset (e.g., inflation rate), testing for stationarity, fitting an appropriate ARIMA model, and generating a forecast for the next 12 periods.

Module 11: Time Series Analysis II: Cointegration and VAR Models

• • Introduction to Vector Autoregression (VAR) models for multiple time series
  • Testing for Granger Causality within a VAR framework
  • Impulse Response Functions (IRF) and Variance Decompositions (VD) for system analysis
  • Understanding Cointegration and the Johansen test for long-run relationships
  • Implementing Vector Error Correction Models (VECM) for cointegrated series
  • Practical session: Running a VAR model on two related macroeconomic variables, testing for Granger causality, and interpreting the Impulse Response Functions.

Module 12: Survival and Duration Analysis (Cox Proportional Hazards)

• • Introduction to survival data: censoring, truncation, and the hazard function
  • Setting up survival data using the stset command
  • Non-parametric estimation of survival curves (Kaplan-Meier estimator)
  • Implementing the Cox Proportional Hazards model (stcox)
  • Testing the Proportional Hazards assumption and utilizing stratified models
  • Practical session: Analyzing a duration dataset (e.g., time to event), running a Kaplan-Meier estimate, and fitting a Cox Proportional Hazards model with covariates.

Module 13: Complex Survey Data Analysis

• • Understanding complex sampling designs (stratification, clustering, weights)
  • Defining the survey design using the svyset command
  • Conducting descriptive statistics and regression analysis using the svy prefix
  • Estimating standard errors and confidence intervals appropriate for complex designs
  • Utilizing svy commands for subpopulation analysis
  • Practical session: Analyzing a provided complex survey dataset (e.g., a household survey), setting the design parameters, and running a weighted OLS regression using the svy prefix.

Module 14: Non-Parametric and Robust Estimation Techniques

• • Introduction to non-parametric tests (Mann-Whitney, Kruskal-Wallis) for non-normal data
  • Implementing Kernel Density Estimation for distribution visualization
  • Utilizing the qreg command for Quantile Regression
  • Interpreting results across different quantiles (e.g., median vs. 90th percentile)
  • Comparing results of OLS and Quantile Regression
  • Practical session: Running a Quantile Regression on income data, comparing the effect of education on income at the 10th, 50th, and 90th percentiles, and visualizing the results.

    ---

Module 15: Causal Inference and Matching Methods

• • Introduction to the Potential Outcomes Framework and Average Treatment Effect (ATE)
  • Understanding Selection on Observables and Propensity Score Matching (PSM)
  • Implementing PSM using the psmatch2 or similar user-written command
  • Conducting balance checks and common support diagnostics
  • Interpreting the estimated Average Treatment Effect on the Treated (ATT)
  • Practical session: Estimating the causal effect of a policy intervention using Propensity Score Matching, checking covariate balance before and after matching.

Module 16: Data Visualization and Custom Graphs (GPL)

• • Mastering the STATA Graphing Programming Language (GPL) for custom plots
  • Creating highly customized scatter plots, bar charts, and box plots
  • Combining multiple plots into a single figure (graph combine)
  • Utilizing color schemes, font styles, and legends for professional presentation
  • Exporting graphs in publication-ready formats (e.g., EPS, TIFF)
  • Practical session: Replicating a complex visualization from a published paper, combining a scatter plot with marginal distributions into a single, highly customized figure.

Module 17: Advanced Output Formatting

• • Utilizing the estout and outreg2 user-written commands for professional regression tables
  • Customizing the statistics displayed (standard errors, t-stats, p-values, N)
  • Exporting formatted tables directly to LaTeX, Word (.rtf), and Excel
  • Generating summary statistics tables with group comparisons (tabstat, table)
  • Creating customized table headers and footnotes for clear communication
  • Practical session: Generating a publication-ready regression table comparing the results of three different models (OLS, IV, FE) and exporting it to a Word document.

Module 18: Simulation and Bootstrapping Techniques

• • • Introduction to Monte Carlo simulations for demonstrating statistical properties
    • Writing simulation programs using STATA's programming features
    • Applying the bootstrap prefix to estimate standard errors for complex statistics
    • Utilizing the simulate prefix for estimating sampling distributions of estimators
    • Understanding the difference between bootstrapping standard errors and confidence intervals
    • Practical session: Using the bootstrap prefix to estimate the standard error and confidence interval for a complex, non-standard statistic derived from a two-stage process.

📌 Participant Requirements

• Participants must have a good command of English.
• Applicants must meet the admission criteria set by Dataex Global Institute.

📄 Terms & Conditions

🎁 Group Discount

Organizations that sponsor four (4) participants will receive a complimentary slot for a fifth participant.

💼 What Course Fees Cover

The training fee includes:

• Comprehensive learning materials
• Daily lunches, teas, and snacks
• Certificate of Participation upon successful completion

Note: Participants are responsible for their own:

• Travel and accommodation
• Visa application and insurance
• Personal expenses

🎓 Certification

All participants will receive a Certificate of Participation at the end of the training.

🔄 Course Content

The program outline provided is for guidance only. Dataex Global Institute reserves the right to update course content as part of our continuous improvement process.

✅ NITA Accreditation

Our programs are approved by the National Industrial Training Authority (NITA). Organizations may claim reimbursement in accordance with NITA regulations.

📝 How to Register

To secure your place:

• Email the Training Officer at training@dataexglobalinstitute.com to request a registration form.
• Or call us directly at +254 725 012 095.

Early booking is encouraged as spaces are limited.

💳 Payment Options

Please make payment at least 5 days before the training start date. Choose from the following options:

1. Cheque Payments (for groups of 5 or more): Payable to
   Dataex Global Training & Development Center Limited
2. Invoice: We can issue an invoice to you or your organization.
3. Bank Deposit: Account details will be provided upon request.

❌ Cancellation Policy

• A non-refundable registration fee of 15% is included in the course fee.
• Cancellations made 14 days or more before the training start date are eligible for a refund (excluding the registration fee).
• No refunds will be issued for cancellations made within 14 days of the training. However, participants may:
• Transfer their registration to a future session, or
• Nominate a substitute participant (subject to eligibility).

🛠️ Tailor-Made Training

Need a customized session for your team? We offer bespoke training for groups of 5 or more, delivered:

• At our Training Centre, or
• At a location of your choice

To request a tailored program, contact us at: 📞 +254 725 012 095
📧 training@dataexglobalinstitute.com

🏨 Accommodation & Airport Transfers

We can assist with accommodation and airport transfers upon request, at an additional cost.
For arrangements, please contact the Training Officer.