Two college profiles can show the same trend with very different visual impact. One line appears almost flat; another rises steeply. Before deciding that one change is more important, read the axis limits and the underlying numbers. Chart design determines how much vertical space represents a given difference.

The UK Office for National Statistics recommends consistent scales for comparable charts and explains why a line chart can sometimes use a non-zero axis. This article develops a fictional comparison. None of the figures describes an actual college, and a non-zero axis is not automatically evidence of manipulation.

Keep the data fixed

Suppose two charts display the same fictional annual outcome rate, 80% in the first year and 84% in the second. Both charts have a plotting area 200 pixels high. Chart A uses a vertical scale from 0% to 100%. Chart B uses a vertical scale from 75% to 85%. Both use a linear scale and the same horizontal spacing.

Chart Vertical scale Four-point rise as share of scale Vertical distance on a 200-pixel plot
A 0 to 100% 4 / 100 = 4% 8 pixels
B 75 to 85% 4 / 10 = 40% 80 pixels

The underlying increase is four percentage points in each chart. Its relative increase from the initial value is 4 / 80 = 5%. Chart B allocates ten times as much vertical space to that difference because its displayed range is one tenth as wide. The pixel calculation applies only to the stipulated equal plot heights and linear axes.

The enlarged line can make a small change easier to see. It can also make the visual slope seem more dramatic if the scale is overlooked. The reader should neither dismiss the change solely because Chart A looks flat nor treat Chart B's slope as a tenfold increase in the underlying outcome.

Check what the line actually measures

Read the axis label, units and time period. A rate of 80 to 84 percent is not the same as a count of 80 to 84 students. Confirm the population, definition and data years before comparing institutions. If an outcome definition or reporting population changes, matching axis scales do not repair that underlying break.

Check the horizontal spacing as well. A change plotted over one year and the same change plotted over five years do not describe the same pace. If one chart skips years, inspect whether the data are genuinely annual observations or selected endpoints. A line between two points does not reveal the values that occurred between them.

Distinguish lines from bar lengths

In a line chart, position communicates the values and a shorter axis can highlight variation. Bar charts commonly encode quantity through the length of each bar from a baseline; cutting that baseline can distort apparent size ratios. Read labels and compare actual values instead of estimating a ratio from truncated bar lengths.

For example, 84 is 1.05 times 80. A bar drawn from a 75 baseline would have visible lengths of 9 and 5 units, a ratio of 1.8. That ratio describes the chosen drawing baseline, not the ratio between the original values. The distinction is especially relevant when comparing two apparently dramatic bars in different college profiles.

Ask for comparable evidence

Where two institutions publish separate charts, put the values into a shared table with the same metric and years. If uncertainty intervals or data-quality warnings are available, retain them beside the numbers; changing the axis cannot establish whether a difference is meaningful. Small cohorts and rounding can also affect how a visible change should be interpreted.

A useful check records the chart type, scale endpoints, units, years and source. Replotting is helpful if the underlying values are available, but copying a point from a small image is not equivalent to obtaining the data. When values cannot be verified, state that limitation. The chart is a starting point for a comparison, and the scale is part of the evidence needed to read it.