Dot Plot

What a dot plot shows, how it differs from a group comparison mean figure and a strip plot, and when to use it.

A dot plot displays the mean of each group as a single point, with an error bar extending above and below to indicate the spread or uncertainty around that mean. It is a compact summary figure — it does not show the underlying distribution or individual observations.

This page is about a summary figure, not a formal statistical result on its own. A dot plot does not produce a p-value or effect size by itself. Its role is to show a compact mean-based summary that can be read alongside the corresponding statistical test and the underlying data review.

What the figure shows

Each dot represents the group mean. The error bar shows one of the following, depending on your analysis settings:

  • Standard deviation (SD) — the spread of values around the mean
  • Standard error of the mean (SEM) — the uncertainty in the mean estimate
  • 95% confidence interval — the range within which the true mean is estimated to fall

Licklider calculates the displayed mean and error bar from the visible numeric data, but those error bar types are not interchangeable. SD describes sample spread. SEM describes how precisely the mean was estimated. A 95% confidence interval is the most direct summary of inferential uncertainty around the mean.

Because these quantities mean different things, the same dot plot can support very different claims depending on which error bar is shown. Error bar overlap alone should not be treated as a formal test of significance or non-significance.

Dot plot vs Group Comparison Mean and SEM Both figures show mean and error bars per group. The visual presentation differs: a dot plot uses a scatter-style dot for the mean, while the Group Comparison Mean figure typically uses bar chart geometry. For most purposes they communicate the same information.

The dot style exists because it is a cleaner way to show the same summary without implying that the filled area below the mean carries meaning. For many continuous outcomes, the point and its interval are the meaningful elements, not the bar area.

Dot plot vs Strip plot A strip plot shows every individual observation. A dot plot shows only the summary (mean + error bar). Use a dot plot when you want to communicate the group summary without showing raw data; use a strip plot when showing every data point matters.

Dot plot vs Box plot A box plot shows the median, interquartile range, and the spread of individual values. A dot plot shows only the mean and a chosen measure of spread or uncertainty. Use a box plot when the distribution shape or outliers are relevant; use a dot plot when a simple mean comparison is sufficient.

That separation is intentional. A dot plot is useful when brevity matters and the mean is the summary you want readers to carry away. A box plot or strip plot is better when you need readers to inspect how the values are distributed within each group.

When to use a dot plot

Dot plots are appropriate when:

  • A clean visual summary of group means is needed
  • You are comparing several groups and do not need to show the underlying data distribution
  • The figure will be used alongside a statistical test result rather than as the primary evidence

Licklider does not determine from the dot plot alone whether the mean is the most honest summary, whether a few outliers dominate the average, whether the groups are skewed or multimodal, or whether the rows are independent. Those questions may require a strip plot, box plot, violin plot, or the linked quality checks.

That limitation matters because very different raw data patterns can produce similar means and similar error bars. A dot plot can therefore look stable even when the underlying data are sparse, skewed, or driven by a small number of influential points.

Design Rationale & References

This page follows a simple rule: use a dot plot when you want a compact mean-based summary, but do not ask the figure to show more than it contains.

That is why Licklider presents the mean as a point with an explicitly chosen error bar type, rather than pretending that a summary-only figure also reveals raw distribution shape. It is also why the dot plot is positioned as a companion to statistical results and data-review figures, not as a replacement for them.

The distinction among SD, SEM, and 95% confidence intervals matters because each answers a different question about the data. If the error bar type is not stated clearly, readers can mistake sample spread for inferential precision or vice versa.

  1. Cumming, G., Fidler, F., & Vaux, D. L. (2007). Error bars in experimental biology. The Journal of Cell Biology, 177(1), 7-11. https://doi.org/10.1083/jcb.200611141
  2. Weissgerber, T. L., Milic, N. M., Winham, S. J., & Garovic, V. D. (2015). Beyond bar and line graphs: time for a new data presentation paradigm. PLOS Biology, 13(4), e1002128. https://doi.org/10.1371/journal.pbio.1002128

These references support two key points reflected in Licklider's design: error bar type changes the claim the figure makes, and summary-only graphics can hide important features of the raw data. The dot plot is therefore best used as a compact summary, not as a standalone guarantee that the underlying data are well behaved.

What this page does not cover