A Taxonomy of Misleading Visualization
Misleading data visualization — using visual representations to create false or distorted impressions of data — ranges from the inadvertent (design errors that mislead without intent) to the deliberately deceptive (visual manipulations designed to support a predetermined conclusion). Both categories are ethically significant, but they require different responses: the inadvertent misleader needs education; the deliberate misleader requires accountability. A visualization practitioner who can identify both forms is equipped both to produce honest work and to critically evaluate the visualizations they encounter in professional and public contexts.
The truncated y-axis (already covered in Module 101) is the most common misleading technique in bar charts. But misleading also occurs in the opposite direction: including irrelevant zero baselines when they are not meaningful. If all data values range from 94 to 100 and the message is the variation within that range, forcing a y-axis from 0 to 100 compresses the data into the top 6% of the chart and visually eliminates the variation. The ethical question is not 'should the axis start at zero' in the abstract — it is 'does this axis choice accurately represent the data's meaningful range for the question being answered?'
Cherrypicking is the selective presentation of data that supports a predetermined conclusion while omitting data that contradicts it. A chart showing only the years in which a trend is positive, or showing only the favorable subgroup in a clinical trial, or ending a time series precisely at a convenient high point — all are forms of cherrypicking. The visualization practitioner's responsibility is to represent the complete picture, or if the chart must be limited in scope, to explicitly acknowledge what is excluded and why.
False causation visualization occurs when correlation between two variables is presented in a way that strongly implies causation. Presenting two time-series that happen to move together (like ice cream sales and drowning rates in summer) in a way that visually suggests the first causes the second — particularly with connection arrows, causal language in titles, or juxtaposition that implies mechanism — is a form of visual misrepresentation even if the underlying data is accurate. Visualization practitioners should use causal language ('X causes Y') only when causation has been established through appropriate methods.