Mathwords logoMathwords

Misuse of Statistics — Definition & Example

Misuse of statistics is when data are presented or interpreted so they give a false impression. The numbers themselves can be accurate while the conclusion they suggest is wrong.

A misuse of statistics is any method of collecting, summarizing, displaying, or interpreting data that supports a conclusion the data do not justify. Common types include biased or unrepresentative sampling, selective reporting (cherry-picking), graphs with distorted or truncated scales, comparing raw counts when proportions are needed, and inferring causation from correlation.

How It Works

Watch for these common tricks. A truncated axis exaggerates small differences, as in the example below. A biased sample, such as an online poll answered only by a show's own fans, overrepresents one group, so its results do not describe the whole population. Cherry-picking reports only the time period, subgroup, or study that supports a claim and leaves out the rest. Raw counts mislead when groups differ in size: a big city can have more car accidents than a small town and still have a lower rate per 1,000 drivers. Correlation is not causation: ice cream sales and drownings rise together because hot weather drives both, not because one causes the other.

Worked Example

Problem: A bar graph compares bike sales. Store A sold 50 bikes and Store B sold 55 bikes. The vertical axis starts at 45 instead of 0. How much more did Store B actually sell, and how much more does it appear to sell on the graph?
Step 1: Find the actual percent increase from Store A to Store B.
55−5050×100%=10%\frac{55-50}{50}\times 100\% = 10\%
Step 2: Find the visible bar heights. The axis starts at 45, so each bar shows only the part above 45. Then compare the two heights.
55−4550−45=105=2\frac{55-45}{50-45} = \frac{10}{5} = 2
Step 3: Store B's bar is twice as tall as Store A's bar. On the graph, Store B appears to sell 100% more.
10−55×100%=100%\frac{10-5}{5}\times 100\% = 100\%
Answer: Store B actually sold 10% more bikes, but the truncated axis makes it look like 100% more. Starting the axis at 0 would show the true comparison.

Why It Matters

In AP Statistics, free-response questions often ask you to spot bias in a survey or to explain why a study cannot show cause and effect. Advertising claims and news charts use the same tricks, so this skill helps you judge evidence outside the classroom.

Common Mistakes

Mistake: Assuming a statistic is trustworthy as long as the numbers are correct.
Correction: Accurate numbers can still mislead. Check how the sample was chosen, where each axis starts, whether counts should be proportions, and whether a cause is being claimed from a correlation.

Related Terms

  • Descriptive Statistics — Summaries whose selective use can mislead readers
  • Bar Graph — Truncated axes most often distort bar graph comparisons
  • Histogram — Unequal bin widths can misrepresent a distribution's shape
  • Scatterplot — Shows correlations that people wrongly read as causation
  • Time Series — Chosen start and end dates can cherry-pick trends
  • Relative Frequency — Proportions prevent misleading comparisons of groups of unequal size
  • Bias — Systematic error that often causes misleading statistical conclusions
  • Biased Sample — A sample that overrepresents some part of the population
  • Correlation vs. Causation — Why two variables moving together does not prove cause