Eight databases. One clean country-year panel. Every step on record.
Data Merge Studio connects straight to the World Bank, the UN, the WHO, the ILO, UNESCO and the big research repositories. Pick the indicators and countries, and it downloads them, lines them up by country and year, checks for gaps and duplicates, and builds one file your software can open. Nothing is estimated, averaged or filled in without you saying so.
Get Data: choose a source, search its indicators, pick countries and years.
The databases you'd otherwise visit one by one.
Each one through its official interface, with the current list of indicators fetched live.
Ten tabs, in the order the work happens.
You can stop at any tab. Export downloaded data as it is, or carry on to a merged panel.
A versioned panel you can defend in a review.
Choose the sources, mark each as observed or projection, and run a dry run first. It reports coverage by country and year, duplicates, mapping problems, the common sample, Caribbean and small-island coverage, and how many observations each merge step would lose. Nothing is built until you approve it.
- Each build goes into its own dated, numbered folder: the panel as Stata, CSV and Excel, a separate projection file if chosen, the variable dictionary, the country concordance, the coverage matrix and the merge report.
- A manifest with a SHA-256 checksum for every file, and a README, so the package can be shared or archived as a unit.
- Data is kept in layers: raw, clean, validated, master and projection. The raw download is never touched.
- No interpolation, no averaging across providers, no silent substitution. Any override you choose is written into the record.
Merge / Build Panel: dry run first, build on approval.
Built for the parts of data work that go wrong.
Country-year alignment
Everything is keyed by ISO3 code and year. A bundled World Bank concordance maps each source's own codes, and a coverage matrix shows, per variable, which countries and years you'd have, and how many countries drop out as you add variables.
Validation that can block an export
Checks for unreadable files, inconsistent columns, missing keys, duplicate keys, gaps in the time series and missing cells. A fail stops the export unless you override it, and the override is recorded.
Cache, compare, freeze
Use the cached copy, download fresh, or compare the two and see what the provider changed. Freeze a snapshot of raw, clean, derived and metadata files when a paper is submitted, so the numbers can be reproduced later.
Provenance on every number
Each download keeps the full request URL, the raw response, the time, the HTTP status, the app version and a checksum. Each export writes an audit file beside it. The Excel export carries Audit, Missingness, Duplicates, Merge Report, Sources and API Provenance sheets.
Recipes
Save the sources, indicators, countries and processing rules as a recipe and run it again when the data is updated. The same panel, rebuilt in minutes, with a new version number.
Your own files too
Bring in Excel and CSV files, inspect them, tidy column names, reshape, and append or join them to downloaded data by key. Compare a downloaded series with a local file exactly or within a tolerance you set.
Honest about the network
Shows you the exact request before it's sent, tests the connection, retries cautiously, and never scrapes a web page. The UN token travels only in a request header and is kept out of URLs, logs and exports.
Nothing leaves but your queries
The only traffic is to the data providers you chose. No analytics, no accounts, no telemetry. Downloads, recipes and exports live in your own folders on the Mac.
Things researchers ask first.
Which file formats does it export?
Stata (.dta, the current format), Excel (.xlsx with the data and the audit sheets) and CSV, in long or wide layout. Files are never overwritten: a new version gets the next free name.
Does it fill in missing values?
No. It shows you where the gaps are and what each merge would cost you in observations, and leaves the decision to you. There is no interpolation and no averaging across providers.
What about IMF, OECD or Comtrade?
Not yet. The eight connectors above are the current set. Tell us which source you need and we'll look at adding it.
Do I need an account anywhere?
Only the UN Population Division asks for a free access token, which the app stores in your Mac's Keychain. The other sources need nothing.
Does it run on Windows?
Not at the moment. Data Merge Studio is a Mac app. It doesn't need Python or anything else installed.
How do I get it?
It's in testing. Email support@mabeleinnovations.com and we'll send you the installer and the user manual. The first launch needs a Control-click, Open, because the app isn't yet notarised by Apple.
Spend the week on the analysis, not the merge.
Ask for a test copy of Data Merge Studio.