Medved used 2005 census numbers.
http://www.townhall.com/columnists/MichaelMedved/2007/01/18/journalistic_malpractice_in_marriage_is_dead_report
Even Michael's debunking arrives at an actual number of 51% married, as opposed to the 56% listed by the census, or the 49% listed by the NYT. From the Townhall article you referenced:
As a matter of fact, the Census Bureau shows that among all females above 15 the majority (51%!) are still classified as married.
So, my point stands. 15-19 year olds were the prime demographic to be married for the vast majority of American history. In other words, if you want to compare apples to apples and not apples to oranges, you need to include them. Excluding them now would lead to statistics as flawed as the nonsense the NYT peddled.