OK you're right. the scales are:

• on a scale of all US adults, the percentage who have a college degree
• on a scale of all readers who primarily get their news from (given magazine), the percentage who have a college degree

So the scales are still different.

I'm guessing they'd make an argument that: "If the college graduate readership were distributed evenly across all news sources, then (given magazine) would have (the percentage of all US adults who have a college degree)." But the labels don't say that, which is why it is confusing.