The 365 Data Science team is proud to invite you to our own community forum. A very well built system to support your queries, questions and give the chance to show your knowledge and help others in their path of becoming Data Science specialists.
Anybody can ask a question
Anybody can answer
The best answers are voted up and moderated by our team

How much is a strong positive or negative skew?

How much is a strong positive or negative skew?


I’m just discovering about skew in the statistics section.
Excel generates a number with the skew function but I’m not sure about the magnitude of this number.
What would be the range of a strong skew and a weak skew?

1 Answer

365 Team

Hi Josh,

Thanks for reaching out!

Actually, we don’t usually talk about ‘strong positive’ or ‘strong negative’ skew. Skewness is just a measure which describes (but doesn’t ‘judge’) a distribution.

For instance a Normal Distribution (and in fact any symmetrical distribution) should have a population skewness = 0. It’s skewness isneither positive, nor negative. Moreover, for a Normal distribution we would expect the skewness to be between -0.8 and +0.8 (that’s a rule of thumb I’ve seen somewhere, but cannot really find the original source).

Now, a sample skewness of 0.5, for instance, would not mean that the distribution is skewed or not skewed. In fact, it would mean that the sample is a bit skewed. If you wish to get an idea about the magnitude (as you asked the question ‘strong negative’/’strong positive’), you could use the 0.8 as a benchmark (for a Normal distribution) and conclude that the sample is not ‘too skewed’. It is important to note here that the population may in fact be symmetrical or much more skewed. There are tests which are designed to check just that (more on that here). While on the topic, please note that while we used the Normal distribution example, there are specific tests for normality, which include more than just skewness (e.g. Jarque–Bera test). Usually, skewness alone is not enough to draw insightful conclusions – it is one of the pieces of the puzzle.

By seeing the lecture you have referenced in your question, I believe that many of the terms used here may be new to you. Therefore, it may make more sense for you to go through the course (at least check all the distribution related lectures, but ideally even go through hypothesis testing) and then come back here.

Hope this helps!


Complete Data Science Education
Get 50% OFF