Statistics Graphs and Charts: Types and How to Choose

Charts & pivot tables · Updated November 22, 2026

Statistics graphs and charts are visual displays of data that make patterns, comparisons, distributions, and relationships easier to see. The best choice depends on what the data represent: categories, measurements over time, or paired numerical values; common options include bar charts, histograms, line charts, scatterplots, and box plots.

What statistical graphs and charts show

A table gives exact values, while a visual display makes the overall shape of those values easier to grasp. A chart can reveal which category is largest, whether a measure is rising, how widely observations vary, or whether two quantities tend to move together. It can also expose unusual values or data-entry problems that are easy to miss in a long list.

In everyday language, graph and chart are often used interchangeably. Some people use graph for displays of mathematical relationships and chart for broader data summaries, but there is no universal distinction that applies to every statistical context. More important than the label is whether the display accurately represents the data and makes the intended comparison understandable.

A visualization does not replace the data or statistical analysis. It summarizes selected features, and its design can affect what viewers notice. A chart of average salaries, for example, may hide large differences within each job group. Reading the chart alongside its labels, definitions, sample size, and source gives it useful context.

How to choose the right chart

Start by asking what question the display should answer. Then identify the type of variable being shown. Categories such as region or product are different from numerical measurements such as height or revenue. Dates can be treated as an ordered sequence, while a pair of numerical variables can be examined for a relationship. The same dataset may support several valid charts, but each emphasizes a different feature.

  • Compare categories: use bars or dots to compare values for groups such as departments or survey responses.
  • Show change over time: use a line chart when the order of dates matters and the observations form a time sequence.
  • Show a numerical distribution: use a histogram, dot plot, or box plot to examine spread, concentration, and unusual values.
  • Explore a relationship: use a scatterplot for paired numerical measurements, such as age and reading time.
  • Show parts of a whole: consider a stacked bar or, for a small number of clear categories, a pie chart.

Think about the audience and the amount of data, too. A chart with dozens of categories may be technically correct but difficult to read. For a quick comparison, sorting bars from largest to smallest can help; for a time series, keeping dates in chronological order is usually essential.

Common types and when they work

  • Bar and column charts: These use separated bars to compare categories. Bar charts usually run horizontally, while column charts run vertically. They work well for counts, totals, or averages by group. Starting the value axis at zero is generally important because bar length encodes magnitude.
  • Histograms: These display the distribution of one numerical variable by grouping measurements into adjacent intervals, called bins. Unlike a bar chart, a histogram’s bars touch because the intervals represent a continuous numerical scale. The choice of bin width can change the apparent shape, so it should be stated or selected thoughtfully.
  • Line charts: These connect values in a meaningful order, commonly time. A line can make a trend easier to see, but connecting unrelated categories can suggest continuity that does not exist. If dates are irregularly spaced, a chart that treats them as equally spaced categories can misrepresent the gaps.
  • Scatterplots: Each point represents a pair of numerical values. A scatterplot can show direction, strength, clusters, and unusual observations in an association. A pattern does not by itself establish that one variable causes the other.
  • Box plots: These summarize a distribution using the median, quartiles, and whiskers, with some conventions marking potential outliers separately. They are useful for comparing distributions across groups, but they do not show every individual value and may be unfamiliar to some audiences.
  • Pie charts and stacked charts: A pie chart shows proportions of a whole, but differences between similarly sized slices are hard to judge. Stacked bars also show components, and can be easier to compare when there are several groups. Too many segments or a large number of groups makes either display harder to interpret.

Understanding distributions and variation

Many statistical questions concern more than a typical value. A distribution describes how observations are spread across possible values. Its shape may be roughly symmetric, skewed toward one end, concentrated in several clusters, or include unusually distant observations. A histogram makes these features visible, while a box plot provides a compact summary useful for comparing groups.

For example, suppose two classes have the same median test score. One class may have scores tightly grouped near the middle, while the other has a much wider range. A chart showing only the medians would make the classes appear more alike than they are. A box plot, dot plot, or histogram adds information about variation. When the dataset is small, plotting individual observations can be more informative than relying only on a summary.

Keep the distinction between counts and percentages clear. If two groups have different numbers of people, raw counts may make the larger group look more prominent even when the proportion with a particular response is lower. Comparing percentages or rates may answer the question better, provided the denominator is clearly identified.

Creating a chart in Excel

In Microsoft Excel for Microsoft 365 on Windows, place the data in a clear table with descriptive column headings. Select the relevant cells, then choose Insert > Recommended Charts. Review the suggested options, choose one that fits the question, and select OK. You can also choose a chart type directly from the chart options on the Insert tab. Suggestions are a starting point, not a guarantee that the chart is statistically appropriate.

After insertion, select the chart to access the Chart Design and Format tabs. Use Chart Design > Add Chart Element to add or adjust items such as axis titles, data labels, and a legend. Give the chart a title that states what is measured and, when helpful, the group or period covered. Check that units, dates, and category names are included. On Excel for Mac, the ribbon has similar chart-design controls, although some menus and controls may be arranged differently.

For a histogram, select the numerical observations and choose Insert > Insert Statistic Chart > Histogram when that option is available in the Microsoft 365 ribbon. Inspect the bin settings: changing the interval width can alter how much detail is visible. For a scatterplot, select the paired numerical columns and choose Insert > Insert Scatter (X, Y) or Bubble Chart. Confirm that the intended variable is on each axis; selecting columns with unrelated or mismatched rows can produce a misleading result.

Design choices that improve clarity

A useful chart makes comparisons easy without adding visual distractions. Use plain backgrounds, readable text, and a restrained color palette. Apply color consistently: if blue represents one group in one panel, it should not represent a different group in another. Do not rely on color alone to communicate meaning, because readers may have color-vision differences or view a printout in grayscale. Labels, patterns, and clear legends can help.

Choose axis scales deliberately. A bar chart with a vertical axis that starts above zero can make small differences look dramatic because bar lengths no longer reflect values from a common baseline. Line charts may sometimes use a narrower scale to show changes, but the axis should be labeled and the choice should not exaggerate the story. If two charts are meant to be compared, consistent scales make comparisons more trustworthy.

Also consider what happens when values are missing or rounded. A line connecting across a missing time point may imply a continuous record where none exists. Rounding values can make labels easier to read, but excessive rounding may conceal meaningful differences. Include units such as dollars, kilograms, or percentage points, and specify the time period when it matters.

Common problems and misleading displays

Several design choices can distort a viewer’s impression even when the underlying numbers are correct. A truncated bar axis, an unclear denominator, or an overloaded legend can change how a comparison is perceived. Three-dimensional effects are usually unhelpful for statistical displays because perspective can make areas and lengths difficult to judge. Decorative images scaled to represent quantities can also mislead if their area, rather than just their height, changes.

Be cautious about combining unlike quantities on one chart. A dual-axis chart may make two series appear to track closely simply because the axis limits were chosen to align their shapes. If both measures are important, separate panels with clearly labeled scales may be easier to interpret. Similarly, a correlation shown by a scatterplot is not proof of causation: other factors, selection effects, or coincidence may explain an observed association.

Finally, check the source and scope. A sample from one city, company, or online survey may not represent a wider population. A chart should identify the population or period it describes when that information is relevant. If uncertainty matters, such as when estimates come from a sample, show suitable uncertainty intervals or explain the limits of the data rather than presenting a point estimate as exact.

Reading a chart critically

Before accepting a visual conclusion, read the title and axis labels, check the units, and identify whether values are counts, percentages, averages, or another statistic. Look for the baseline and scale, and note whether the chart covers all observations or a selected subset. If it shows an average, ask whether the distribution or sample size might change the interpretation.

For example, a report might show that a town’s average commute increased from 24 to 27 minutes. That is a three-minute increase in the reported average, but it does not reveal whether most residents’ commutes changed slightly or a small number of very long commutes raised the average. A distribution chart, median, or additional breakdown could help answer that follow-up question. The right chart depends on the question, not on which format looks most impressive.

Frequently Asked Questions

What is the difference between a graph and a chart?
In common statistical use, the terms overlap and are often interchangeable. Some contexts reserve graph for a plotted mathematical relationship and chart for other visual summaries, but the distinction is not consistent. The chart type and data shown matter more than the name.

Which chart is best for comparing data?
For comparing values across categories, a bar or column chart is often a clear choice. A dot plot can also work well, particularly when showing averages or individual observations. The best option depends on how many categories there are and whether you need to show counts, proportions, or distributions.

When should I use a histogram instead of a bar chart?
Use a histogram to show the distribution of a numerical variable grouped into intervals, such as ages or measured weights. Use a bar chart to compare distinct categories, such as departments or product names. Histogram bars usually touch because neighboring bins represent neighboring ranges of a continuous scale.

Can a chart prove that one thing causes another?
No. A chart can reveal a pattern or association, but that alone does not prove causation. Establishing a causal claim requires appropriate study design and analysis, and may require considering other explanations or confounding factors.

Why can a graph be misleading if its numbers are correct?
Scale choices, truncated axes, unclear units, selective data, or confusing labels can influence the visual impression without changing the underlying values. Read the axis and source details, and check whether the display answers the question it appears to address.

Related

Try it: the Percentage Formula Generator: percent of, percent change, increase and decrease, with the formula.

More in Charts & pivot tables

Sources: Function behaviour and menu paths: Microsoft Support documentation for Excel. Method & sources

Spotted an error in this guide? Let us know.