That's covered in the article; the projects both use git, so libav commits are merged with the original committers intact, not under someone else's name.
But the merge commits are still authored by Michael. Not having looked at the git repository, but if merges happen often, that can very much artificially inflate the number of commits attributed to him, except if the stats were gathered with --no-merges.
Are the commits from libav merged or cherry-picked/rebased? In any case, I would hope the stats were gathered with --no-merges; if not, that would indeed skew the stats.
There's a merge commit for each commit pulled from libav (and it's frequently not empty, as the codebases have diverged and so conflicts are common).
While it's true that Michael generates lots of merge commits, if you click the link in LWN article to Andreas Cadhalpun's mail [1], then you can read that following commands were used to generate statisticts: