Your math is disingenuous. In the "Not Paying" category for Direct Loans, you included the 7.0 million students who were (cell D32 in your linked Federal Reserve spreadsheet).

No shit they weren't paying. They're in school.

Removing them gives you 3.9 million in non-payment, which drops the non-payment % to 43%.

43% in non-payment is bad enough to not fudge the number so you can use the clickbait term "over half" in the headline. Using the right data will gain you a lot more followers than this mathematic spin.