A truncated graph is a chart where the vertical axis (y-axis) doesn’t start at zero. Instead, it begins at some higher value, cutting off the lower portion of the scale. This makes differences between data points look much larger than they actually are. Truncated graphs appear frequently in news media, business presentations, and political messaging, sometimes by accident and sometimes to make a trend look more dramatic than the data supports.
How Truncation Changes What You See
Imagine a bar chart comparing two companies’ quarterly revenue: one earned $98 million and the other earned $100 million. On a chart starting at zero, those two bars look nearly identical, because a $2 million gap is tiny relative to the full scale. Now imagine the same chart with the y-axis starting at $97 million. Suddenly, one bar appears roughly three times taller than the other, even though the underlying difference hasn’t changed at all.
That’s the core mechanic of truncation. By starting the axis from a value other than zero, the chart designer compresses the visible range and over-emphasizes small differences that would otherwise appear minor. The data itself is accurate, but the visual story it tells is distorted. A slight uptick becomes a dramatic spike. A modest decline looks like a freefall.
This works on line graphs too. When the y-axis is truncated, the slope of a trend line gets steeper. A gradual increase that would look nearly flat on a full-scale chart can appear to shoot upward when the bottom of the scale is lopped off.
Why It Fools Nearly Everyone
You might think that being aware of truncated axes would protect you from misreading them. Research published in the Journal of Applied Research in Memory and Cognition tested exactly this across five separate studies, and the results were striking. Viewers consistently judged the differences shown in truncated graphs as larger than the same differences on full-scale graphs. This “truncation effect” showed up in 83.5% of participants across all five studies, regardless of how well they understood graphs in general.
Even training didn’t fully solve the problem. When participants were taught how truncation distorts perception, the effect shrank somewhat, both immediately and after a one-day delay. But it never disappeared. PhD students in quantitative fields like statistics and engineering were slightly more resistant than those in the humanities, but they were still susceptible. The effect sizes ranged from moderate to large, meaning this isn’t a subtle bias. It’s a powerful visual illusion that overrides what people know intellectually about how charts work.
Truncated Graphs in the Real World
Crime statistics are a common target for this technique. A bar chart of U.S. murders that starts its y-axis at 16,340 instead of zero can make a small year-over-year increase look like a public safety crisis. The same data plotted from zero reveals that the change is barely visible relative to the total. Similarly, a line graph of U.S. robberies that starts at 316,000 can make a modest uptick over a few years look like a steep, alarming trend, even when zooming out to the full dataset reveals that robberies actually dropped significantly in the years that followed.
These aren’t hypothetical scenarios. They reflect documented examples using FBI Uniform Crime Report data, where the same numbers can tell wildly different visual stories depending on where the y-axis begins. Someone arguing that the country is becoming more dangerous would naturally prefer the truncated version. Someone making the opposite case would show the full scale. The numbers are identical in both charts.
This pattern shows up everywhere: cable news segments showing economic indicators, corporate earnings reports highlighting revenue growth, and political ads comparing candidates’ records. Whenever the stakes of perception are high, truncated axes tend to appear.
When Truncation Is Actually Appropriate
Not every truncated axis is misleading. In some cases, starting at zero would actually make a chart less useful. If you’re tracking a patient’s body temperature over several days, a y-axis from 0 to 110°F would compress all the data into a tiny band near the top, making it impossible to see clinically meaningful fluctuations between 98°F and 103°F. The same logic applies to stock prices, atmospheric CO₂ concentrations, or any dataset where the meaningful variation happens within a narrow range far from zero.
Line graphs get more leeway here than bar charts. The height of a bar is supposed to represent the full quantity it shows, so cutting off the bottom fundamentally distorts the visual comparison. A line graph, on the other hand, primarily communicates the shape of a trend rather than absolute magnitude. Research suggests that for line graphs, starting the y-axis about 1.5 to 2 standard deviations below the mean of the data allows viewers to interpret the trend correctly without the chart becoming unreadable.
The key distinction is intent and context. A truncated axis paired with clear labeling, appropriate scale markings, and an honest purpose is a legitimate design choice. A truncated axis designed to make a small change look dramatic, especially with minimal labeling, is manipulation.
How to Spot a Truncated Graph
The single most reliable habit is checking where the y-axis starts before reacting to any chart. If it doesn’t begin at zero, mentally ask yourself: what would this look like on a full scale? That question alone can prevent most misreadings.
- Check the axis labels. Look at the lowest value on the y-axis. If it’s not zero, the chart may be exaggerating differences.
- Look for axis break symbols. Some charts include a zigzag line or double slash near the bottom of the axis to signal that part of the scale has been removed. Many don’t bother, though.
- Compare the visual to the numbers. If two bars look dramatically different but the actual values are close together (say, 47% vs. 49%), truncation is doing the heavy lifting.
- Consider the chart type. Truncation on a bar chart is almost always misleading. On a line graph, it may be reasonable depending on the data range.
Remember that even knowing about this trick doesn’t make you immune to it. The research is clear on that point. Your brain processes the visual proportions of a chart faster than it reads the axis labels, so the distorted impression lands first. Checking the axis is a deliberate, conscious correction you have to make every time.