Fixed numbers, different pictures
Use four artificial observations in order: 10, 11, 12 and 11 units. They represent no actual business or experiment. Display them once on a vertical range from zero to twenty and consider a narrower range from nine to thirteen, keeping the physical plotting area the same.
The values remain unchanged while the visible distance between the highest and lowest occupies a different share of the picture. It is easy to confuse the strength of an impression with the size of a difference. Correct numbers in a table do not remove the need to inspect the way they occupy a screen or page.
Chart type belongs to the decision
For bars, length from a baseline encodes magnitude, so official statistical guidance recommends a zero starting point. Starting these bars at nine would give the value ten a displayed length of one unit and twelve a length of three, although their actual magnitudes are not in that ratio.
A dot or line plot does not rely in the same way on the length of a filled shape from the axis. A narrower range can help reveal small changes when its scale is explicit and suits the question. A non-zero axis is not automatically deceptive; examine the visual encoding and the comparison requested of the reader.
Three cards retaining the example's specifications
The diagram summarises the original values, zero-based bars and dots within the narrower range. These are specification cards, not an axis-based quantitative chart. They retain what someone needs to rebuild the comparison: four observations in the same order, an unchanged plotting area and two declared vertical ranges.
Draw bars from zero to twenty and then dots from nine to thirteen. The choice raises a question: are we comparing magnitudes from a baseline or following changes between nearby observations? Select the scale for that purpose rather than to amplify a story already decided before examining the values.
What changing a scale cannot do
An altered axis neither improves the data nor explains the third observation's rise. Making a difference visually prominent does not establish that it matters for the task. Units, the purpose of the comparison and a criterion of meaningful difference are needed, not merely a description such as steep or rising.
An excessively wide range can also obscure a distinction worth examining. For a question about limited variation, explicitly scaled dots may be more useful than bars occupying much space while hiding the change. There need not be one compulsory picture for every question; explain what the selected view reveals and what it makes less apparent.
Publish a reproducible specification
Supply values, units, order, chart type and axis limits. When comparing pictures, retain the plotting area and other choices or state their differences. If scale, dimensions, colour and grouping all change together, it becomes difficult to identify which decision produced the impression under discussion.
Ask a reader for the numerical difference they understood, not only an emotional description. Published values let that reading be checked. We need not begin by accusing an author of bad intentions. We need display choices whose consequences can be examined and whose contribution to understanding is made clear.
Numbers and their represented relationship
Within the Unified Framework of Systemic Boundaries, an axis range sets a boundary on the visual space available for differences. It is not a boundary in the phenomenon and does not change the arithmetic relationship between values. The example shows representation mediating a reading of unchanged data.
After seeing a chart, ask what the scale makes clear and what becomes harder to judge. Retaining values and display specifications supports a picture chosen for understanding, without turning visual force into evidence of a large change or a cause that has not been investigated.



