Empirereading Best Practices: How to Improve Your Reading Skills

A growing share of what we read is not prose but data: charts, graphs, tables, and dashboards. Yet most readers who would never accept a written claim without scrutiny will accept a chart at a glance, because a picture feels objective. It is not. A chart is an argument made with pixels, and it can mislead as easily as any sentence — often more, because the manipulation is invisible to the untrained eye. Improving your reading skills for the modern world means learning to read data as critically as you read words.

Want expert help putting this into practice? EmpireReading can guide you through it.

Read the axes before you read the shape

The first thing most people notice in a chart is the shape of the line or the height of the bars — the dramatic rise, the steep drop. This is exactly backwards. Before you register the shape, read the axes, because the axes determine what the shape means, and a striking shape drawn on deceptive axes tells you nothing.

Check what each axis measures and in what units. Above all, check where the vertical axis starts. A bar chart whose y-axis begins at zero shows honest proportions; one that begins partway up exaggerates small differences into dramatic ones. A change from 100 to 102 can be made to look like a doubling simply by starting the axis at 99. Also check the scale type: a logarithmic axis, where each step multiplies rather than adds, will flatten explosive growth and can be entirely appropriate or quietly misleading depending on context. Read the axes first, every time.

Ask what the numbers actually count

Related: Empirereading Tips and Strategies for Effective Reading.

A chart is only as good as the data feeding it, and the labels rarely tell the whole story. Before drawing any conclusion, ask what is really being counted and how. "Cases" might mean confirmed diagnoses or merely tests taken. "Average" might be the mean, which a few extreme values can drag around, or the median, which they cannot. The same word can hide very different measurements.

Be especially alert to counts presented without their context. A raw number of events is often meaningless without the base it came from. "Region A had 500 incidents, Region B had 300" tells you little if Region A has five times the population — per person, it may be far safer. Whenever you see an absolute count, ask "out of how many?" and whenever you see a rate, ask "based on what raw numbers?" The most common data deception is simply choosing whichever of these framings tells the more dramatic story.

Match the chart type to the question

Different chart types answer different questions, and a mismatch is often a red flag. Learn what each is for so you can tell when one has been chosen to obscure rather than reveal:

• Line charts show change over time — appropriate when the horizontal axis is genuinely time.
• Bar charts compare quantities across categories — the length of each bar should be readable against a zero baseline.
• Pie charts show parts of a single whole, and become unreadable past a few slices.
• Scatter plots show the relationship between two variables — the right tool for spotting correlation, and for seeing how noisy it is.

When a presenter uses a pie chart for things that do not sum to a whole, or a line chart across non-time categories, treat it as a signal to look harder. The wrong chart type is sometimes incompetence and sometimes misdirection, but either way it should lower your trust.

Distinguish the trend from the noise

See also: Empirereading Best Practices: Enhancing Accuracy and Efficiency.

Data wobbles. Any real-world measurement varies from point to point for reasons that have nothing to do with the underlying trend, and a skilled reader learns to see through this noise rather than reacting to every bump. A single point that breaks a pattern is far more likely to be random variation than a meaningful signal.

Be particularly wary of conclusions drawn from short stretches of data or from the endpoints of a series. Choosing a start date carefully — a technique sometimes called cherry-picking the baseline — lets someone show a rise or a fall from the very same dataset just by where they begin. Ask to see the longer series. If a claim rests on the change between two specific points, ask why those two, and what the full picture around them looks like. Genuine trends survive being viewed over a longer window; artefacts of noise do not.

Separate what the data shows from what it is claimed to show

The final and most important step is to distinguish the data itself from the interpretation laid over it. A chart may accurately show that two things rose together, while the caption claims one caused the other — a leap the data does not support. Correlation shown in a scatter plot is not causation, however tight the cloud of points.

Read the data and form your own reading of it before you read the accompanying claim, then compare. Ask what else could explain the pattern: a hidden third factor, a change in how the data was collected partway through, a reporting artefact. Ask what is not shown — the categories left off the chart, the timeframe cropped away, the uncertainty bars omitted. A chart that shows point estimates with no sense of their margin of error is presenting a false precision, and you should mentally add the uncertainty back in.

Build a data-reading checklist

These habits become reliable when you turn them into a routine you run on every chart. Before accepting what a visualisation tells you, ask: Where do the axes start and what are the units? What is actually being counted, and out of what base? Is this the right chart for the question? Am I reacting to a trend or to noise? And does the data genuinely support the claim attached to it, or only a weaker version of it?

Running that checklist takes under a minute and will spare you from a large fraction of the misleading data that fills news, marketing, and social media. The reader who can interrogate a chart as sharply as a sentence holds a real advantage in an age awash with data, and developing that fluency with numbers and visuals is a core part of the reading skill set that EmpireReading helps you build. Data does not speak for itself; it says whatever its presentation is designed to make it say, and your job is to hear what it actually shows.