Negatively Skewed — Definition, Formula & Examples
Negatively skewed describes a distribution whose long tail stretches toward the smaller values on the left. Most of the data piles up on the right, and a few unusually low values pull the tail out to the left. It is also called left-skewed.
A distribution is negatively skewed when its coefficient of skewness is less than zero. The standard measure is the third standardized moment, . Cubing keeps the sign of each deviation, so large negative deviations from the mean dominate and make this value negative.
Key Formula
Where:
- = sample (moment) coefficient of skewness
- = the i-th data value
- = sample mean
- = number of data values
Worked Example
Problem: Is the data set 2, 8, 9, 10, 11 negatively skewed? Compare the mean and median, then compute the skewness coefficient .
Step 1: Find the mean. The data are already in order, so the median is the middle (3rd) value, 9. The mean is less than the median, because the low value 2 pulls the mean left.
Step 2: Find the deviations from the mean: . Sum their cubes and their squares.
Step 3: Divide each sum by and substitute into the formula.
Answer: Yes. The skewness coefficient is , and the mean (8) is less than the median (9). Both point to a negatively skewed data set.
Visualization
Why It Matters
On the AP Statistics exam, you must describe a distribution's shape, center, spread, and unusual features. Shape tells you which summary to report. For a skewed distribution, the median and IQR resist the tail better than the mean and standard deviation. Exam scores on an easy test and age at retirement are typical left-skewed variables.
Common Mistakes
Mistake: Calling a distribution right-skewed because its peak sits on the right side of the graph.
Correction: Skew is named for the direction of the tail, not the peak. A tall cluster on the right with a tail trailing left is negatively (left) skewed. Also, 'mean < median' is only a rule of thumb. It usually holds for left-skewed data, but some distributions break it, so check the graph or too.
