Original Research & Editorial Storytelling | Ilya Nikitin
I turn unanswered questions into research worth reading. I develop testable questions, audit original evidence and data, and build clear editorial stories around what survives scrutiny. The conclusion is not chosen in advance.
↓ Download portfolio PDF (5 pages · 87 KB)
Selected research and editorial work
Three published projects show different parts of the process: constructing and testing a new comparison, auditing a scientific claim, and turning data into an interactive tool.
01 · Original secondary analysis
Finland is #1 for life evaluation, but #114 in an exploratory smiling model. Why?
The finding: The two rankings measure different constructs. In the Smiling Observation Synthesis Index (SOSI), an experimental model based on selected sources, Finland placed 114th of 125. Across overlapping source comparisons, median agreement on which country smiled more was only 59.1%. The striking contrast became a story about what rankings can and cannot establish.
“The survey might be flawless. But satisfaction with one’s running fitness is not the same as running speed.”
Evidence boundary: The 10,740 comparison rows are derived within-study comparisons from selected source families, not independent observations of people. SOSI is not a representative measure of national happiness or smile frequency.
Read the published feature · Research report, methods and data
02 · Scientific claim audit
At the celebrated 8.8-hour coffee cutoff, the model still predicts about 12 minutes of lost sleep.
The finding: A widely cited caffeine timing rule was derived from a model’s confidence interval reaching zero. Reconstructing the central prediction from published rounded coefficients shows that the predicted effect does not disappear at the quoted cutoff. Statistical uncertainty is not the same thing as a biologically safe boundary.
“Falling asleep tells you how the night began, not what happened next.”
Evidence boundary: This reconstructs a published model estimate. It does not measure an individual’s sleep loss or establish a universal cutoff.
03 · Interactive research feature
The same move looks different when you change the income assumption.
The finding: For Finland to Spain, the model illustrates a +33.8% change in purchasing power when spendable income is held constant, versus a −11.0% change under a population-level local-income proxy. These are different measures, not competing personal salary forecasts. Making that distinction visible is part of the story.
“A Finn who moves to Spain does not become a statistical Spaniard at passport control.”
The tool: An interactive, 54-country, seven-dimension comparison that lets readers explore assumptions instead of accepting a single country ranking.
What you can commission
Original Data Investigation
A focused question, public or client-supplied data, a transparent analytical method, robustness checks and a finished evidence-led feature. Typical deliverables: analysis note, graphics, source ledger and article.
Scientific Claim Audit
A close examination of an influential result or public claim: primary sources, reproduced calculations, competing explanations, limits of inference and a reader-friendly account of what actually holds up. Typical deliverables: reproducibility memo, explanatory graphics and article.
Every project begins with a written brief and agreed scope. Source access, tools (including any AI use), deliverables, review rounds and deadlines are set in writing. A narrowly scoped public-data pilot may be feasible within 3–5 working days; larger investigations require separate scoping.
About Ilya Nikitin
I am a Finland-based researcher and research-focused author at Mike’s Balance, and founder of The Null Institute, a research initiative operated by IMT First Oy. My academic qualification is Candidate of Technical Sciences (microelectronics, Saint Petersburg Electrotechnical University “LETI”). My work focuses on measurement, quantitative reasoning, model assumptions and making uncertainty understandable. My degree is in engineering, not clinical medicine.
Discuss a project
Bring a question. Get a defensible story.
For a concrete brief, dataset or scientific claim worth testing, email contact@mikesbalance.com. Please include the intended audience, available sources and desired publication or decision.