EXPERIMENT 012 / THE STORY MACHINE SAHILWHO.ONLINE

DATA JOURNALISM × HUMAN PERCEPTION

SAME NUMBERS.
DIFFERENT
STORIES.

The facts don't move. Your interpretation might. Change one decision at a time and watch a chart turn into an argument.

NO TRACKING. NO ACCOUNT. ALL EXAMPLE DATA IS FICTIONAL.

01 / INTERACTIVE EXPERIMENT

THE FRAMING MACHINE.

Meet a fictional neighborhood library. Its monthly visits are fixed; your chart decisions are not.

EXHIBIT A / FICTIONAL DATASET

A library gets busier

ZERO BASELINE

Visits recorded by month · Example year

Illustrative library usage chartInteractive chart for twelve months of synthetic library usage. Exact values are provided in the data table below.
FIRST MONTH860
LAST MONTH1,530
CHANGE+77.9%
THE INTERPRETATION

A rising number isn't a complete story.

Visits increase in this fictional dataset. But changes in opening hours, measurement definitions, and available alternatives might change what that increase means.

INTEGRITY NOTE

Changing an axis cannot change the underlying values. Truncating a bar chart's baseline exaggerates apparent differences. We expose that misleading option here only to teach recognition, and label it visibly when selected.

02 / DECODE THE TRICK

EVERY CHART
HAS AN EDITOR.

Before a number reaches your eyes, someone has decided what to count, who to include, which period to show, how to group results, and what deserves visual emphasis.

Some decisions are necessary. Others can distort. The scientific habit is not to distrust every chart—it is to ask which choices shaped this one.

Take the Field Guide ↗
03 / FIELD NOTES

FIVE QUESTIONS.
BETTER STORIES.

01 / DENOMINATOR

Compared with what?

Raw totals can rise even while the rate per person, hour or opportunity falls. Ask which comparison actually fits the question.

02 / TIME WINDOW

Why these dates?

A different start or end point may change the apparent pattern. Inspect a longer period when it's available.

03 / DEFINITION

What exactly is counted?

Visits are not unique people. Event attendance is not total footfall. Definitions shape what conclusions are possible.

04 / SCALE

How big does it look?

Bars encode quantities through length, which should begin at zero. Lines can sometimes use a tighter axis, clearly labelled.

05 / CAUSALITY

Why did it change?

A pattern can suggest a question; it cannot, by itself, establish a cause. Look for mechanisms and alternative explanations.

06 / THE WILD CARD

What's missing?

Every dataset is a selection. Find out which people, time periods or experiences never became a number at all.

04 / TRY THIS OFFLINE

THE TWO-HEADLINE CHALLENGE.

Take one data table. Make two accurate-sounding headlines with opposite emotional tones. Then write a third that is fair to the uncertainty. No inventing numbers; no hiding the source.

↓ DOWNLOAD THE COMPLETE FIELD GUIDE
THE RESPONSIBLE QUESTION

“What would I need to know before I believe the story this chart seems to tell?”

05 / RESEARCH & PROVENANCE

RECEIPTS, NOT VIBES.

All numbers in the interactive experiment were created as instructional examples. They do not describe any real library, city, institution or population. The reading below concerns general visual reasoning and chart design.

  1. UK Office for National Statistics — Axes and gridlines ↗Practical guidance on chart baselines, axes and visual interpretation.
  2. Datawrapper Academy — Why bars start at zero ↗Explanation of the perceptual problems caused by cropped bar charts.
  3. Cleveland & McGill (1984) — Graphical Perception ↗Foundational experimental research on decoding quantitative visual displays.