The data cited there looks a lot like some of the other sources I happened on and some of them noted in their analysis that the data from the 1860 Census they examined was incomplete in one way or another due to events back then (e.g., Jeff Davis’ Secession Speech Jan 1861).
So you have an 1850 Census and an 1870 Census to try to fill in/augment the blanks. I read in some sites that the population increased statistically significantly between 1850 and 1860 (and from 1860 to 1870) I'm assuming due to immigration.
Another factor in reconstructing a true picture is availability of local records. If you drive through a lot of the small county seat towns here in Georgia, in some of the towns you'll see an empty square, others have courthouses built after the war. Marietta is one example. There used to be a courthouse there but Sherman burned it (and the city along with Atlanta and others). The square to this day is empty but for the accouterments of a public park.
http://marietta.patch.com/groups/around-town/p/the-burning-of-marietta
Nonetheless, I think what your analysis boils down to as an average, about 1 in 5 slave owners overall, is probably the most accurate estimate of the ones I've seen.
Yes, overall roughly one in five Southern families owned slaves -- ranging from one in three in Deep South States like South Carolina and Mississippi (#1 & #2 to secede), one in nine in Border States like Maryland and Missouri (which never voted to secede).
On previous threads I posted miscalculated numbers as high as 50% instead of the correct 33% maximum.
No excuses, just my sloppy work.
Hopefully this can start to correct any misimpressions... ;-(