Puzzle
Five rounds of two charts. Each pair shows the same factually correct data. One chart frames it honestly. The other twists the framing to push a story. Pick the skewed one. We'll explain why.
Which chart is the skewed one?
Technique Truncated Y-axis
Why it's skewed
Both charts plot the same numbers: 1.7M, 1.8M, 2.0M offences. The first starts the Y axis at zero, so the bars look similar in height — a 17% rise. The second starts the Y axis at 1.6M, so the 2023 bar appears roughly four times taller than 2019. The change in real terms is identical. The visual story is not.
ONS itself flags that police-recorded crime data is heavily influenced by changes in recording practices, police activity, and public reporting — not just by what's actually happening. HMICFRS reviews have driven much of the apparent rise.
Real-world use
This is the most common chart trick in political graphics. A truncated Y axis turns a modest change into an apparent crisis (or, depending on framing, vanishes a real one). Watch for it on stock charts, polling graphs, and crime statistics — especially anywhere the headline depends on the bars looking dramatic.
Technique Cherry-picked date range
Why it's skewed
The two-year chart shows wages outpacing inflation by 2024. Technically true. The five-year chart shows the years it leaves out: inflation peaked at 11.1% in October 2022, and wages trailed inflation for roughly twenty consecutive months. Workers lost ground in real terms across that period, and the cumulative shortfall doesn't disappear because a single year now beats the index.
Real-world use
Government and treasury press releases throughout late 2023 led with "wages now outpacing inflation" without acknowledging the prior cumulative real-terms loss. The headline metric was true. The reading it invited was misleading. Whenever a chart starts at a date that's suspiciously convenient, ask what came before.
Technique Spurious correlation by dual axis
Why it's skewed
Two unrelated rising series, plotted on independently-tunable Y axes, can almost always be made to track each other. Here the eye sees correlation because both lines slope upward and end near the top of the chart. The chart never claims causation. The visual does the work.
The actual rise in recorded crime is driven heavily by fraud reclassification: in 2022, fraud accounted for 41% of all crimes, almost entirely because reporting improved and crime moved online. It has nothing to do with population movement.
Real-world use
This pairing was a recurring feature of UK tabloid coverage between 2016 and 2023. Other classics: Nicolas Cage films vs. swimming pool drownings, US science spending vs. suicides by hanging. The technique works because two unrelated time series, scaled freely on dual axes, can be made to look like one's chasing the other.
Technique Metric substitution
Why it's skewed
Capacity tells you what the system could do at full output, all the time. Generation tells you what it actually did. They aren't the same number, and the gap matters: a low-wind year drops generation even though capacity has only grown. The capacity chart looks like steady, unbroken progress. The generation chart shows a real 2021 dip caused by unusually low wind speeds.
Real-world use
Press releases from energy companies, governments, and grid operators routinely lead with capacity figures. "52 gigawatts of renewable capacity" is a more impressive headline than "42% of electricity generated." The IEA has flagged this as a systemic communication problem in climate reporting globally. When you see "capacity" in a headline, ask what was actually produced.
Technique Cumulative versus periodic framing
Why it's skewed
A cumulative line can't fall. Once a death is counted, it stays counted. The line will keep rising as long as the metric is being measured, even if the daily rate drops to single digits. The shape implies an ongoing escalation, but it's an artefact of the chart type, not the underlying trend.
The daily chart shows what's actually happening: three waves with sharp peaks and sharp falls, and a much smaller third wave. By early 2022 the rolling average was approaching pre-pandemic background levels.
Real-world use
Cumulative charts dominated media and government dashboards for two full years of the pandemic. They're appropriate for historical totals but actively misleading when used to communicate current trajectory. Epidemiologists repeatedly noted this in public — the daily rate was the trajectory; the cumulative total was the inventory.
Five rounds in. Here's where you landed.
0 of V correct
- I Truncated Y-axis Missed
- II Cherry-picked date range Missed
- III Spurious correlation by dual axis Missed
- IV Metric substitution Missed
- V Cumulative versus periodic framing Missed
If we get our reports wrong, our clients lose money or trust. Spotting these tricks isn't optional in our line of work, it's the floor.
Tell us what you're working on.
A paragraph is plenty. We usually reply within two working days, with our take on whether we're the right people, where we'd start, and roughly what it'd cost. If it's not a fit, we'll say so, and point you somewhere better.
Write to zak@tahleelconsultancy.com