But the general point is well made.
Nobody has a personal memory of where they were born. They just know what they have been told ... IF they ever asked ... which they might not have done. If they didn't ask, they might have made an assumption, which could be correct or could not.
In the modern day and age, exact date of birth is important to so many things, and we keep meticulous records. Before the welfare state, however, nobody was avidly counting the days to their 60th or 65th birthday when they would qualify for their state pension. In the days before modern education laws, nobody was carefully tracking the date on which a child should start school, or before which they could not enter the workforce. In the days before modern laws on alcohol, tobacco, glue, blade and other sales, nobody was obsessively interested in whether somebody could prove that they were over 18 or not.
So in an age when nobody cared that much about exact age, few people monitored it closely (consider the scene in "Far From The Madding Crowd", when they are in the pub trying to figure out JUST how old the old man actually was ... Hardy was not trying to write a humorous parody ... this was genuinely the kind of way in which people would try to figure it out).
And then, once every ten years, somebody comes along and asks "How old are you? Where were you born? How old is your wife? Where was she born? How old are each of your children? Where was each of them born?" People couldn't remember exactly ... and neither could they remember what answers they had given last time around. So the answers given may vary from census to census. Not really any great surprise there ...
I have always worked on the basis that the most accurate information as to approximate date of birth and place of birth is to be found on the EARLIEST census that a person appears in, unless that is the 1841 census in which case the 1851 census is likely to be more useful (especially if 1841 says they were not born "in county"). That is closest to the event when memory is likely to be freshest and, let's face it, you are FAR less likely to make an erroneous report that a 1 year old is 3 or 4 than you are to make an erroneous report that an 11 year old is 13 or 14.
And of course, the less schooling the person making the census return has had, the greater the likelihood of egregious errors. If they "made their mark" on their marriage certificate ... well ... they're not going to be keeping written records in a family bible somewhere that they can consult when the census form pops through their letter box, are they??