Common Census Errors
Home > Genealogy > Reference > Census > Census Errors
The Federal Census records are a wonderful source of family information, but like all records, they can (and often do) contain mistakes. To me, census records usually fall into the category of a Primary source, meaning the data obtained was obtained from a witness, a reliable source, someone intimately associated with the information. As noted below, there are exceptions that can change a census record into a Secondary source.
Census takers went from door to door and interviewed one or several family members in order to obtain the census data. Generally speaking, I'd imagine they talked to the man of the house, or the lady of the house. If neither of these were available, they might interview one of the older children or other residents. As you can imagine the accuracy of the data decreases depending on who the census taker got the information from. If from father or mother, then the data is probably highly accurate. If from a child or other household member, then the data might only be fairly accurate.
The exceptions come about from a variety of causes. What if the only person at home was a hired hand or domestic servant who wasn't related to the family. They might not provide the most accurate information. Another likely scenario is that the family was not at home when the census taker came around. If this were the case, if the family was known to the census taker, he might estimate all the ages of the family members, and inadvertently leave out members he didn't know about. If the family were unknown to the census taker, then he would probably ask a neighbor about them, and then the results could be quite off quite a bit. Do your neighbors know your exact age and birth place? And that of your other household members?
Another problem comes when the family is being deliberately evasive about ages and birth places of origin. This might be done for a variety of reasons, but it has been known to happen. People then, as now, were not always forthcoming with personal information, especially when the government is involved.
So, in a nutshell, the accuracy of any census entry depends on who the census taker got the information from. Unfortunately, the census records do not make any notation as to who gave the data in each case, so we have no way of determining the original source.
The above does not explain all census errors. Some are caused by the census taker himself. For example, the spellings of names would almost always be written the way the census taker thought they should be spelled which might not be the families preferred spelling.
Another possible problem is with the Race or Color category. Census takers used their own opinions when filling in this category. (See my "Race Codes" page for more on this subject).
The worst census problem of all is one that most people are not aware of. The census records that you look at (usually on microfilm or digital microfilm or online) on not the original census records!. The census taker made the census record as he went from house to house. He may have used pre-printed forms or just wrote on whatever paper was available. When done he probably compiled the records onto the pre-printed forms each night or much later on. He was required to make another copy to be submitted to the Federal Government. He did this by hand and it would seem it was often done hurriedly and carelessly. The new copy (full of new errors) is what we have now. The original copies were mostly discarded, or filed away locally, and lost over the years. Some remain, but as far as I know, there is no easy access to them (at least not all in one place).
Other researchers have made comparisons between the original census records (where available) and the "copies" we have now, and have found numerous transcription errors. In some cases the surnames as originally noted were totally different on the copy. Given names were different, ages, birth places, etc. were often changed from the original to the copy. I'm sure the census taker did not do any of this intentionally, but it happened, and it is one of the several problems we have with the census records we use.
Based on all the above, any census record should be considered a Primary Source, but one that might be either highly suspect or highly accurate. It is always helpful if the data can be confirmed from another source (family bible, other records, including other census records). Of course, unfortunately, sometimes a census record is all we have.
Now, as you probably know, the 1790 through 1840 census records did not give the names of everyone in the family, only the person considered the head of household. The 1850 and later census records gave the name of every freeperson living in each house. In 1850, 1860, and 1870, no relationship information was given, so, even though all the names were given, it isn't always clear how the family members were related. Marital status was not listed until 1870.
It is often assumed (in 1850, 1860 or 1870) that if there are two adults (man and woman) and several children, that this represents a husband and wife and their children. This is probably correct in most cases. However, those children listed might actually be grandchildren, or nieces and nephews, or even cousins. Starting in 1880 the relationship to the head of household was given. That makes it easier to figure out the family structure.
Another common problem is that of different surnames in the same household. A houseful of people may all be listed with the same surname (either spelled out or with ditto marks), but some of those persons might actually have a different surname but not be listed as such. I've noticed this problem quite a few times.
There are probably other types of census mistakes that could be noted. I believe that someone has written a book on this subject, but I have not seen it, though I imagine it would be quite useful to have.