Data visualization is frequently treated as a neutral medium, a method for seeing through to the significance of underlying data. However, the act of creating a visual representation requires a series of subjective decisions that can be as significant as the data itself. Much like photography, visualization cannot be perfectly neutral because it requires constant choices about framing: what to include, what to exclude, and which specific aspects of a subject to highlight. Even when the goal is to capture a subject 'neutrally,' the designer must decide on the charge range, the numeric scale, and the visual hierarchy, all of which guide the viewer toward a specific conclusion.
Technical adjustments in a plot can serve as a double-edged sword between clarity and bias. For instance, adjusting axis ranges, choosing specific color palettes, or applying smoothing to a trend line can be used to make a specific point more legible to an audience. While these choices are often necessary to make a figure useful—especially in scientific contexts where a figure's purpose is to highlight a specific finding—they also mean the analyst is actively shaping the viewer's perception. A visualization that omits the broader context to focus on a minor fluctuation, for example, may be technically accurate in its data points but subjective in its presentation of importance.
Beyond the visual mechanics, linguistic and structural choices further influence interpretation. A designer might present the exact same figures but change the perception of the data by simply renaming a category—such as switching from 'The State' to 'Government'—or by choosing how to group data points. These choices do not change the raw numbers, but they do change what a reader walks away believing. Furthermore, there is a tension between providing a 'clean' visual and the risk of omitting necessary context. Some argue that a visualization may contain objective information that is simply misinterpreted by an uninitiated audience; in this view, the lack of neutrality is not a failure of the visualization itself, but a gap in the user's knowledge required to decode the objective information provided.
Ultimately, the quest for a perfectly neutral visualization faces a fundamental hurdle: the trade-off between clarity and completeness. A visualization that includes every possible data point and every nuance of context may become unreadable, while a visualization that is easy to digest must inevitably discard some information. Because every design choice involves a trade-off, the resulting visual is always a curated narrative rather than a raw, unfiltered reflection of reality. While some argue that pure mathematical functions can be represented without bias, most real-world data—such as stock prices or population trends—require a human-led curation that inherently favors one perspective over another.