6 Data Visualization

Choose a chart for the question

A chart is useful when it makes a pattern easier to see and understand than a table. Choose a chart according to the question you want to answer, rather than simply choosing a familiar-looking chart. Before creating one, write down the main point you want readers to notice.

• Comparing categories: A bar chart compares values across categories, such as the number of books borrowed from several libraries. A sorted horizontal bar chart can make long category names easier to scan.

• Showing change over time: A line chart shows how a value changes over time, such as monthly temperatures across a year.

• Showing a numerical distribution: A histogram shows frequencies across value ranges, while a box plot summarizes the median, quartiles, and possible outliers. Either can help explore the spread of exam scores.

• Showing whether two numerical variables move together: A scatter plot compares two numerical variables, such as hours studied and exam scores. Each point represents an observation.

• Showing parts of a total: A stacked bar chart can show how a total is divided among categories. A pie chart may also work, especially when there are only a few clearly distinct parts, such as categories in a budget.

These are common choices, not rigid rules. If a chart is difficult to read or makes values difficult to compare, try another form.

Read charts carefully

Begin by checking the chart’s title, axes, units, legend, and notes. Establish what each observation represents and which place and time period the chart covers. Check whether values are counts, percentages, rates, or averages.

A percentage needs a denominator. For example, $20\%$ of $50$ people and $20\%$ of $5{,}000$ people represent different numbers of people.

Compare like with like: check that categories use the same units and that groups or time periods are defined consistently. Pay attention to the axis range as well. Bar charts comparing magnitudes should generally start at zero because bar length encodes value. A line chart may use a narrower range to reveal small changes, but its axis should be clear so the change is not overstated.

Do not rely only on averages. A mean can hide variation or extreme values, while a histogram or box plot may show whether data cluster, spread out, or include unusual observations. A scatter plot can show that two variables vary together, but that relationship alone does not establish that one causes the other; other factors may explain the pattern.

Estimates based on samples are not exact. When relevant, show an uncertainty interval or explain limitations. Avoid presenting small differences as certain when they may fall within the margin of error.

Communicate patterns accurately

Make the chart’s message clear without claiming more than the data show. Use a concise headline to describe the main pattern, and use a subtitle or caption to identify the measure, population or geography, and time period. Label axes with units, identify series clearly, and cite the specific data source.

A short annotation can direct attention to an important change, but it should not imply a cause unless the analysis supports that explanation. Keep the design simple by removing decoration that does not help readers understand the data. Use color consistently and accessibly, and do not rely on color alone to distinguish important groups; labels, patterns, or distinct line styles can help.

Provide a text description or data table when needed so that people who cannot see the chart can still access its message or values.

For example, suppose a chart shows that a store’s monthly sales rose from $\$10{,}000$ to $\$12{,}000$. A clear headline could say “Sales increased $20\%$ from January to February,” with a caption specifying the store and year. If the vertical axis starts at $\$9{,}500$, bars may make the increase appear much larger than it is. A zero baseline or a clearly labeled line chart makes the scale easier to judge; the chart shows an increase, but not what caused it.