About ELSTAT
We work in the gap between academic research and the decision that has to be made next week.
ELSTAT started from a recurring complaint: important decisions get made on data nobody has checked the collection of. A number circulates, gets quoted, gets used — and no one can say where it came from.
We take the opposite position. Methodology is written before data is collected. Assumptions are stated out loud. Computation is made reproducible by someone other than its author.
The consequence is that our work is often slower at the start and considerably cheaper at the end, because it does not have to be redone when the result is challenged.
Working principles
Method first, data second
Sampling design and indicator definitions are written and agreed before a single respondent is approached. Choosing the method after seeing the data is the subtlest way to fool yourself.
Every number is traceable
Any figure in a report can be traced back to the raw row it came from and the line of code that computed it.
Uncertainty gets reported
An estimate without a confidence interval is an opinion in a lab coat. Margins of error are always reported, including when the result is inconvenient.
Results have to be usable
A report that is correct but unread is the same as no report. We design outputs for the person who has to decide, not for fellow analysts.
Have a question that needs a defensible answer?
Tell us the context. If the question is not sharp yet, sharpening it is the first part of the work.
Start a conversation