Interesting data in the link above, however yet another piece of data in a large database shows that 5% of the aborts above 100 kts (without relation to V1) result in major incidents during the abort. This might suggest pilot technique in the causal link rather than just too fast to stop.
It would be difficult for me to examine the incident data above V1 simply because most incidents don't accurately report actual decision speed vs V1 (almost invariably they report the incident *at* or near V1 (but rarely do they report it as *above*).
I suspect that most of the Boeing data supporting the 2% rate they speak of is limited to corroborated data using DFDR recorders.