20.5 Skewness and Kurtosis

Skewness

Skewness occurs when a majority of data points is at one end of the distribution. If the data are largely positioned toward the left of the distribution, the data is right- or positively-skewed. If the data are largely positioned toward the right of the distribution, the data is left- or negatively-skewed.

The remaining data points that are not in the majority positioned toward one end of the distribution or the other, are called the tail because they trail off to the other side of the distribution.

Analytics software will produce a value for skewness. If the value is negative, then you can say that the data is “negatively skewed”. If the value of skewness is positive, then you can say that the data is “positively skewed”.

Kurtosis

Kurtosis is the presence of many data points toward the ends of the distribution. If the value of kurtosis is negative, the tails are said to be “thin”. If the value of kurtosis is positive, the tails are “thick”. Analytics software differ in the values they display for kurtosis.

It is important to note that JMP centers Kurtosis on 0, meaning you can differentiate between thin and thick tails based on whether the value is greater or less than 0. In other software, Kurtosis is centered on 3, so you differentiate between thin and thick tails based on whether the value is greater or less than 3.