what does it mean if a statistic is resistant?
A statistic is resistant if its value does not change much when the data set includes extreme values (outliers) or when a few data points are altered.
Quick Scoop: What “resistant” means
In statistics, resistant basically means “hard to throw off with outliers.”
A resistant statistic stays fairly stable even if:
- You add a very large value to the data set.
- You add a very small value to the data set.
- You change one or two data points to be extreme compared with the rest.
Formally, a statistic is resistant if extreme values do not affect its value substantially.
In other words: resistant statistics “ignore the drama” in your data and focus on the typical pattern.
Examples: Resistant vs non‑resistant
Common resistant statistics
These are the “calm” ones that do not react much to outliers.
- Median (middle value):
- For data, median = 3.
* Change 5 to 500 →; median is still 3.
- Interquartile range (IQR) (middle 50% of the data):
- IQR uses only the central half of the data, so extreme highs or lows barely matter.
Common non‑resistant statistics
These react strongly to outliers and can be pulled far away from the “typical” data.
- Mean (average)
- Standard deviation
- Range (max − min)
For the same data, mean = 3.
If 5 becomes 500 →, the mean jumps to 102, a huge change caused by just one extreme value.
Here’s a quick comparison:
| Statistic | Resistant? | Effect of outliers |
|---|---|---|
| Median | Yes | Changes little when extreme values are added. | [9][1][3][5]
| Interquartile range (IQR) | Yes | Uses middle 50% of data, mostly ignores extremes. | [6][1][3][5]
| Mean | No | Can move a lot if one value is very large or very small. | [1][5][6][9]
| Standard deviation | No | Becomes large when there are extreme values. | [5][6][1]
| Range | No | Directly depends on the largest and smallest values. | [10][6][5]
Why resistance matters (in practice)
Resistant statistics are especially useful when:
- Your data has outliers or errors
- Example: A typo in someone’s salary (e.g., 50,000 entered as 500,000) can wreck the mean but barely touch the median.
- Your data is skewed
- For incomes, house prices, and many real‑world quantities, a few very large values stretch the distribution.
- In such cases, median and IQR often describe “typical” values better than mean and standard deviation.
- You want robust summaries
- Resistant statistics give a more reliable picture when you cannot fully trust every data point.
Tiny story-style example
Imagine a small class where most students score between 70 and 85 on a test, but one student forgets to answer most questions and gets 10, while another guesses everything correctly and somehow gets 100.
- The mean will swing noticeably because those two extreme scores pull it down and up.
- The median will still sit near the middle of the typical scores, barely changing at all.
That steadiness is exactly what it means for a statistic to be resistant.
TL;DR:
If a statistic is resistant, extreme values (very large or very small) do
not substantially change its value, making it a more stable and reliable
summary when your data contain outliers.
Information gathered from public forums or data available on the internet and portrayed here.
