It bears repeating - the information on the census isn't always correct. One common source for inaccuracies is when the census taker didn't find the family at home, and not wanting to have to come back on a different day, asked a neighbor for the information.
But even when the census taker found an adult member of the household, that individual could still provide inaccurate information. One strange example:
Below is a screen snapshot of the 1900 census record for my great grandfather, Melvin (M.E) Van Every, and his family. (Caldwell County, Texas, Justice Precinct 1, District 116, page 23)
Even those who don't know the family are likely to question whether Abigail is really a son. It's not too uncommon, however, for individuals to undergo a gender change on the census. That's not what this is about.
Reading the 1900 census: Melvin and Margaret had seven children ranging in age from 15 to 3 months: Minnie, Samuel, Abigail, Willa, Delbert, Eva and Myrtle.
Here's the list of children from the Van Every Family Record (which I believe came from the family Bible).
Abigail lived for two weeks in 1888, and Delbert was either stillborn, or died within 24 hours in 1890.
When the census taker asked my great grandmother or great grandfather to name their children...why would they name Abigail and Delbert as well? (They didn't name their son, Melvin Theodore, who had died in 1899.)
My first thought was that it had to have been one of them, as no neighbor would have known the months and years of birth for the children - which all match the Family Record perfectly. Then I realized -- if a child (or anyone else) answered the door, they could have retrieved the Bible for the information. Even assuming it was one of the older children, it's still a little strange that they would have given the census taker the information for two out of three of their deceased siblings.
[It should be noted that I am not relying completely on the Family Record, as Melvin did not list either Abigail or Delbert as children in his testimony in front of the Dawes Commission in November of 1900, five months after the census.]