The upper half is a line of monthly temperatures, April to September. The lower half is eight test scores as a histogram, with the median beside it. Edit a number card and the chart and the median move with it. The chart type is a menu on the card itself: bar, pie, line or histogram.
How to read this board
On a line chart the order of the connections is the horizontal axis. April, May, June and so on appear in the order they are wired, which is what makes it a chart for ordered data — reorder the points and the meaning changes.
A histogram chooses its own bins (round widths like 60, 70, 80). It resembles a bar chart, but a histogram leaves no gaps between the bars: the values are continuous and what is being counted is how many fall in each interval.
Where it trips you up
The places that are easy to get wrong, next to how to check each one on the board.
| Where it trips you up | Why that happens | How the board shows it |
|---|---|---|
| A bar chart is not a histogram | They look alike. A bar chart compares categories (apples, oranges); a histogram counts an interval (how many scored in the sixties). | Set the score card to histogram and the bars close up. Switch to bar and the gaps come back. That gap is the difference in meaning. |
| When a line chart is the wrong chart | Put unordered data on a line — average scores by class, say — and the segments between the points suggest a progression that does not exist. | The temperature cards are wired in month order. Rewire them and the shape changes, which is the order carrying meaning. |
| The mean and the median part company | Watch only the mean and a single extreme value moves your impression of the whole set. | Change one score from 93 to 5. The median hardly moves; the mean drops sharply. |
| Bin width changes the picture | Widen the bins and the data can look like one hill; narrow them and the same data can split into two. | The histogram picks its bins from the spread, so widening the range of the scores changes the bins too. |
| The vertical range changes how big a difference looks | A bar chart that does not start at zero makes small differences look large. A line chart is a different case: narrowing the range to show detail is normal, and not starting at zero is not by itself an error. Read the axis along with the shape. | The line card draws its dashed zero baseline only when zero is inside the range, so that temperatures below freezing still plot naturally. |
| When there is not much data | Eight people is eight people. What breaks is the step from there to a claim about a larger group — and the fewer the data points, the wider that uncertainty. | Delete a score card. The histogram changes shape from one removal. Read the number of data points alongside the result. |
The same data, and each way of showing it makes a different feature easy to see. A line suits change along an order, a histogram suits the distribution of values, and the mean or median stands in for the set as a whole. All three are here so that no single view has to carry the judgement.
Try this
- Push one score to an extreme — the mean moves and the median holds. "Typical value" covers two things that behave differently.
- Drop one temperature sharply — the line's vertical range widens and the other months' differences flatten out, because the scale adjusts itself.
- Switch chart types — view the same scores as bar, pie and histogram. As a pie they become "each score's share of the total", which is not what the data means.