From data to political strategy
Why a dashboard is not enough: how to combine statistics, discourse analysis and OSINT to turn public information into diagnoses and decisions.
Matías Gaglio ·
Leer en españolThe bottleneck is no longer access to data.
Collecting, organizing and connecting scattered information is an advantage. But handing over a database or a dashboard without interpretation can give the client a new problem: more information, plus the time, tools and judgment needed to process it. The value appears when you can rank what matters and separate relevant signals from noise.
Quantitative and qualitative, together.
An algorithm can detect a regularity; a qualitative reading explains what it means and in what context it happens. Spearman correlations, factor analysis or clustering help find structure, but counting mentions or auto-classifying comments is not enough to understand a conversation: the same expression can be support, irony or rejection.
Multidimensional political spaces
Compare declared identity, concrete positions and chosen leaders, and study the spread, borders and overlaps between groups.
Which answers and topics appear together
Network analysis shows how conversation communities are structured and which topics connect positions.
Open sources as a cross-cutting method
Connect news, public records, documents and posts with verifiable sources, context and traceability.
What a third party can find
Records and statements still circulating, impersonations and outdated information about a public figure.
Non-obvious patterns, with discipline.
I am especially interested in relationships we would not have thought of, which emerge when many variables are analyzed together. But finding an association is where the research starts, not where it ends.
- Bias and confounding: check selection bias and variables that drive both sides at once.
- Multiple comparisons: with many tests, some relationships appear by chance.
- Out-of-sample validation: a convincing explanation written after seeing a pattern is still a hypothesis.
- Scenarios, not prophecies: prediction is useful for estimating probabilities that can be checked later.
Continuous public opinion measurement.
I see room in Argentina for a stable measurement product that combines polls, digital conversation, open sources and qualitative analysis, with a transparent methodology. Measuring well means knowing what each source adds: social networks show discourse and topic circulation; polls need sample design and weighting.
One research infrastructure can support recurring reports for public agencies, universities, civil society and companies, and add specific studies or strategic support case by case.