Why Use a Bar Chart?
A frequency table lists how many observations fall in each category. A bar chart turns those counts into a visual comparison: each category gets a separate bar, and the bar’s height shows its count. This makes it easier to see which categories are most common, which are less common, and how large the count differences are.
Bar charts are used for categorical data. The categories might be favorite activities, transportation choices, or types of appointment. A bar chart displays the observed counts; by itself, it does not explain why a category is common or establish that the pattern applies to a wider population.
Check the Counts Before Drawing
A bar chart can only represent its frequency table accurately. Before plotting, check that the categories are defined consistently and that each observation is counted once in exactly one category. The categories should cover the observations being described. If the table lists counts \(f_1, f_2, \ldots, f_k\), their sum is the total number of observations represented:
Here, \(k\) is the number of categories and \(f_i\) is the count for category \(i\). This total provides a useful check: if the counts should include everyone surveyed but sum to fewer observations than were surveyed, the table needs investigation before a chart is drawn.
Constructing a Count Bar Chart
Start with the frequency table, then translate each count into a bar. The following steps describe a vertical chart; a horizontal version follows the same principle, with bar length rather than height representing the count.
Use the categories in the frequency table, keeping their labels clear and consistent with the data.
Put zero at the bottom of the count axis. Choose evenly spaced values that reach at least the largest count, and use that same scale for every bar.
Draw one bar above each category. Use equal widths and leave visible gaps between bars to show that the categories are distinct rather than intervals on a continuous scale.
Make each bar’s top correspond to its category count. Give the chart a title or clear description, label the category axis, and label the other axis as a count of observations.
Categories do not need to be numerical for this chart to work. They can be placed in any readable order, as long as every bar remains associated with its category label. The height—not the bar’s horizontal position—encodes the count.
Worked Example: Favorite Activities
Suppose a hypothetical class of 20 students each names one favorite activity. The recorded counts are reading, 8; sports, 5; music, 4; and drawing, 3. Construct a count bar chart and describe its main features.
Check the total. The category counts add to \(8+5+4+3=20\), matching the number of students. The chart can therefore represent all 20 responses, provided each student’s response was assigned to exactly one of the four categories.
Choose a scale. The largest count is 8, so a scale from 0 to 8 in steps of 1 can show every count exactly. In this compact text rendering, each filled cell represents one student, and the category columns are the bars:
| Count level | Reading | Sports | Music | Drawing |
|---|---|---|---|---|
| 8 | █ | |||
| 7 | █ | |||
| 6 | █ | |||
| 5 | █ | █ | ||
| 4 | █ | █ | █ | |
| 3 | █ | █ | █ | █ |
| 2 | █ | █ | █ | █ |
| 1 | █ | █ | █ | █ |
| 0 |
The reading bar reaches 8, sports reaches 5, music reaches 4, and drawing reaches 3. Reading is the most frequently named activity, and drawing is the least frequently named. The gap in count between reading and sports is \(8-5=3\) students. These are descriptions of this hypothetical class’s responses, not claims about students generally.
Why Height Represents Count
A bar chart turns a numerical count into a visual length. If every unit on the vertical axis stands for one observation, a bar of height 5 represents five observations; a bar of height 8 represents eight. Because both bars use the same scale and begin at zero, their height difference corresponds to their count difference.
More generally, suppose the vertical scale uses \(u\) units of height per observation, where \(u\) is a fixed positive amount for the whole chart. A category with count \(f_i\) has height \(h_i=u f_i\). For two categories \(i\) and \(j\), subtracting their heights gives:
Since \(u\) is the same for both categories, a difference in bar heights represents the corresponding difference in counts, expressed in the chart’s height units. This justification depends on using one consistent scale. If the scale changes between bars, height differences no longer have this interpretation.
Reading Heights, Gaps, and Labels
Read a bar chart by connecting each bar to its category label, then matching the top of the bar to the count axis. The height answers “how many observations are in this category?” Comparing two heights answers “how many more or fewer observations are in one category than another?” The labels answer “which categories are being compared?”
A gap between bars is a visual separation between categories. It is not a missing count or an unobserved value between categories. Bar charts use gaps because categorical values are distinct groups, not neighboring positions on a continuous numerical scale. The vertical difference between bar tops, by contrast, corresponds to the difference in their counts when the same scale is used.
A tie occurs when two bars reach the same height: the categories have equal counts in the data shown. A bar with height zero has a category label but no visible bar above the baseline. Including such a category can be informative if it is part of the set of categories being displayed.
Worked Example: Transportation Choices
Imagine that 30 hypothetical commuters each report one usual way of getting to work. A frequency table gives walking, 4; bicycle, 6; bus, 8; and car, 12. Use the counts to describe what a chart would show.
Check the table. The sum is \(4+6+8+12=30\), the number of commuters. Each response is assigned to one transportation category, so these counts account for the full set.
Read the heights. With a vertical count scale from 0 to 12, the car bar reaches 12 and is tallest. The bus bar reaches 8, the bicycle bar reaches 6, and the walking bar reaches 4. The difference between car and bus counts is \(12-8=4\) commuters; the difference between bicycle and walking counts is \(6-4=2\) commuters.
Interpret the gaps correctly. Separate bars distinguish the four transportation categories. The gap between the walking and bicycle bars does not represent a fifth transportation choice. The bars summarize the reported choices of these 30 hypothetical commuters; a count chart alone does not establish the pattern among all commuters.
Worked Example: Equal Counts and a Zero Count
In a hypothetical survey of 15 students, the counts for favorite activities are music, 6; reading, 6; sports, 3; and gardening, 0. Explain how ties and a zero count appear in a chart.
Check the total. Here \(6+6+3+0=15\), so the counts sum to the number surveyed.
Plot the counts. Use one scale from 0 to 6. The music and reading bars both reach 6, so the chart shows a tie for the largest count. The sports bar reaches 3. Gardening has a category label at the baseline but no bar rising above it, because its count is zero.
State the conclusion in context. In this hypothetical group, music and reading are tied as the most frequently named activities, with six students each; no student named gardening. The empty gardening bar does not mean that gardening was excluded from the survey: its label and zero count communicate that it was an available category with no recorded responses.
Common Mistakes
- Using unequal count scales. If one unit has a different meaning for different bars, their heights cannot be compared fairly. Use one common count scale throughout.
- Starting the count axis above zero. This can exaggerate visible height differences. For a straightforward count bar chart, include zero as the baseline.
- Confusing a bar gap with a count gap. Space between bars marks distinct categories. Compare the bar tops, using the count axis, to find count differences.
- Leaving out category labels or the count-axis label. Without labels, a reader may not know which group a bar represents or whether the vertical scale shows counts. Label both axes and make the variable clear.
- Treating unequal bar widths as if they change the height scale. When all bars share one count scale, their heights still represent comparable counts even if their widths differ. But different widths can make bar areas misleading to compare. Equal widths are preferable, so the visual emphasis stays on height rather than area.
- Claiming more than the count chart shows. A bar height reports an observed count in the data represented. It does not, by itself, explain the reason for the count or justify a conclusion about people who were not observed.
Check Your Understanding
Use the definitions and examples above to answer the following questions.
- A hypothetical group of 24 people chooses one favorite fruit. The counts are apples, 9; bananas, 7; oranges, 5; and pears, 3. What should the total of the bar heights be, and why?
- In a count bar chart, one bar reaches 11 and another reaches 7. What is the difference in the category counts, assuming both bars use the same scale?
- Why is the space between two category bars not a category with count zero?
- Two bars have the same height. What does that show about the categories in the data represented?
- If bars have different widths but share one count scale, can their heights still be compared? What visual comparison might their unequal widths make misleading?