This entry is part of the Chart Index, the reference library for the Chart Design Field Guide.

The dot plot is Cleveland's refinement of the bar chart, and the form he argued was perceptually superior for nearly every ranking task. A single dot per category, sitting on a horizontal value axis, with a thin reference line connecting the dot to the category label. No bars, no stacking, no fill. Just position.

Cleveland's case was empirical. His 1984 experiments with McGill ranked perceptual tasks: position along a common scale was the most accurate channel. A dot plot uses only that channel; a bar chart adds length, which is slightly less accurate. For small differences and dense displays, the dot plot wins by a measurable margin.

What it is

A dot plot maps a categorical variable to position along a category axis (typically rows) and a quantitative variable to position along a value axis. Each data point is rendered as a single dot. Categories are usually sorted by value. Optionally, a thin guide line (dotted or solid) anchors each dot to its row label.

The hero chart above ranks fourteen API endpoints by median response time. Because there are no bars, the axis does not need to start at zero — though here it does, since the slowest endpoint is forty times the fastest and the range is the point. Read the dots as a curve: a steep top of three slow endpoints, a shallow middle where six endpoints sit within 80 milliseconds of one another, and a floor of health checks and token refreshes. The eye can compare the middle six with a precision that bars of nearly equal length would not have allowed. Randomise the data and the dots slide along their rows.

When to use it

Dot plots are the right choice when:

  • You have many categories (15–50) and a bar chart would feel too heavy.
  • The values are close together and you need to read small differences.
  • The reader's question is "rank these" or "which is largest / smallest?"
  • You want two values per category — a before / after dumbbell is a natural extension.
  • The chart is editorial or analytical rather than presentation-graphic.

When not to use it

  • Few categories. Three or four dots look sparse and underweight. A bar chart provides more visual substance.
  • Stacked or grouped composition. Dot plots resist stacking. For composition, use stacked bars or treemaps.
  • Audiences unfamiliar with the form. Business stakeholders may default to expecting bars; dot plots can read as incomplete to readers who have not encountered them.
  • Values that span orders of magnitude. A linear axis with one value at 10 and another at 10,000 puts most dots in a thin band. Use a log axis or a different form.

Design principles

Range the axis sensibly

Unlike bar charts, dot plots do not require a zero baseline — there is no length to encode. Use the data range with modest padding. This often reveals differences that a zero-anchored bar chart compresses. The pair below holds nine regional offices whose satisfaction scores all sit between 73 and 87.

Nine close values as bars, from zero
Bars must start at zero, so the fourteen-point spread between the best and worst office is squeezed into the last sixth of each bar. The ranking is technically visible; the sizes of the gaps are not.
The same nine values as dots, on a ranged axis
Position needs no zero, so the axis runs from 70 to 90 and the same fourteen points now fill the width of the chart. The one-point gaps in the middle of the table are visible, and the three-point jump from Melbourne to Sydney is obviously the largest.

Make the dot deliberate

Use a 6–10 pixel solid filled circle. Filled, not hollow. A solid dot reads as a definite value; a hollow ring reads as an uncertain or sample marker. Cleveland used solid dots for exactly this reason.

Add a thin guide line

The original Cleveland dot plot includes a faint horizontal dotted line from the category label to the dot. The guide helps the reader's eye track from the long row label to the dot's position. Without it, dots can feel adrift in white space. Faint vertical gridlines at the axis ticks, as in every figure in this entry, do part of the same job.

Sort by value

Like lollipops, dot plots reward sorting. Random or alphabetical order produces noise. Sorted by value, the chart becomes a ranked list and the message — who leads, who lags — emerges.

The hero's endpoints in alphabetical order
The same fourteen medians, listed A to Z. Each endpoint can be looked up, but the shape of the distribution — three slow outliers, a dense middle, a fast floor — has disappeared. Compare the hero chart above, where the sorted dots draw it as a curve.

Use a second dot for paired comparisons

Two dots per row, joined by a thin line, become a dumbbell plot — a powerful way to show before/after or target/actual for many categories at once. The line shows the gap; the dots show the endpoints. This is where the dot plot most clearly beats the grouped bar: with two bars per category the reader compares heights and subtracts, whereas two dots on one row show the difference as a length.

Before and after as a grouped bar
Two bars per endpoint. Every pair is a separate subtraction, and because the bars start at zero the changes — some of them a third of the value — read as small differences in the height of tall columns.
The same before and after as a dumbbell
Grey dot before, blue dot after, a connector between them. The connector's length is the improvement, so the reader sees at once that search and analytics gained most, that products gained proportionally most, and that invoices got slightly worse — the one row where the blue dot sits to the right of the grey.

Direct-label dots when precision matters

A dot plot with an axis gives approximate values; a dot plot with values written next to each dot gives exact ones. The eye reads position for rank and the number for precision.

Avoid heavy gridlines

Like all dense displays, a dot plot benefits from quiet scaffolding. A single axis at the bottom, a few tick marks, dotted guide lines per row — and nothing else.

Anatomy

The composition of a dot plot
An anatomical guide

The dot plot is structurally minimal: an axis, a set of dots, optional guide lines. Cleveland's bare-bones display in its purest form.

  • Bar chart — the older alternative; better for audiences unfamiliar with dot plots and for charts where visual weight is desired.
  • Lollipop chart — adds a line from baseline to dot. Compromise between bar and dot.
  • Dumbbell / range plot — two dots per row, joined. For from–to comparisons.
  • Forest plot — dot with horizontal error bars. Used for confidence intervals in meta-analysis.
  • Strip plot — many dots per category, showing the distribution. Different question.

Reading list

  • Cleveland, W. (1985). The Elements of Graphing Data. The original argument for the dot plot.
  • Cleveland, W. & McGill, R. (1984). Graphical Perception. The empirical foundation.
  • Few, S. (2012). Show Me the Numbers. Practical application of dot plots in business charts.