You ask for a cite, and I'm talking about math.
Math doesn't really require a cite. It just needs you and me to verify it.
For instance, look here, in the section called Turnout Statistics. In the table, look at the last column, called Turnout as % of VEP (voting eligible population). It's a short column, it starts in 1980.
en.wikipedia.org
Looking carefully at that data, one notices the distribution is bimodal, which means there are two clusters, one from 1980 thru 2000, and another from 2004 thru 2016. The first mean is around 55, the second around 60. But in 2020 we're about 67%, which is 20% larger than the delta between the first two means.
That number, is much larger than the number of new teenagers. The first delta corresponds with the advent of motor voter laws. The second does not.