Skip to content
AP Statistics · Exploring One-Variable Data and Collecting Data

Outlier

Outlier appears in 4 reviewed AP Statistics questions and sits in Exploring One-Variable Data and Collecting Data. Below are worked examples showing exactly how the exam tests it, with the answer marked and every distractor explained.

How Outlier is tested

Example 1Medium
The IQR of a data set is 18. What is the length of the 'whiskers' in a standard box plot if there are no outliers?
  1. AThe whiskers extend at most 27 units (1.5 × IQR) beyond Q1 and Q3.Correct
  2. BThe whiskers extend exactly 18 units on each side.
  3. CThe whiskers extend to the mean plus or minus one standard deviation.
  4. DThe whiskers always extend to the minimum and maximum values, regardless of outliers.

Why A is right

In a standard box plot, whiskers extend from Q1 and Q3 to the most extreme data points that are still within the fences: Q1 - 1.5(IQR) and Q3 + 1.5(IQR). With IQR = 18, the fences are 1.5(18) = 27 units away from Q1 and Q3. If no values fall beyond the fences, the whiskers reach the actual min and max, which are at most 27 units from Q1 and Q3.
Why the other 3 choices miss
B
Whiskers extend 1.5 × IQR, not 1 × IQR, beyond Q1 and Q3. With IQR = 18, the maximum fence distance is 27, not 18.
C
Box plot whiskers use the IQR-based fences, not the mean and standard deviation. The SD is not used in constructing box plots.
D
Whiskers do NOT always extend to the minimum and maximum. In a modified box plot (the standard AP Statistics version), whiskers extend only to the most extreme NON-OUTLIER values. Points beyond the fences are plotted individually as outliers.
Example 2Harder
A sample of n = 10 measurements has a boxplot showing a strong right skew with one extreme high outlier. The researcher wants to test a claim about the population mean. Which of the following is the best advice?
  1. ARemove the outlier and proceed with the t-test on the remaining 9 observations
  2. BProceed with the t-test, since t-procedures are robust to any skewness or outliers
  3. CAvoid the t-test; with n = 10, the skew and outlier violate normalityCorrect
  4. DUse a z-test instead, since the z-procedure does not require a normal population

Why C is right

With only 10 observations, the t-procedure requires the population to be approximately normal. Strong right skew and an extreme outlier indicate a substantial departure from normality that the t-procedure cannot tolerate at this sample size. The researcher should consider nonparametric alternatives or data transformations rather than applying the t-test to data that violate its conditions.
Why the other 3 choices miss
A
Incorrect. Removing an outlier without substantive justification is not appropriate statistical practice. The outlier may be a legitimate observation, and removing it changes the research question and can introduce bias.
B
Incorrect. t-procedures are not robust to any amount of skewness or outliers. For small samples, they require approximate normality. Strong skewness and extreme outliers can produce misleading results.
D
Incorrect. The z-test also assumes normality (or large samples via CLT) and additionally requires knowing sigma. Switching from t to z does not solve the normality problem.

Studied alongside

Other concepts from Exploring One-Variable Data and Collecting Data.