There's a larger proportion of older males in the population now than there has been in most of history. Testosterone decreases as we age. The average is across _all_ males, even those outside of reproductive years. I assume if we controlled for outliers, that the results would be fairly comparable across time.