A lot of the tables and math around demographics seems to be set around the family unit, but the tables on page 235 seem to be based on individuals. That seems to be resulting in a lot of leveled characters per household.
I'm modeling a realm of 400,000 humans, or 80,000 households. After I run the numbers through a spreadsheet based on the Demographics of Leveled Characters based on individuals I'm coming up with 32,347 leveled characters. That is resulting I. A little more than 4 in 10 households having a leveled character, which seems quite high. I'd say that those ratios should be based on the family unit, but the text describing the purpose of the table states the table shows "the approximated frequency of leveled characters per person".
Keep in mind that at least 75% of those are going to be 1st level fighters who're just war veterans.
That would be a lot of war veterans. The numbers I came up with has has 20,000 out of the 32k+ being first level. The class distribution is usually modeled as 4 (fighters):2 (clerics):2 (theives):1 (mages). From that I get (from the top of my head – I'm at a game now) around 25% of the leveled NPCs are 1st level fighters, 12.5% 1st level clerics, 12.5% are 1st level theives (a whole lot of robbing going on) and 6.25% are 1st level mages. Alternatively you could work it out to out of every 36 households (still doing math in my head so it could be off):
4 have 1st level fighters
2 have 1st level clerics
2 have 1st level theives
1 has a 1st level wizard
5.4-ish have higher level NPCs
You seem to be doing the math correctly. Keep in mind the following:
1. Leveled characters tend to cluster. The number of politicians living in Washington DC is vastly higher than the number across America. The number of US Army Rangers living in Ft. Benning, GA is disproportionate to the number across America. So it's not likely that 40% of households have a 1st level character.
2. Characters reach middle age and old age quite fast in ACKS. Life expectancy at age 20 in Antiquity was approximately 60. Assuming level 1 at 20, approximately 37.5% will be adult, 50% middle aged, and 12.5% old, e.g. over half are middle aged or above, and thus not at their peak.
Anyway, if that's too many leveled characters for your taste, simply adjust to suit. But if you do so, it has radical implications for the number of clerics available to cast healing spells, the number of fighters available to lead troops and rule domains, etc.
This has 1 in 60 characters possessing divine levels. Not entirely unreasonable?
I can't get reliable numbers from the internet with a quick search, but wild estimates:
Population of England in the 12th century: ~2 million (1.5e6 and growing at Domesday in 1086)
Number of monks in England in the 12th century: > 40,000 (there were > 400 monasteries founded; I have no figures on average sizes, and have read of some being 10 or 20 residents, but looking at some of the commonly-posted floorplans of well-known monasteries > 100 residents also seems reasonable; assuming closure of new foundations balances out any pre-existing establishments).
Thus density of monastics is 2% or so.
Then another 2% for clerics (http://faculty.history.wisc.edu/sommerville/123/123%20132%20church.htm); "many of these were in "minor orders" ... who could conduct services but not officiate at mass".
If you model them all as ACKS clerics, 2/3rds of them will be 1st level and have no spellcasting, which fits sub-deacons, deacons, and monastic novices just fine.
I tried to get into an argument with Alex once that felt a bit inconclusive, but if I recall correctly the gist was that I'd expect that you'd over-represent leveled fighters among active military, so maybe more than 1 in 7 would be above 0 level, but the discrepancy needn't be huge; I figure most of us aren't trying to exactly locate and describe every one of those 6 million people, but get an internally-consistent distribution to apply.