[matthewdgreen]

You can look at the charts themselves to determine whether they're different. And then if you determine they are, then talking about the slopes (or coefficients) of variables doesn't seem like plagiarism to me.

More generally, "the large numbers of incidents" here is exactly what I'd expect if someone ran a text detector, but didn't do any quality control to see if the result actually represented theft of original research.

[dekhn]

I honestly think you are either intentionally or mistakenly ignoring the large swaths of direct copied text, as in entire sentences/paragraphs.

False positives in plagiarism detection is a problem but my inspection of the examples found roughly 3-4 examples of directly copied sentences, usually with just one word changed, from a paper that she explicitly cites elsewhere in the document. Under my understanding (I'm an ex-academic with published papers) this counts as real plagiarism, and since it's repeated throughout multiple works, I think it's safe to conclude she did this intentionally and just thought she wouldn't get caught. I could be wrong.