Beautiful data is data a viewer can read at a glance without being misled. Design serves the number, never the other way around.
The quickest way to lie with a chart is to leave out the baseline or to start the axis someplace flattering. Truncated scales make small changes look huge. The viewer feels the drama without knowing the trick. That is what makes it dangerous.
Start from the honest version of the chart, then make that version clear and attractive. If the honest chart is boring, the finding is probably modest, and that is worth saying. Beauty is not the goal by itself. Clarity that happens to look good is the goal.
Decide what single comparison the viewer should take away, then build the chart around it. A chart that tries to show everything shows nothing. Make one point and make it obvious. The rest of the data can live in your notes.
Highlight the line or bar that matters and mute the rest. Color is a spotlight, and a spotlight pointed at everything is just a bright room. Use contrast to direct the eye. The viewer should know where to look without being told.
Label directly on the chart instead of forcing the viewer to match a legend color to a line. Every glance saved is attention kept. Legends cost the viewer a small mental step each time. Remove that step and the chart reads instantly.
Color is a spotlight. A spotlight pointed at everything is just a bright room.
Bar charts should start at zero, because bar length carries the comparison. Line charts can zoom in when the question is about change over time, but mark the cutoff clearly. The rule changes with the form. Know which one you are using and why.
Keep the scale consistent across charts you place side by side. A silent rescale between two panels is one of the easiest mistakes to make and the hardest for a viewer to catch. Consistency is a form of honesty. Break it only with a visible note.
When you crop an axis, say so on screen. A quick note protects the viewer and your own credibility at the same time. The note costs one line of text. It buys back the trust the crop could have damaged.
A build that shows data appearing step by step can help a viewer follow the growth. A build that hides the scale until the end can trick the eye instead. The difference is whether the frame of reference is visible. Show the axes and the animation teaches.
Keep the axes visible throughout the animation so the viewer always knows the frame of reference. Motion without a stable frame reads as drama, not data. A moving point means nothing without fixed lines. Hold the reference steady while the data moves.
Let the viewer sit with the finished chart for a moment. The last frame is the one they will remember, so give it room to land. A held chart becomes a mental image. A chart that vanishes is forgotten before the next sentence.
Most numbers carry a range, and hiding it makes the figure look more exact than it is. A shaded band or an error bar tells the truth without clutter. The band is often the most important part. It shows how much the data can wobble.
When the sample is small, say so near the chart. A single number from a handful of cases should not look like a population estimate. Scale claims to the size of the sample. A note is enough to keep the viewer calibrated.
Honest data design is not less beautiful. Done well, it makes the shape of what people know and what they are guessing visible in the same picture. That shape is the real story. A chart that shows it is doing its job.
Good science content turns a hard idea into something a curious newcomer can hold. This guide shows how to explain complexity without flattening the truth.
Animation can carry a science idea that words alone cannot. This guide shows how to plan visuals for your science content so they clarify, not decorate.
Citing a study is easy to botch. This guide shows how to source research in your science content so claims rest on evidence you actually read.
Analyze a channel, extract its structural DNA, and generate scripts engineered from patterns that actually work.
Analyze a Channel