The comparison at a glance
| Setting | Width 10 | Width 20 |
|---|---|---|
| Start | 40 | 40 |
| Bin count | 6 | 3 |
| Frequencies | 2, 3, 4, 5, 4, 2 | 5, 9, 6 |
| Total observations | 20 | 20 |
A width change is a data grouping change
The twenty fictional scores in our example use width 10 from a start of 40. Their counts are 2, 3, 4, 5, 4 and 2. With width 20 from the same start, adjacent counts combine into 5, 9 and 6. The observations have not changed, but the grouped description has become coarser.
Start and width work together
A width of ten beginning at 40 is not equivalent to a width of ten beginning at 35. Different boundary positions can place observations into different intervals and shift a visible peak. Report the start or full boundaries when reproducibility matters. The calculated table gives both endpoints and makes the inclusion convention explicit.
Watch for unstable stories
If a claimed cluster appears only under one carefully chosen width, inspect the raw data and alternative widths before emphasizing it. Very narrow bins in a small sample can create isolated spikes. Very wide bins can combine distinct groups. A useful choice communicates the relevant scale without pretending that a histogram is independent of its settings.
Use the controls predictably
In this maker an explicit positive bin width overrides the requested bin count. The final interval includes the maximum observation. Clearing width lets the requested count determine an equal width over the chosen range. Settings that would create more than fifty bins are rejected, so no observations disappear merely to satisfy a visual limit.
A related chart you can inspect
References and reproducibility
The worked arithmetic and teaching examples on this page are original. Background reading: NIST histogram background. Source code and reuse terms expose the implemented calculations.
Written by SupaMakers for ChartsAI · September 8, 2026.