Skip to content
AP Statistics · Inference for Quantitative Data: Means

Robustness

Robustness appears in 5 reviewed AP Statistics questions and sits in Inference for Quantitative Data: Means. Below are worked examples showing exactly how the exam tests it, with the answer marked and every distractor explained.

How Robustness is tested

Example 1Harder
A student wants to test whether the mean commute time in her city differs from the national average of 26 minutes. She collects a random sample of 16 commuters and finds a sample mean of 29 minutes and a sample standard deviation of 8 minutes. Before conducting a t-test, she checks conditions. The sample data show a strong right skew with one extreme outlier. What should she do?
  1. ABe cautious about the t-test, since with n = 16 it is not robust to strong skewness and outliers; consider a nonparametric test or a larger sampleCorrect
  2. BProceed with the t-test because the sample size of 16 is large enough
  3. CIncrease the significance level to alpha = 0.10 to account for the skewness
  4. DRemove the outlier and proceed with the t-test on the remaining 15 observations, since without the extreme value the remaining data can be treated as approximately normal

Why A is right

With n = 16, the t-test is not robust to severe departures from normality. Strong right skew with an extreme outlier violates the normality condition and cannot be overcome by a sample size this small. The researcher should either collect a larger sample (so the CLT provides approximate normality), use a nonparametric test, or investigate the outlier.
Why the other 3 choices miss
B
Incorrect because n = 16 is not large enough for the CLT to compensate for strong skewness and outliers; larger samples (n >= 30-40) are needed.
C
Incorrect because changing alpha does not address the violation of the normality condition; it only changes the decision threshold.
D
Incorrect because removing an outlier without justification (such as a data entry error) can bias the results and is not recommended practice; moreover, the strong right skew remains even after the outlier is dropped, so the data still cannot be treated as approximately normal.

Studied alongside

Other concepts from Inference for Quantitative Data: Means.