Reading graphs and statistics

Reading graphs and statistics: a chapter of the Thinking Clearly Book, with the mind traps and tricks that are worth knowing by name.

11 cards · Science

Start learning

Each Zap teaches up to 5 new cards, so you learn the chapter a few at a time.

Preview

  • A bar chart whose axis doesn't start at zero

    Show answer: A bar chart whose axis doesn't start at zero

    A truncated axis

    Learn more: Cutting off the bottom of the axis exaggerates differences; always check where the scale starts.

  • Axis where each step multiplies by 10

    Show answer: Axis where each step multiplies by 10

    Log scale

    Learn more: Useful for epidemics and money, but it can hide how fast things are really rising.

  • Choosing only the dates that tell your story

    Show answer: Choosing only the dates that tell your story

    Cherry-picking

    Learn more: Start or end a chart at the right moment and almost any trend appears; ask what the full record shows.

  • "Doubles the risk", with no starting number

    Show answer: "Doubles the risk", with no starting number

    Relative risk

    Learn more: Always ask for the absolute risk: going from 1 in a million to 2 in a million is "double".

  • The ±3% that comes with a poll

    Show answer: The ±3% that comes with a poll

    Margin of error

    Learn more: A 2-point lead inside a ±3% margin is too close to call.

About this topic

A chart can mislead without a single false number. An axis that doesn't start at zero makes small gaps look huge, a log scale tames fast growth, and a well-chosen start date can make almost any trend appear. This chapter teaches the questions to ask of any graph or statistic: where the scale starts, what the absolute numbers are, how big the margin of error is, who was asked, and whether two things moving together really have anything to do with each other.

This is a chapter of the Thinking Clearly Book. Each card asks one question, with a hint before you answer and a short reason after. Once you know most of this chapter, the next one starts arriving in your Daily Zap.

All 11 cards

  1. A bar chart whose axis doesn't start at zero

    A truncated axis

    Learn more: Cutting off the bottom of the axis exaggerates differences; always check where the scale starts.

  2. Axis where each step multiplies by 10

    Log scale

    Learn more: Useful for epidemics and money, but it can hide how fast things are really rising.

  3. Choosing only the dates that tell your story

    Cherry-picking

    Learn more: Start or end a chart at the right moment and almost any trend appears; ask what the full record shows.

  4. "Doubles the risk", with no starting number

    Relative risk

    Learn more: Always ask for the absolute risk: going from 1 in a million to 2 in a million is "double".

  5. The ±3% that comes with a poll

    Margin of error

    Learn more: A 2-point lead inside a ±3% margin is too close to call.

  6. A poll that only asks people online at noon

    Selection bias

    Learn more: If the sample isn't like the population, even a huge poll can be wrong.

  7. From 10% to 12% is a rise of 2…

    Percentage points

    Learn more: From 10% to 12% is 2 percentage points but a 20% rise: both are true, and they sound very different.

  8. A pie chart's slices should add up to

    100%

    Learn more: If the slices add up to more than the whole, the chart is mixing up categories.

  9. One value far from all the others

    An outlier

    Learn more: One billionaire in a room lifts the average income hugely; the median barely moves.

  10. Cheese eaten tracks deaths by bedsheet tangling

    Spurious correlation

    Learn more: With enough data sets, some move together by pure chance; that is not cause and effect.

  11. Comparing countries fairly by dividing by population

    Per capita

    Learn more: China emits the most CO₂ in total; per person, the US emits more.

More science decks

Each deck takes about 1 min a day.

All science decks