Interpreting and Comparing Skewness

Both coefficients are relative measures — pure numbers — and are read the same way regardless of the data's units.

The size of the number (ignoring its sign) tells the degree: the closer it is to $0$ the more nearly symmetric the distribution, and the closer to its limit ($\pm 3$, practically $\pm 1$, for Karl Pearson's; $\pm 1$ for Bowley's) the more strongly skewed it is. A coefficient of $+0.8$ describes a distribution far more skewed than one of $+0.1$, and the positive sign in each case fixes the direction as right-tailed.

Because both coefficients are dimensionless, they let an analyst compare the skewness of two different distributions even when the raw data are measured in different units or on different scales — which is the whole practical purpose of a relative measure. When comparing, keep the two coefficients separate: report a Karl Pearson value against another Karl Pearson value, and a Bowley value against another Bowley value, since the two methods use different parts of the data and are not on the same numerical scale. …