Math high-school May 21, 2026

Why Do Graphs Sometimes Lie?

How design choices can change a data story

$A classroom-style data display showing two line graphs of the same data with different vertical scales so students can compare how the visual message changes.$

Graphs can mislead when their scales, axes, or selected data ranges change how the pattern looks. The numbers may be real, but the picture can make a small change look huge or hide a large change. Careful readers check the axes, units, time range, and missing data before trusting the graph.

Big Idea. Common Core 8.SP.A supports using scatter plots and patterns in data while judging whether a display gives a fair picture.

Graphs turn numbers into pictures. That makes them powerful, but it also makes them easy to misuse. A bar chart can make one value look twice as large just by cutting off the bottom of the scale. A line graph can make a slow change look steep by showing only a short time window. A scatter plot can seem to show a strong pattern if the points that disagree are left out. None of these tricks require fake numbers. The lie often comes from the frame around the data. This matters in news, sports, health, school reports, and social media posts. Data literacy means asking how a graph was built, not just reading the tallest bar or the steepest line. In math class, this connects to ratios, percent change, scatter plots, and claims based on evidence.

The axis can change the drama

Two bar charts show the same two values. One chart starts the vertical axis at zero, while the other starts near the values and makes the difference look much larger. — Same data, different starting point

A vertical axis is the measuring stick of a graph. If it starts at zero, the height of each bar is easier to compare fairly. If it starts near the data values, small differences can look much larger. This is called a truncated axis. It is not always wrong. A scientist may zoom in to study tiny changes in temperature or voltage. The problem comes when the graph is used to make a normal difference look like a crisis. Bar graphs are especially sensitive because people compare the heights of the bars. If the bottom is missing, the bar height no longer matches the full size of the value. Line graphs can also be affected, but they are often used to show change over time. A good reader checks where the axis starts and how far each tick mark moves.

A cut-off axis can turn a small difference into a big-looking one.

Scale spacing matters

Two line graphs show the same rising data with different vertical scale spacing, making one line appear steep and the other appear gradual. — Same trend, different scale

A graph scale tells how much each step on an axis is worth. Equal spaces should usually mean equal amounts. If a vertical axis jumps by 1, then 10, then 100, the shape can become hard to read unless the graph is clearly labeled as a special scale. Even when the scale is even, the size of the step changes the feeling of the graph. A narrow scale makes changes look steep. A wide scale makes the same changes look gentle. This is why two line graphs can use the same data and still give different impressions. The slope you see depends on the units shown on each axis. The safest habit is to read the numbers before reacting to the shape. Look at the tick marks, the units, and the total range. A steep-looking line is not evidence by itself.

The slope you notice depends on the scale you are shown.

Time windows can hide the pattern

A long line graph shows ups and downs over time, with a highlighted short section that rises even though the full pattern is mixed. — A selected window can change the story

A graph can be honest about the points it shows and still hide the larger story. This happens when the time range is carefully chosen. A stock price, temperature record, or test score trend may rise over one short period but fall over a longer period. The reverse can also happen. Choosing only the part that supports a claim is called cherry-picking. In school math, this connects to describing patterns and judging whether a claim is supported by data. A fair graph should make the selected range clear. It should also explain why that range matters. Readers can ask what happened before and after the displayed window. If the graph starts at a convenient peak or ends at a convenient low point, the claim may be weaker than it looks. The missing time can be just as important as the time shown.

A short range can make a mixed pattern look simple.

Missing points change the message

Two scatter plots compare the full set of data points with a version where several points are missing, making the relationship look stronger. — Leaving out points can strengthen a claim

Scatter plots show pairs of data values. Each point matters because the overall pattern comes from the cloud of points, not from one example. If someone removes points that do not fit the claim, the pattern can look stronger than it is. A weak relationship can appear strong. A strong relationship can appear weak if key points are hidden. Outliers need careful treatment. Sometimes a point is removed because it came from a measurement error. Sometimes it is removed because it is inconvenient. Those are very different choices. A trustworthy graph should say how the data were collected and whether any points were excluded. In math class, students can compare a scatter plot before and after removing a point. They can see how the line of best fit and the interpretation can change. The graph is only as fair as the data behind it.

A pattern built from selected points is not the same as the full data.

Percent change needs a base

A comparison display shows two equal increases of one unit, where one small starting value has a large percent change and one large starting value has a small percent change. — Equal increases can have unequal percents

Percent change can be useful, but it depends on the starting amount. A change from 1 to 2 is a 100 percent increase. A change from 100 to 101 is a 1 percent increase. Both changes add one, but the percent stories are very different. Graphs can use percent change to make a small count look dramatic, especially when the starting number is small. They can also use raw counts to hide how large a relative change is for a small group. Neither raw numbers nor percents are automatically better. They answer different questions. Raw counts tell how many. Percents tell how large compared with the starting amount. A fair explanation often shows both. When reading a graph, check the denominator or base group. If the base is not clear, the percent may be easy to misunderstand.

Percent change is hard to judge without knowing the starting value.

Vocabulary

Axis: A reference line on a graph that shows the scale and units for the data.
Scale: The set of values marked along an axis and the amount represented by each step.
Truncated axis: An axis that does not show the full range from zero or from the natural starting point.
Cherry-picking: Selecting only the data that support a claim while leaving out data that may change the conclusion.
Outlier: A data point that is far from the main pattern and may need extra investigation.
Percent change: A comparison of how much a value changes relative to its starting value.

In the Classroom

Redraw the same data

20 minutes | Grades 8-12

Give students one small data table and ask them to make two graphs that use different vertical scales. Students compare how the visual message changes, then write a fair caption for each graph.

Find the missing window

25 minutes | Grades 8-12

Show students a short time-range graph and the full time-range graph from the same data set. Students identify what claim the short graph supports and what the full graph adds or changes.

Data honesty checklist

30 minutes | Grades 9-12

Students review graphs from articles, ads, or class-created examples. They check axes, units, time range, missing data, and whether raw numbers or percents are being used.

Key Takeaways

• A graph can use real numbers and still create a misleading impression.
• Truncated axes can exaggerate differences, especially in bar graphs.
• Changing the scale can make the same trend look steep or flat.
• Cherry-picked ranges and missing points can hide the larger pattern.
• Careful graph reading starts with axes, units, ranges, and the data source.

Sign in to save