Data Collection
KoBoToolbox · ODK Collect · Survey Solutions
Most organisations set up their data collection tools incorrectly. Forms are poorly designed, skip logic breaks, validation rules are missing, and data exports are messy. We design and deploy your data collection system correctly from the start, so the data that arrives is clean, complete, and ready to analyse.
"Clean data starts with a properly built collection system, not with cleaning after the fact."
What We Build
- Survey and form design with skip logic and validation
- Server setup and user management
- Quality control protocols and real-time monitoring
- Clean data pipeline from collection to analysis
- Enumerator training and field supervisor protocols
- Pilot testing and instrument validation
Platforms
- KoBoToolbox, structured household and census surveys
- Survey Solutions, CAPI/CAWI for large surveys
- Web scraping and online data extraction — collecting structured data from websites, government portals, and online databases using Python (BeautifulSoup, Scrapy) and R (rvest)
- Custom web-based data entry for offline contexts
Available for: NGOs, health programmes, government surveys, research institutions. Also available for individual researchers setting up their first data collection project.
Statistics Services
Sample Size · Survey Design · Data Cleaning · Descriptive and Inferential Statistics · Statistical Reporting
"Good statistics are not just about running the right test. They are about asking the right question, collecting the right data, and reporting the results honestly."
Most data problems start before data is collected. A questionnaire with poor question design, a sample that is too small, or variables that were never coded correctly produce results that cannot be trusted — no matter how sophisticated the analysis that follows. We work with you from the design stage through to the final write-up.
Rigorous statistics. Results you can defend.
What We Do
- Sample size and power calculations — before data collection begins, determining how many participants are needed to detect a meaningful effect
- Survey and questionnaire design — question wording, response scales, skip logic, piloting, and validation
- Data cleaning and management — recoding, merging datasets, handling missing data, creating analysis-ready datasets
- Descriptive statistics — frequencies, percentages, means, medians, standard deviations, cross-tabulations
- Inferential statistics — hypothesis testing using t-tests, chi-square, ANOVA, Mann-Whitney, and related tests
- Statistical reporting — results written clearly for reports, donor submissions, journal articles, and publications
- Peer statistical review — checking an existing analysis for methodological errors, incorrect test selection, or reporting problems
When You Need This
- You are designing a baseline or endline survey and need to know your sample size
- You have collected data but are unsure which statistical test is appropriate
- Your analysis needs to meet the standard of a donor, ethics committee, or peer reviewer
- You have results but need them written up clearly for a non-technical audience
- You want a second expert opinion on statistical work before it is published or submitted
- You are a student or researcher preparing a thesis, dissertation, or journal submission
Available for: NGOs, health programmes, government agencies, research institutions, and individual researchers and students preparing thesis work, publications, or donor reports. Also available as a Data Consultation Session for one-off statistical questions.
Data Analytics
R · Stata · Python · SPSS · Excel
We handle complex, messy, real-world datasets and produce analysis that stands scrutiny, from a field officer's debrief to a World Bank evaluation panel. Survey data, programme records, financial figures, longitudinal evaluations. We analyse all of it using the right statistical tool for the question.
What We Analyse
- Complex survey and programme data
- Statistical modelling and regression analysis
- Inferential statistics and hypothesis testing
- Longitudinal data and trend analysis
- Mixed-methods data integration
- Financial and operational data
What We Produce
- Data cleaning and quality assurance reports
- Statistical analysis outputs with interpretation
- Professional visualisations and charts
- Analytical reports for any audience
- Reproducible code in R, Python, or Stata
- Executive summaries for decision-makers
Available for: Organisations with data they cannot use. Also for individual researchers and students who need expert analysis for theses or research projects.
Note: Data Analytics covers descriptive analysis, exploratory work, and standard reporting. For regression modelling, mixed models, time series, and advanced inference, see Statistical Modelling below.
Statistical Modelling
R · Stata · Python · SPSS · Regression · Mixed Models · Time Series · Survival Analysis · Clustering · Classification · Pattern Discovery
Statistical modelling goes beyond descriptive analysis. We build models that explain relationships, test hypotheses, account for confounding, and produce estimates your team and your donors can defend. Whether you need a logistic regression for a health outcome, a mixed-effects model for longitudinal programme data, or a time series forecast for resource planning, we choose the right model for your question and your data structure.
Every model we build comes with documented code, interpretation of results in plain language, and guidance on limitations. We do not just run the model. We explain what it means and where it should not be trusted.
Text Analysis and Text Mining
Structured data is only part of the picture. Organisations collect enormous amounts of unstructured text — open survey responses, community feedback, interview transcripts, policy documents, and reports. Text mining extracts patterns, themes, and quantifiable insights from this text.
We use text mining for: thematic analysis of open-ended survey responses, sentiment analysis of community feedback, topic modelling across large document collections, keyword frequency and co-occurrence analysis, and extracting structured data from unstructured reports.
R (tidytext, tm, quanteda)
Python (NLTK, spaCy, gensim)
MAXQDA
Types of Models We Build
- Linear and multiple regression — outcomes, associations, predictions
- Logistic and probit regression — binary and ordered outcomes
- Mixed-effects and multilevel models — clustered and longitudinal data
- Time series analysis — trends, seasonality, forecasting
- Survival and event history analysis — time-to-event data
- Factor analysis and structural equation modelling
- Propensity score and matching models
- Poisson and negative binomial — count data
- Clustering and segmentation — K-means, hierarchical, DBSCAN — grouping records without predefined categories
- Classification models — decision trees, random forests, gradient boosting — predicting categories from data
- Association rule mining — discovering patterns and relationships across large datasets
- Anomaly and outlier detection — identifying unusual records that do not fit the expected pattern
What We Deliver
- Model specification document — why this model for this question
- Clean, documented code — R, Stata, Python, or SPSS
- Diagnostic checks and assumption tests — with results
- Output tables formatted for reports and journal submission
- Plain-language interpretation for non-technical audiences
- Sensitivity analysis where appropriate
- Guidance on limitations and what the model cannot answer
Available for: Health programmes, research institutions, NGOs running evaluations, government agencies, universities, and individual researchers building models for theses or publications. Also available as a Data Consultation Session for one-off modelling support.
Data Consultation Sessions
On-Demand Expert Sessions · Organisations and Individuals Welcome
For organisations and individuals who need expert data support on a specific problem fast. Book a session with a senior data specialist. We diagnose the problem, work through it with you, and leave you with a solution and the understanding to handle it yourself next time.
No organisation? No problem.
The Data Consultation Sessions is open to individuals. Students, researchers, professionals, and anyone with a data problem they need expert help solving. Book a single session or a block of sessions.
Data Cleaning
Statistical Analysis
Survey Review
Visualisation
MEL Data Processing
Research Support
How to Book a Data Consultation Session
Sessions are available in person in Dar es Salaam and online via Zoom or Google Meet. Individuals and organisations are both welcome. You do not need a long brief — just a clear description of the problem you need help with.
Single Session
One focused session — 60 to 90 minutes — on a specific data problem. Best for a targeted question or a stuck point in your analysis.
Block of Sessions
Three to five sessions over a week or more — for ongoing support through a research project, thesis, or analysis assignment.
To book, send us a short message with your name, the problem you need help with, and whether you prefer in-person or online. We will confirm availability and session fee within one business day.
Monitoring, Evaluation and Learning
Donor-Grade MEL · USAID · World Bank · EU · GIZ Standards
MEL is ongoing monitoring throughout a programme. Impact Evaluation is a one-time rigorous test of whether the programme caused the change. If you need the latter, see Impact Evaluation below.
We design MEL frameworks that satisfy rigorous donor standards. We collect baseline data, run evaluations, and produce evidence that holds up to scrutiny. Every MEL system is built to generate the evidence your donor requires and the learning your programme needs.
Framework Design
- Theory of change development
- Results frameworks and logframes
- Indicator selection and data dictionary
- MEL plans in USAID, EU, GIZ, World Bank formats
- Data quality assessment systems
Data Collection and Analysis
- Baseline, midline, and endline evaluations
- Mixed-methods, quantitative and qualitative
- Focus groups, key informant interviews
- Statistical analysis using R, Stata, Python, SPSS
- Donor reporting packages formatted to requirements
Impact Evaluation
Experimental and Quasi-Experimental Designs · Econometric Methods
We conduct rigorous impact evaluations using control groups, counterfactual analysis, and econometric methods to prove your programme caused the change. Built to the evidence standards required by USAID, World Bank, EU, and GIZ.
Evaluation Designs
- Randomised Control Trials
- Difference-in-difference
- Propensity Score Matching
- Regression Discontinuity Design
- Instrumental Variables
- Pre-post with comparison group
What We Deliver
- Evaluation design document and protocol
- Validated data collection instruments
- Full quantitative and qualitative data collection
- Statistical analysis with documented code
- Final report to donor standards
- Findings presentation to donors and boards
Database Design and Development
MySQL · PostgreSQL · Client Registries · Beneficiary Systems · Programme Data Stores
Many organisations collect data but store it in spreadsheets that break, duplicate, and contradict each other. A properly designed database solves this permanently. We design, build, and hand over relational databases tailored to your specific programme or organisational needs, with full documentation and staff training included.
What We Build
- Beneficiary and client registries
- Programme monitoring data systems
- Financial and grant tracking databases
- Asset and inventory management systems
- Research and survey data repositories
- Custom data entry forms and interfaces
- Reporting and dashboard data backends
What We Deliver
- Database design document — tables, relationships, data dictionary
- Built and tested database ready for use
- Data entry interface appropriate to your team's skills
- Data migration from your existing spreadsheets
- User roles and access control setup
- Staff training — your team manages it independently
- Full technical documentation for future developers
Available for: NGOs replacing spreadsheet systems, government agencies building data infrastructure, health programmes needing patient registries, research institutions managing longitudinal datasets, and startups building data foundations.
BI and Report Automation
Power BI · Python · R · Automated Pipelines
Finance teams and programme managers in most organisations spend two to four days every month pulling numbers from multiple sources and building the same report they built last month. We build automated reporting systems that eliminate this entirely.
"Stop building the same report every month. Automate it once. Run it forever."
What We Build
- Connect multiple data sources, databases, Excel, APIs
- Automated data transformation and cleaning
- Formatted reports generated automatically
- Scheduled delivery to management, boards, donors
- Full handover, your team runs it independently
Tools Used
- Power BI, interactive dashboards and reports
- Python (pandas, openpyxl, reportlab)
- R (rmarkdown, officer), reproducible reports
- SQL, database queries and extraction
Real-Time Monitoring Dashboards
Built Once · Monthly Maintenance · Always Live
We build and maintain live monitoring dashboards that pull your programme data automatically and display it in clear, professional visuals for leadership, field managers, and donors. One-time build. Monthly subscription maintenance. Always current.
What We Build
- Programme performance dashboards
- Beneficiary tracking and mapping views
- Budget monitoring and financial oversight
- Donor-shareable views to their format requirements
- Multi-level access, management, field, donor
Platforms
- Power BI, interactive, enterprise-grade
- Python (Dash, Streamlit), custom interactive apps
- R (Shiny, flexdashboard), statistical dashboards
- Custom web dashboards, fully branded
GIS & Geospatial Services
Location Data · Spatial Analysis · Remote Sensing · Web GIS · Consultancy
Location adds a powerful dimension to data. We turn coordinates, boundaries, satellite imagery, field observations, and programme records into clear maps and spatial evidence that support planning, monitoring, research, and decision-making.
"Turning location data into maps, analysis, and decision-making insights."
Core GIS Services
- Spatial Data Collection — GPS and mobile collection of coordinates, facilities, assets, households, project sites, and other georeferenced information.
- Spatial Analysis — proximity, buffering, overlay, accessibility, service coverage, hotspot, spatial joins, other location-based analysis, and production of analytical and thematic maps.
Advanced & Capacity Services
- Remote Sensing — satellite-image analysis for land use, land cover, environmental monitoring, change detection, and related applications.
- Web GIS & Dashboards — interactive online maps and geospatial dashboards for monitoring, communication, and decision support.
- GIS Training & Consultancy — practical GIS training, workflow design, technical support, spatial-data management, and institutional GIS guidance.
Available for: NGOs and development programmes, government and local authorities, research institutions and universities, agriculture and environmental projects, health programmes, businesses, and individual researchers who need mapping or spatial analysis support.