DevData Practice Guide
What Is DevData Practice?
DevData Practice is ImpactMojo's synthetic dataset generator — a tool that creates realistic, ready-to-use practice datasets modelled on real development surveys. It produces 840,000+ rows of data across 36 generators covering the kinds of data development professionals work with every day.
If you've ever needed practice data for a training session, a classroom exercise, or testing an analytical approach — but couldn't use real data because it's sensitive, restricted, or messy in ways that distract from learning — DevData Practice solves that problem.
Access: Professional tier (₹999/month) — Open DevData Practice
What Is Synthetic Data? (And Why Does It Matter?)
Synthetic data is data that has been artificially generated to resemble real data — it has the same structure, patterns, and statistical properties as actual survey data, but it doesn't contain any real individuals' information.
Why not just use real data for training?
Privacy and ethics — real household survey data contains sensitive information about real people
Synthetic data contains no real individuals — no consent issues, no anonymisation needed
Access restrictions — many important datasets require institutional agreements or IRB approval to use
DevData Practice datasets are freely available for anyone to use
Messy and distracting — real data has missing values, coding errors, and inconsistencies that confuse learners
Synthetic datasets are clean and consistent, so learners can focus on methods
Difficult to find exactly what you need — real datasets may not have the right variables for your exercise
You can generate datasets tailored to specific topics and methods
When should you use real data instead?
Synthetic data is for learning and practice. When you're doing actual research, programme evaluation, or policy analysis, you should use real data. DevData Practice is a training tool, not a substitute for genuine evidence.
What Datasets Can You Generate?
DevData Practice includes 36 generators organised into 10 categories:
1. Household Surveys & Consumption
Datasets modelled on structures like India's National Sample Survey (NSS) and household consumption expenditure surveys. Variables include household size, income, expenditure categories, asset ownership, and poverty indicators.
Use for: Teaching poverty measurement, consumption analysis, welfare economics.
2. RCT Experimental Data
Treatment-control datasets with baseline and endline measurements, attrition patterns, and treatment effects. Modelled on the structure of real randomised controlled trials.
Use for: Teaching impact evaluation, difference-in-differences, intention-to-treat analysis.
3. Health & Nutrition
Datasets similar to the Demographic and Health Survey (DHS) and India's National Family Health Survey (NFHS). Variables include maternal health indicators, child nutrition (stunting, wasting, underweight), immunisation coverage, and healthcare access.
Use for: Teaching health data analysis, nutrition assessment, public health programme evaluation.
4. Education & Learning Outcomes
Datasets modelled on education surveys like ASER (Annual Status of Education Report). Variables include learning levels, school attendance, teacher characteristics, and education expenditure.
Use for: Teaching education data analysis, learning outcome measurement, school effectiveness research.
5. Agricultural & Livelihood
Datasets covering agricultural production, landholding, crop patterns, input use, and market access — modelled on agricultural census and livelihood survey structures.
Use for: Teaching agricultural economics, livelihood analysis, rural development programme evaluation.
6. Gender & GBV Indicators
Datasets with gender-disaggregated variables including women's decision-making autonomy, time use, labour force participation, and gender-based violence indicators.
Use for: Teaching gender analysis, Women's Economic Empowerment (WEE) measurement, GESI audits.
7. Climate & Environmental
Datasets with temperature, rainfall, emissions, and environmental quality indicators alongside socioeconomic variables — useful for studying climate-development interactions.
Use for: Teaching climate vulnerability analysis, environmental impact assessment, adaptation programme evaluation.
8. WASH Coverage
Water, sanitation, and hygiene datasets with household-level coverage indicators, water quality parameters, and sanitation access variables.
Use for: Teaching WASH programme monitoring, coverage estimation, service delivery analysis.
9. Humanitarian Response
Datasets modelled on rapid needs assessments and humanitarian response surveys — displacement, food security, shelter, and protection indicators.
Use for: Teaching humanitarian data analysis, needs assessment methods, emergency response evaluation.
10. IRT Psychometric Data
Item response theory datasets for testing psychometric and assessment methods — useful for advanced evaluation courses.
Use for: Teaching measurement theory, assessment design, test item analysis.
How Educators Can Use DevData Practice
For Data Analysis Workshops
Generate a dataset that matches your workshop topic. Teaching MEL? Generate an RCT dataset with baseline-endline data and have participants estimate treatment effects. Teaching gender analysis? Generate a gender-disaggregated household survey.
The key advantage: Every participant works with the same clean dataset, so you can focus on teaching the method rather than troubleshooting data problems.
For University Courses
Assign students datasets as homework. They can practise statistical techniques on realistic data without the ethical and logistical complexity of working with real survey data.
For Testing Analytical Approaches
Before applying a new analytical method to real programme data, test it on a synthetic dataset with known properties. This lets you verify your code and approach before working with sensitive data.
For Building Training Materials
Create datasets to accompany your handouts, slides, or course modules. A handout on "Introduction to Regression" is more useful when it comes with a practice dataset students can analyse.
Tips
Start with the dataset closest to your teaching topic. Don't try to use all 36 generators — pick the one that matches your curriculum.
Combine with ImpactMojo courses. Pair the RCT dataset with the MEL course's impact evaluation module. Pair the health dataset with the public health course.
Tell participants it's synthetic. Be transparent that practice data is generated, not real. This is an honest teaching practice and avoids confusion about what the data represents.
Use it for exam questions. Generate a unique dataset each time so exam questions can't be recycled from previous years.
840,000+ rows means you can practice with large datasets. If you're teaching data management or big-data techniques, the volume is useful.
Last updated
Was this helpful?