Is correlation a percentage? The Pearson correlation coefficient ranges from -1 to 1. Oftentimes, people take this number, multiply it by 100 and call it

-17.3% correlation

or

63% correlation

I was once dinged by reviewers for this, in a STEM journal. Is this really a mistake?

Examples of real-world usage:
Book: Spurious Correlations:

Textbook: Clear-Sighted Statistics: Module 18: Linear Correlation and Regression:

 A: It’s wrong, and if a reviewer wants to tell you to change it, you have no argument. I would not, however, consider that to be more than a typo (minor revision), even if I said it should be changed.
I see an argument that it’s just a slang that perhaps has no place in formal writing like a scientific article but is fine for casual discussions. However, squaring the correlation has an interpretation as a proportion or percentage (it’s $R^2$ in a regression involving your two variables), so I think I do not like such a slang. If you mentioned having a correlation of $81\%$, that could correspond to $r=0.81$, $r=0.9$, or $r=-0.9$.
A: Informally, you can do whatever works for you/your group. I can see a lot of reason why it can be easy to visualize a set of positive correlations as proportions towards two ideal states (0 and 1).
However, formally, I think this is wrong on many conceptual levels. Most important: correlations measures are (usually) not additive. It means that the difference in information between $r = .5$ and $r = -0.5$ is not the same as the difference between $r = 1$ and $r = 0$ even if the metric difference is the same. And this holds whatever two couples of points in the scale you take, roughly.
Personally, I came to the idea that if you are not at the market, adding $%$ to numbers is always a bad choice because it is misleading.
