Getting back to the question of 1 year date differences, here is my take on it, and it has proved correct many times.
Oftentimes (especially in US before 1900) censuses are the primary - if not only - sources available. Required birth registrations were late-coming by the states, and church records often don't exist. Plus, as people in the US and Canada moved West, the availability of their "church of choice" often wasn't there. I have seen families who were strict Presbyterians in Scotland and Nova Scotia, attending Anglican, Methodist, and even Baptist churches for baptisms, marriages, and burials.
Censuses in Britain seemed to be taken (normally) in March; in the US the norm was June, but I've seen anywhere from early January to September. Say an individual shows up as 20 years old in an 1860 census, and 30 yo in 1870: the tendency is to subtract 20 or 30 and come up with a birth year of 1840. For a census taken in March, you only have a 25% chance of this being correct, and if taken in June, only a 50% chance.
I have found this "off by one" error many times in my own research. What I now do is put the birth year (in the above example) down as "ca 1839-1840" until I find definitive proof.
I would guess that the IGI entries go for the easy fix by simply subtracting.
Nick