Loading collection…

Last revised Sep 24, 2026

Think Like Data Analysts

Don't ask what the number is. Ask where it came from.

Most people meet data analysis at the end of the pipeline — a dashboard, a percentage, a model output with a confident number attached. By the time it arrives, every decision that determines whether the number means anything has already been made by someone else: who was counted, what the baseline was, which cuts survived. This collection works backwards up that pipeline and hands the reader the questions that working analysts run automatically. Across 30 articles and 6 sections, it covers the interrogation moves that separate a finding that will hold from one that will evaporate — without formulas, without software, and without a statistics background. The moves work in a meeting, with a chart on screen and no tools in hand.

Field GuideReframe

Earn4CreditsinCritical ThinkingScientific ThinkingEpistemology

6Modules30Sessions275Cards70Quizzes

Learn more about the subscription plan

Short on time?

What you can earn

Certificates, credits, and a record of your progress.

Completion certificate

Reach 60% of its base credits to earn a Certificate of Completion.

View example certificate

4 base learning credits

Credits accumulate in your learning tracks, building a subject-specific record.

Learning transcript

See your credits, reading progress, quiz performance, and available exam results.

View a sample report

Boost your score with a speed exam · Where available, take a short, timed speed exam in the mobile app to add up to 30 points to your Achievement Report score. This collection does not currently offer a speed exam.

Modules in this Collection’s System

One session. One clear takeaway.Each short session develops one distinct idea with a clear takeaway. Together, the sessions show how those ideas connect into a structured understanding of the topic.

The Question Before the Data

Settling what you are actually asking before touching the dataset — because a clean analysis aimed at the wrong question is wasted entirely.

5Sessions

Where Numbers Come From

How data is generated, who falls out of it, what the instrument actually measured, and the judgement calls buried in every cleaning step.

5Sessions

Reading a Number Honestly

The baseline, the distribution, the noise floor, and the aggregation traps that make an accurate number support a false conclusion.

5Sessions

From Correlation to Cause

What a relationship in observational data licenses you to claim — and the reasoning machinery that gets you toward something worth acting on.

5Sessions

Uncertainty and Being Wrong

Stating what you do not know, resisting the mechanisms that manufacture findings, and keeping score of your own past confidence.

5Sessions

Analysis Meets Decisions

What happens to analysis once it enters an organization — where measurement becomes a target and independence erodes one accommodation at a time.

5Sessions

What You'll Walk Away With

6interrogative frameworks for reading a number before it has a chance to mislead you
5questions analysts run automatically that almost no one asks out loud in a meeting
4causal reasoning tools for moving from a correlation to a claim worth acting on
3uncertainty practices for stating what you don't know without leaving the room empty-handed
6organizational failure patterns where analysis loses its independence one accommodation at a time

You'll Have Answers To

?Why does changing the denominator — without touching a single count — reverse a conclusion?
?How do working analysts separate a genuine signal from the routine variation that surrounds every process?
?What does it actually take to move from an observed correlation to a causal claim worth acting on?
?Why does statistical significance so often coexist with practical irrelevance — and which number should be on the slide?
?What is the professional move when the data genuinely cannot answer the question being asked?

Critical Concepts Explored

Denominator DependencySelection BiasOperationalizationDefinition DriftCounterfactual ReasoningRegression to the MeanSimpson's ParadoxGarden of Forking PathsCalibrationGoodhart's Law

Editor's Note

“The sharpest short course in statistical scepticism we have built yet.”

Most numeracy education teaches you to compute. This collection teaches you to interrogate — to ask what the denominator is before reading the rate, who is absent from the sample before trusting the pattern, and when the honest answer is that the data cannot settle the question. Thirty articles, zero formulas, and fully usable in any meeting.

Editor's Brief

Who it's for: Anyone handed data and expected to act on it — managers, researchers, founders, and consultants who work upstream of the analysis and downstream of its consequences.
What stands out: Unlike statistics courses, this collection requires no tools, no software, and no prior training in probability — every move works in a meeting with nothing but a slide on the screen. It also treats analyst failure modes — organizational pressure, Goodhart's Law, narrative capture — as core content rather than closing caveats.
Read if: Read if you would rather ask the right question about slide 12 than nod at a chart you privately do not trust.

Gold Quotes

The question every analyst runs before anything else is not 'what does the data say' — it is 'who is missing from it.'

Every dataset is a record of who and what survived the collection process. The absent cases — churned customers who skipped the survey, failed projects that went undocumented, candidates who never made it to the interview — are usually the ones most correlated with what you are trying to understand. Reading the data without asking who is missing from it is like reading a verdict without knowing who was excluded from the jury.

A point estimate is a range with the honesty removed. The width of the interval is the finding, not the embarrassment.

When an analyst reports a range instead of a single number, most audiences hear indecision. What they are actually hearing is precision — a statement about how much the evidence can and cannot settle. A narrow interval is high confidence; a wide one is a genuine finding about the limits of the data. The pressure to compress a range into a point is the most common source of false certainty in organizational decision-making.

An analysis nobody can retrace is a rumour with a chart attached.

Reproducibility is not a hygiene chore — it is an epistemic property of the analysis itself. When you know a colleague must be able to retrace every step of your pipeline, the decisions you make change while you are still making them: which exclusions you record, which cuts you report alongside the one that worked, which assumptions you state rather than leave implicit. The audit trail does not just verify the result; it makes the result more likely to be correct.

About the Curator

TThink Like Great Minds

Think Like Great Minds is a LearningFirst original series that borrows the reasoning habits of world-class practitioners — analysts, mathematicians, diplomats, game designers — and translates them into transferable moves for anyone who has to think carefully in a room where the answer isn't obvious.

Readers also explored

Ranked by semantic similarity