Sea stack plots: Replacing bar charts with histograms

Area of Science:

• Data Visualization
• Statistical Graphics
• Scientific Communication

Background:

• Graphs enhance data comprehension but poor design can mislead readers.
• Existing plot types (e.g., boxplots, density plots) have limitations with large or unevenly distributed datasets.
• Uninformative plots like bar charts and dot and whisker plots are prevalent in ecology and conservation literature.

Purpose of the Study:

• To introduce a new plot type, the sea stack plot, for accurate and efficient representation of large univariate datasets.
• To compare the effectiveness of sea stack plots against commonly used plot types.
• To analyze current data visualization practices in ecology and conservation journals.

Main Methods:

• Comparison of five common plot types (dot and whisker, boxplots, density, univariate scatter, dot plots) for data distribution representation.
• Analysis of figures in four ecology and conservation journals to identify prevalent visualization methods.
• Development of the sea stack plot and an accompanying R package ('seastackplot').

Main Results:

• Commonly assessed plot types are difficult to read with large sample sizes or can misrepresent data distributions.
• Bar charts and dot and whisker plots constitute 60% of univariate data panels for comparison in analyzed journals.
• 16% of panels combined plot types (e.g., boxplots with density plots) for improved data display, indicating a need for better visualization tools.

Conclusions:

• Sea stack plots offer an improved method for visualizing large and unevenly distributed univariate data, overcoming limitations of existing plots.
• The development of user-friendly and accurate plot types like sea stack plots is needed to meet the demand for effective data visualization in scientific fields.
• The 'seastackplot' R package provides a tool for researchers to implement this new visualization technique.