Of course he's talking about studies that use significance tests, which his own paper isn't directly using to prove his point.