Our services / 02
Data Analytics
We turn existing and newly collected data into clear, defensible answers, using methods matched to the question and to the way the data were collected. Results will be further triangulated with other sources.
01 / Scope
What it covers
- In-depth analysis of secondary data: national surveys such as DHS and PMA, routine health information (HMIS/DHIS2), census and administrative records
- Survey-weighted estimation, with standard errors and design effects
- Statistical modelling: regression, multilevel, nonparametric and multivariate methods
- Spatial and geostatistical analysis and mapping
- Data management, cleaning and harmonisation of datasets
- Dashboards and data visualisation
02 / Method
How we do it
We identify the right sources, obtain data-use permissions, and check each dataset's documentation, coverage and quality before any analysis begins.
We agree a written analysis plan with you first: the questions, indicators, comparisons and methods. Survey data are analysed with their design in mind, using sampling weights, clusters and strata, so that estimates and their standard errors are correct.
Depending on the question, we use trend and equity analysis, regression and multilevel models, nonparametric and multivariate methods, decomposition, and spatial analysis and mapping. We compare results with recognised standards and published figures, test how sensitive they are to key choices, and state plainly what the data can and cannot show.
All analysis is written in code, in Stata, R or Python, so every figure can be traced and repeated.
03 / Deliverables
What you receive
- An agreed analysis plan
- Clean, documented datasets and analysis code
- Tables, charts and maps of the results
- A report or brief written for decision-makers
- Dashboards, where you need to track results over time
Built around your needs
Who it is for
Government agencies, development partners, non-governmental organisations, researchers and universities that need evidence from data they already hold or that is publicly available, often without new data collection.