The goal of data visualization is Communication, not Decoration. Common pitfalls: excessive use of 3D effects (distorts data proportions); dual-axis charts (implies correlations that don’t exist); truncated Y-axes creating exaggerated growth impressions (a Y-axis starting at 90% makes a 5% increase look like a visual “doubling”); rainbow color schemes (completely fails in color printing or for color-blind readers); too many pie chart slices (human eyes can’t accurately perceive angle differences between more than 5 segments).
## Chart Type Selection Framework
The right chart type depends on what type of data relationship you want to show:
**Comparison** → **Bar/Column charts**: ideal for comparing numerical values across multiple categories; keep Y-axis starting at 0; sort categories meaningfully (by size or time).
**Trend/Change over time** → **Line charts**: shows time series data, time on X-axis, line continuity conveys the change process; avoid scatter plots when more than 20 points (lines are clearer).
**Part-to-Whole** → **Stacked bar charts** or **100% stacked bar charts**: easier to precisely compare parts than pie charts; use pie charts only when categories ≤5 and primarily showing the largest category.
**Distribution** → **Histogram** or **Box plot**: shows numerical data distribution shape, skewness, and outliers; box plots are particularly suited for comparing distributions across multiple groups.
**Correlation** → **Scatter Plot**: shows the relationship between two numerical variables; can add trend lines (linear regression lines) to reinforce correlation visual perception.
## Tufte’s Minimalism Principle
Edward Tufte proposed the “Data-Ink Ratio” principle in “The Visual Display of Quantitative Information”: maximize each “ink drop” (pixel) to display data itself rather than decorative elements. Practice: remove grid lines (or replace heavy black grids with very thin gray lines); remove borders around data points; reduce legends (directly label series names on the chart); remove chart background colors; replace legends with data labels.
See [Workplace Data Literacy](https://sunqi.org/data-literacy-workplace-en/), [Datawrapper chart design tool](https://www.datawrapper.de/), and [Storytelling with Data by Cole Nussbaumer Knaflic](https://www.storytellingwithdata.com/).
