Then perhaps the peak identification is wrong -- surely they haven't tested solid comparison groups for such claims, like individuals that didn't receive education later in life.