George,

This is excellent advice, especially with regards to the census. I'm in the middle of abstracting data now for Norfolk, VA. I've had to wait until I could get a sizeable list of names from city/state records before I attempted to view the census records. Armed with a preliminary name index from deeds and other records, I've been able to identify people who were previously enumerated as white. So far the numbers aren't large, but they are significant. And to me, significant is defined as few as one name. I always think from the perspective of the researcher looking for that one ancestor. If the name is omitted, then possible all future leads are lost.

But getting back to your point, we need to be armed with more evidence as we search through the federal censuses. Or at best, if we've already viewed them, we need to return to them AFTER searching through local/state records.