A resistant statistic is a measure that changes very little when a dataset contains an extreme value, also called an outlier. This matters because real data often include unusually large or small observations from measurement error, rare events, or natural variation. If a statistic is strongly pulled by an outlier, it may describe the extreme value more than the typical data.
Comparing resistant and non-resistant statistics helps you choose a summary that matches the shape of the data.
Understanding Statistics: Resistant vs Non-Resistant Statistics
The key difference comes from how a calculation uses each observation. The mean gives every value a full share of the total, so one unusually high or low result changes that total before the final division. Standard deviation is even more sensitive because it measures distances from the mean and gives large distances extra weight through squaring.
The range uses only the two endpoint values. A single mistaken endpoint can therefore change the range completely. These measures can be useful when the extreme value is real and important, but they need careful interpretation.
Imagine recording the time needed for eight students to finish a short task. Most finish between nine and thirteen minutes, while one student takes forty minutes because their computer freezes. The mean time rises noticeably, even though it no longer matches what most students experienced.
The median stays near the center of the usual times. The middle half of the results stays similar too, so the IQR gives a steadier picture of the spread among typical students.
If the forty minute result came from a real accessibility need rather than a technical problem, it should not simply be deleted. The data needs an explanation before anyone chooses a summary.
Outliers are not automatically errors. A very high rainfall total may show a severe storm. An unusually large test result may come from a student who already knew the material.
A low reading from a sensor may signal that the instrument failed. First check the source, units, and method of collection. A misplaced decimal point can create a false outlier.
Then look at a graph such as a dot plot, histogram, or box plot. Graphs reveal gaps, clusters, skewed shapes, and isolated values that a single summary number can hide.
In school problems, the best choice depends on the purpose. For roughly balanced data with no strong extremes, the mean and standard deviation describe the center and variation well. For income, house prices, waiting times, and many environmental measurements, the distribution often has a long tail on one side.
Median and IQR usually give a fairer summary of a typical case in those situations. When comparing two groups, use the same kind of summary for both groups.
Pay attention to whether an outlier changes the conclusion. Reporting both a graph and suitable numerical measures makes the evidence easier to judge.
Key Facts
- Mean = (sum of all data values) / n.
- Median = the middle value when data are ordered, or the average of the two middle values if n is even.
- Range = maximum value - minimum value.
- IQR = Q3 - Q1, where Q1 and Q3 are the first and third quartiles.
- The median and IQR are resistant because they depend mostly on position in the ordered data.
- The mean, standard deviation, and range are non-resistant because extreme values can strongly change their sizes.
Vocabulary
- Resistant statistic
- A statistic that is not greatly changed by extreme values or outliers in a dataset.
- Non-resistant statistic
- A statistic that can change a lot when an outlier is added, removed, or changed.
- Outlier
- A data value that is unusually far from the rest of the dataset.
- Median
- The middle value of an ordered dataset, or the average of the two middle values when there is an even number of values.
- Interquartile range
- The spread of the middle 50 percent of a dataset, found using IQR = Q3 - Q1.
Common Mistakes to Avoid
- Using the mean for a strongly skewed dataset with outliers, because the mean can be pulled toward the extreme values and may not represent a typical observation.
- Saying the median ignores all data values, because the median still depends on the ordered positions of the values, but it is not strongly affected by how far an outlier is from the center.
- Treating the range as resistant, because the range uses only the minimum and maximum values and can change dramatically from one extreme observation.
- Choosing standard deviation to describe spread when outliers dominate the data, because standard deviation uses distances from the mean and extreme distances can make the spread look much larger.
Practice Questions
- 1 For the dataset 4, 5, 5, 6, 6, 7, 8, find the mean, median, range, and IQR.
- 2 For the dataset 4, 5, 5, 6, 6, 7, 30, find the mean, median, and range. Compare your answers with the dataset 4, 5, 5, 6, 6, 7, 8 and identify which measures changed the most.
- 3 A teacher records quiz times for most students between 10 and 15 minutes, but one student takes 45 minutes because of a technical problem. Explain whether the teacher should report the mean and standard deviation or the median and IQR to describe the typical quiz time, and justify your choice.