Audit the Graph Before Describing It
A bar chart can look neat and still leave out information needed to understand it. Without a useful title, category labels, or a clear vertical axis, a reader may not know what the bars represent. Even when every bar has a value, an irregular or inconsistent scale can make comparisons unreliable.
In Describing a Bar Chart in Context, you learned to name the group and variable and report what the bars show. Before doing that, check whether the graph gives you enough information to do so accurately. This tutorial focuses on four common problems: missing titles, missing labels, missing units, and inconsistent scales. As discussed in Misleading Axes and Scales in Bar Charts, a graph’s scale can affect how a difference appears. Here, the goal is to spot the information or scale problem and say how to correct it.
Not every graph needs a long caption. But a reader should be able to tell what group is represented, what variable is displayed, what each axis means, and what the values measure. If any of those details are absent, do not fill them in by guessing. Explain what cannot be determined from the graph as drawn.
A Four-Part Graph Check
Use the following questions when checking a bar chart on an exam. They work together: a title provides context, category labels identify the bars, the vertical-axis label identifies the measure, and the scale shows how displayed positions correspond to values.
Does it identify the group or setting and the categorical variable? A title such as “Survey Results” may be too vague to tell the reader whose responses or which response variable are shown.
Can you tell what each bar represents? The labels should name the categories clearly. Unlabeled bars cannot be reliably connected to particular responses.
Does the axis say whether it shows counts, proportions, or percentages? If the values measure something else, such as distance or time, are the units named?
Are numerical intervals equal when the tick marks are equally spaced? If two graphs are meant to be compared, do they use the same scale for the same measure?
A good title is specific enough to supply useful context without trying to replace axis labels. For example, “Preferred Lunch Option Among 80 Surveyed Students” names the group and variable. The horizontal axis should still identify the lunch categories, and the vertical axis should still say whether the bars show counts or percentages.
Units are the meaning attached to a numerical value. A count is measured in individuals or items; a percentage is a share of a group; a duration might be measured in minutes. A vertical axis labeled only “Value” does not tell the reader which of these is intended. If the graph is missing units, do not assume that a bar height of 12 means 12 people rather than \(12\%\), 12 minutes, or another quantity.
For a numerical axis, equal distances between tick marks should represent equal numerical changes. If marks that are equally spaced are labeled 0, 5, 10, and 20, the intervals do not represent equal changes: the final jump is 10 while the earlier jumps are 5. Either the tick marks need to be repositioned according to their values, or the displayed values need to follow equal increments.
A separate issue arises when two graphs are intended to be compared. If both display counts for similar groups, their vertical axes should use the same scale for a fair visual comparison. A bar reaching halfway up a graph that goes to 20 is not the same count as a bar reaching halfway up a graph that goes to 100. Read the values and scales, not just the relative bar heights.
Worked Example: A Chart With Missing Context and Labels
Worked Example: A Chart With Missing Context and Labels
A fictional chart shows four separated bars with heights 18, 15, 9, and 6. It has no title, the horizontal axis has no category names, and the vertical axis is marked 0, 5, 10, 15, and 20 but has no axis label. Identify the problems and describe what a corrected graph needs.
Check the title. There is no title, so a reader cannot tell which group the data describe or what variable was recorded. The graph might concern surveyed students, library visitors, or something else. None of those settings can be chosen from the bars alone.
Check the category labels. The four bars have no names. Although their positions and heights are visible, a reader cannot identify which category has a height of 18 or connect any bar to a response.
Check the vertical axis. Its numerical ticks make the bar heights readable as values from 0 to 20, but the label does not explain what those values count or measure. The height 18 could represent 18 responses or another quantity. Calling it a count without evidence would be a guess.
Recommend a correction. Add a specific title naming the group and variable, label the horizontal axis with all four categories, and label the vertical axis with the measure and units, such as “Number of responses.” The example does not provide the group, variable, or category names, so those must come from the original data or question; they cannot be recovered from the bar heights.
The tick marks are evenly spaced and increase by 5 each time, so the numerical intervals themselves are consistent. That does not fix the missing information. A scale can be regular while the graph remains too incomplete to interpret in context.
Check Scale Intervals and Comparisons
A scale error is not always an obviously dramatic visual trick. Sometimes the numbers are printed next to the axis, but their positions do not match their numerical spacing. Inspect both: look at the tick labels and at the distances between the tick marks. If the axis is meant to be numerical, equal distances should correspond to equal increases.
For instance, if 0, 10, and 20 appear at equal distances, each step represents an increase of 10. A value of 15 belongs halfway between 10 and 20. If the graph instead puts 15 as far from 10 as 20 is, the scale is inconsistent. The displayed position no longer represents the stated value accurately.
When graphs are shown side by side, first determine what is being compared. If each graph shows a different group but the same measure, a shared vertical scale helps the reader compare bar heights fairly. If one graph uses counts and another uses percentages, the units differ; simply making the numerical tick labels match does not make those measures interchangeable. State the mismatch rather than treating their bar heights as directly comparable.
Worked Example: Unequal Intervals on One Axis
Worked Example: Unequal Intervals on One Axis
A fictional bar chart displays the number of reusable bottles sold in four weeks. The vertical axis has tick marks at 0, 5, 10, and 20, and all four marks are drawn the same distance apart. The bars are labeled with values 8, 12, 15, and 18. Identify the scale error and describe a correction.
Check the numerical steps. From 0 to 5 is an increase of 5; from 5 to 10 is also an increase of 5; from 10 to 20 is an increase of 10. The intervals are \(5\), \(5\), and \(10\), even though the marks are equally spaced. The last interval therefore represents twice the numerical increase of either earlier interval.
Explain the effect. The bar heights are supposed to communicate the numbers of bottles sold, but the plotted positions do not consistently represent numerical distance. A height of 15 should be halfway between the 10 and 20 positions on a correctly spaced axis. If the tick marks are equally spaced, the current display places values on an inconsistent scale and can give a misleading visual impression of differences.
Recommend a correction. Keep the vertical axis as a numerical scale and use equal increments at equal distances. For example, marks at 0, 5, 10, 15, and 20 can be spaced evenly. Then a bar at 15 will be halfway between 10 and 20. The axis should also be labeled “Bottles sold,” and the graph should have a title that identifies the weeks and the measure.
The bars’ stated values do not need to change; the scale positions do. The issue is not that 8, 12, 15, and 18 are impossible counts. It is that the axis spacing does not faithfully represent numerical differences.
Use a Shared Scale for a Fair Comparison
When two graphs display the same kind of measurement, compare their axis ranges and tick increments before comparing their bar heights. The graphs may each be internally readable but still invite an unfair comparison if one uses a much larger scale. A clear correction is usually to redraw them with the same vertical-axis range and equal tick intervals.
This is particularly useful when two panels have separate axes. A reader’s eye naturally compares how much of each panel a bar fills. If the panels use different scales, that visual comparison can suggest a difference that the counts do not support—or make a real difference hard to see. Report the actual values when available, and identify the scale mismatch explicitly.
Worked Example: Two Graphs With Different Scales
Worked Example: Two Graphs With Different Scales
A fictional community center compares the numbers of weekly equipment checkouts at two locations. One bar chart shows 24 checkouts at North Center on an axis from 0 to 30. A second chart shows 30 checkouts at South Center on an axis from 0 to 60. Each chart has evenly spaced tick marks within its own axis, but the two scales differ. What is wrong with the comparison, and how could it be improved?
Check what is being compared. Both charts display counts of weekly checkouts, so the units match. The North Center axis runs from 0 to 30, while the South Center axis runs from 0 to 60. Because the ranges differ, the bar heights as a fraction of each graph’s height are not directly comparable.
Compare the values with the scales. North Center’s bar represents 24 checkouts, which is \(24/30=0.8\), or \(80\%\), of its displayed maximum. South Center’s bar represents 30 checkouts, which is \(30/60=0.5\), or \(50\%\), of its displayed maximum. So the North bar fills a larger share of its panel even though its count, 24, is less than South Center’s count of 30. The panel-height comparison would be misleading.
Recommend a correction. Put both locations on the same vertical scale, such as 0 to 60 with equal tick intervals. On that shared scale, the graph will show 24 checkouts for North Center and 30 for South Center in directly comparable positions. The title should identify the weekly checkout measure and locations; each panel or bar should be clearly labeled.
The original values are not contradictory: 30 is greater than 24. The problem is the visual comparison created by separate scales. Naming the different ranges and recommending a shared scale explains the error more precisely than saying only that “the graph is confusing.”
Common Mistakes and AP Exam Tips
- Saying only “the graph needs labels.” Specify which labels are missing: a title, category names, the vertical-axis measure, or units. Then state what information those labels should provide.
- Guessing what an unlabeled axis means. A set of numbers does not tell you whether the graph shows counts, percentages, or another measure. Say that the measure or units are not identified; do not supply an interpretation unsupported by the display.
- Calling any different scale an irregular scale. A single graph can have equal numerical intervals even if another graph uses a different range. For comparisons, explain that the graphs use different scales; for an irregular axis, explain that equally spaced marks represent unequal numerical changes.
- Checking labels but not tick spacing. Correctly printed numbers can still be placed at incorrect distances. Compare the numerical differences between successive labels with the physical spacing between the tick marks.
- Claiming that different scales change the data values. A scale problem affects how the data are displayed and compared; it does not change the counts recorded. Name the displayed values when they are provided.
- Giving a correction that is too vague. “Fix the scale” is less informative than “use the same 0-to-60 vertical scale for both count graphs, with equal intervals.” A full-credit response names the defect and gives an actionable correction.
A strong exam response is concise but specific: “The vertical axis has equally spaced marks labeled 0, 5, 10, and 20, so the final interval represents 10 while the others represent 5. Use equal numerical increments at equal distances.” For a comparison, say which graphs use different ranges and recommend a common scale when they show the same measure. For missing labels, name the absent information and explain why the graph cannot be interpreted without it.
Check Your Understanding
For each situation, identify the graph problem and state a useful correction or limitation.
- A bar chart has a title reading “Results,” but no information about the group surveyed or the variable recorded. What should a more informative title identify?
- A graph has category names but a vertical axis labeled only “Amount.” Why is that label inadequate, and what should be added?
- Equally spaced tick marks are labeled 0, 4, 8, and 16. What is inconsistent about the numerical intervals?
- Two charts compare counts for two teams. One vertical axis goes from 0 to 40 and the other from 0 to 80. Why can comparing how much of each panel the bars fill be misleading?
- A chart has no category labels, but its bars have readable heights. What can you report about the categories, and what can you not identify?