# RE: Hive Posts Length | Number of posts, words per posts and payouts | March 2023

You are viewing a single comment's thread:

Whenever you present averages, if you could also present the medians that would be super helpful.

Several of the distributions are clearly skewed. Whenever a distribution is skewed, the median is a better measure of central tendency than the mean (average).

0
0
0.000

Thanks for the suggestion.
Which of the charts you think is with skewed data?

0
0
0.000
(Edited)

Post length is clearly skewed. With over half the posts being less than 500 words. You've mostly accounted for that by showing average payout by post length instead of just the overall average.

However, if you showed average and median payout by post length that would be a bit more informative, especially for posts below 500 words. I suspect that that distribution is heavily skewed in and of itself.

Basically, my suggestion would be to show median every time you show average. If the underlying distribution is symmetrical, those numbers will be the same. If it's skewed, they will be different, and the median will be a better representation of central tendency, because it tells you half are above and half are below that value.

In other words, the median tells me, if I'm an average blogger and this is my average post length, the median value is what I can expect my typical payout to be.

Median values are much less influenced by extreme outliers than are the mean values.

0
0
0.000
(Edited)

I agree with the skewness of the data. Since it's skewed to right, I suggest to lower interval from 500 to maybe 200. I think that will tell a different story. The data might follow a normal distribution curve and averages can represent the data.

0
0
0.000

Reporting both means and medians is ideal. If they're equal, you know the distribution is symmetric. If they're not, then you know skewness exists, which usually warrants a deeper dive to discover the underlying reasons, as you have suggested.

If you can only report one or the other, then report medians.

0
0
0.000