How is the Altmetric score calculated?

While the most important part of an Altmetric report is the qualitative data, it's also useful to put attention in context and see how some articles are doing relative to others.

The Altmetric score for an article can give you an idea of the attention that it has received.

The score is a weighted count

The score is a weighted count of the different sources (newspaper stories, tweets, blog posts, comments) that mention the paper. 

Why is it weighted? To reflect the relative importance of each type of source. It's easy to imagine that the average newspaper story is more likely to bring attention to the paper than the average tweet. This is reflected in the default weightings:

Example default score contributions for different sources
News Blogs Q&A forums Twitter Google+ Facebook
8 5 2.5 1 1 0.25

To produce these default weightings we asked groups of researchers to rank different sources in order of how important they were to an article's broader impact.

So if your paper has been mentioned by a news story and three tweets it might score:

8 (default News contribution) + 3 (3 x default Twitter contribution) = 11

We round scores up. If it has one blog post and a Facebook mention it may score:

5 (default Blog contribution) + 0.25 (default Facebook contribution) = 5.25 = rounded up to 6

Dealing with edge cases by weighting individual mentions

So far so good. However, what if the newspaper in the first example above is a local title that nobody reads and one of the tweets is from Steven Hawking (or Barack Obama, or Brian Cox, or anybody else with a large following)? The type of each source is important but so are details of the actual person or outlet mentioning the article.

Wherever possible we look at reach (how many people is the mention going to reach?), promiscuity (how many other articles have been mentioned by the same source in a short time period?) and bias (are all the articles from a single journal or publisher?)

Where there are marked deviations in any of the above we weight the mention. In practice while most blog posts contribute 5 some with large readerships may contribute up to 9 (more than an average news story). Equally, mentions in newspapers with a large reach may score higher, or much lower if they are too promiscuous: sites that simply reproduce large volumes of press releases will contribute less than average to the score.

Remember

The Altmetric score is useful to rank articles based on attention - it can't tell you anything about the quality of the article itself, though reading the linked discussions might.

It is important to know that the score is based on the kinds of attention that Altmetric tracks (specifically links to or saves of scholarly articles, books and datasets) and to be mindful of potential limitations.

You should also bear in mind that different subject areas usually aren't directly comparable: a "popular" physics paper may have a far lower Altmetric score than an "average" genetics paper.

We don't use downloads, citations or reader counts from Mendeley or CiteULike in the score calculation as we're unable to audit those sources (for all other sources you can see each individual mention on the details page).

Finally, you should keep in mind that in some rare cases, the Altmetric score may fluctuate slightly over time. Fluctuations can happen for various reasons, such as when tweets get removed by the original tweeter. We're also trying to improve the accuracy of our scoring algorithm and sometimes need to adjust the weights of certain mentions to appropriately reflect the source's reach (this is particularly true for newspapers).

Feedback and Knowledge Base