Data Lab · Unit 5 of 8
Unit 5: Charts that tell the truth
Choosing the right chart, honest axes, clear labels and spotting misleading charts.
Opens the first lesson of this unit in Data Lab. Every lesson is open from the start.
Lesson 1: Choosing the right chart
Goal: Match the question to the chart type that answers it.
A chart is an answer to a question. Pick the chart by asking what kind of question it is:
| Question | Chart | Example |
|---|---|---|
| How did something change over time? | Line chart | Website visitors each month |
| How do categories compare? | Bar chart | Sales by product |
| What are the parts of a whole? | Pie chart (only a few slices) or a stacked bar | Share of budget spent on rent, food, transport |
| Are two numbers related? | Scatter plot | Hours studied versus test score |
| How are values distributed? | Histogram | How many deliveries took 0–10, 10–20, 20–30 minutes |
Bar charts are the safest default for comparisons: people judge the length of bars accurately. Pie charts are hard to read with more than about five slices, or when slices are similar sizes, because people judge angles poorly. If you are not sure, use a bar chart.
Hands-on lab in the room: Chart lab. For each question, choose the chart type and see whether it answers the question clearly.
- Line chart
- Points joined by lines, usually showing change over time.
- Bar chart
- Bars whose lengths compare values across categories.
- Scatter plot
- Dots placed by two numeric values to show whether they are related.
- Histogram
- Bars showing how many values fall into each range, revealing the shape of the data.
Remember: Over time: line. Compare categories: bar. Related numbers: scatter. Parts of a whole: pie, only with few slices.
Lesson 2: Honest axes
Goal: Recognize truncated axes and decide when a zero baseline is required.
In a bar chart, the length of each bar stands for its value. So the value axis must start at zero. If it starts at, say, 95, a small difference looks enormous.
Worked example: the same data, two pictures
Store A sold 98 units; Store B sold 100.
- With the axis starting at 0, the bars look almost equal, which is true: B sold 2% more.
- With the axis starting at 97, Store B's bar looks three times as tall as Store A's.
- Nothing about the data changed; only the axis did.
Hands-on lab in the room: Chart lab. Toggle the axis start and see how the same numbers can tell two different stories.
Line charts are different: they show change through the slope of the line, not its height, so zooming in on the range of the data is often fine, such as a body temperature chart from 36 to 39 °C. Just label the axis clearly so no one is fooled.
Other axis tricks to watch for: uneven time intervals (2010, 2015, 2016, 2017), two different y-axes on one chart that make unrelated lines look connected, and axes with no labels at all.
- Baseline
- Where the value axis starts. Bar charts need a zero baseline.
- Truncated axis
- An axis that starts above zero, which can exaggerate differences in a bar chart.
Remember: Bar charts start at zero. Line charts may zoom in, but the axis must be clearly labeled.
Lesson 3: Titles, labels and one clear message
Goal: Design a chart so a busy reader gets the point in five seconds.
Assume your reader is busy. They should understand your chart in about five seconds. That means:
- A title that states the finding, not just the topic. Not "Sales by month" but "Sales doubled after the summer menu launched".
- Labeled axes with units: "Visitors (thousands)", "Delivery time (minutes)".
- One main message per chart. If you have three points to make, make three charts.
- Highlight what matters. Use one strong color for the bar or line you are talking about and grey for the rest.
- Remove clutter: 3D effects, heavy gridlines, backgrounds and decorations that do not carry information.
Worked example: before and after
- Before: "Q3 Data", rainbow-colored 3D bars, no axis labels, a legend with 8 entries.
- After: "Online orders overtook in-store in September", two simple lines, online in orange, in-store in grey, axis labeled "Orders per week".
The second chart tells the reader what to notice and proves it at a glance.
Also think about accessibility: do not rely on red versus green alone (some readers cannot tell them apart), use enough contrast, and describe the main point in text near the chart.
Remember: Title = the finding. Label axes with units. One message, one highlight color, no clutter.
Lesson 4: Spotting misleading charts
Goal: Read charts critically and catch common distortions.
Charts can mislead by accident or on purpose. Train yourself to check these every time:
| Trick | What to check |
|---|---|
| Truncated bar axis | Does the bar chart's axis start at zero? |
| Cherry-picked time range | Would a longer time range tell a different story? A "huge rise" may just be a normal bounce after a dip. |
| Missing context | Totals versus per person: a big city will have more of almost everything simply because more people live there. |
| 3D and pictures | Do 3D slices or icons make some values look bigger than they are? |
| Correlation dressed as cause | Two lines moving together does not prove one causes the other (Unit 8). |
| No source | Where did the data come from, and when? |
Worked example: per person, not totals
- City X had 900 bike accidents; Town Y had 90.
- City X has 1,000,000 people; Town Y has 50,000.
- Per 10,000 people: City X 9 accidents, Town Y 18.
- Town Y's rate is twice as high, the opposite of what the totals suggested.
Rates make fair comparisons between groups of different sizes: per person, per 1,000 people, per customer, per hour.
- Cherry-picking
- Showing only the part of the data that supports a point, such as a convenient time range.
- Rate
- A value divided by a size, such as accidents per 10,000 people, which allows fair comparisons.
Remember: Check the axis, the time range, totals versus rates, and the source.
Practise it in Data Lab
Each lesson ends with two or three questions that check you can use the idea, not just recognise it. Wrong answers explain why and come back in Review. The unit ends with a 5-question quiz; pass it with 80% or more to master the unit. Hands-on lab: Chart lab.
Every lesson is free: no ads, no lives and no paywall.