Why are ages rounded up on the 1900 US Census?
Before I post this question in the Indexing Group, I thought I'd ask it here because this little problem is so pervasive that there might be a reason for this other than errors by the indexers.
It has to do with the rounding up of ages that appear on the 1900 US Census but it applies only to those individuals born after the actual date of the census (June 1st). For example, an individual born in Dec 1895 is indexed as 5 years old but the actual data on the original form is correct in that it shows 4, the last birthday of that person.
This rounding up of the indexed age for those born after June 1st happens to 100% of the listings anywhere in the 1900 US Census, not just a few people or a few areas.
Surely not every single indexer made the same exact mistake and ignored the data on the record and arbitrarily rounded up the age. So there must be some sort of algorhythm going on here that supercedes the indexers' entries. Did I not get the memo about this or is there something else going on?
Example:
https://www.familysearch.org/ark:/61903/1:1:M9Z8-SGR?from=lynx1UIV8&treeref=L9HQ-VVK
The head of the family is indexed with an age of 41 but on the actual document his birth date was entered as Sep 1869 and an age of 40
This anomaly continues through several of the children in this family, those born after June 1st.
Answers
-
lots of people didn’t know their exact age , Many times the person giving the information would be a family member, landlord or neighbor; census enumerators were instructed to round off the numbers if “exact” age couldn’t be obtained. Also, the dates had to correspond with their age as of the “official census date” which wasn’t always the date listed on the form. Please see attachment
0 -
0
-
Thanks for that image of instructions to the census enumerators but it does nothing to address my queswtion of why the indexing age has been increased one year over what what written on the original image document. Your image of instructions was the same given to the enumerators every census in that the age must be reported at the date of the census (and not the date of the enumeration) and that the age was to be that attained at the last birthday (and not the closest next birth date). It also talks about enumerators rounding ages off such as "about 70" whenever the exact age was not know. That was not my question but rather how or why FS is rounding up the ages above what was already reported by the enumerators.
0 -
Brian,
It might be as simple as them just subtracting the year of birth from the census year. It is a simple and fast calculation that doesn't involve the month and day of the month. Also, some of that index data may have been calculated after the fact by computer once all the manual inputting of the index data was completed.
And X24mom is very correct in that when you go back a couple generations the older someone was, usually the more incorrectly they remember their age. If you want more reliable ages, check the censuses closer to when they were born. It harder to mistake the age of a 3 year old than it is an 83 year old😊
0 -
I agree, Jeff, that certainly appears to be the computational method in which the ages were indexed ignoring the months and day of the month. But that would mean that the computer doing the calculation would somehow have to know the year of birth. And how would it know that without somebody inputting the index data. And then that methodology is not used on any of the other census years --- only the 1900 US Census. I'm trying to figure out as to why this one particular database was handled so differently than all the others. Its just so un-nerving to notice the age on the index record does not match what was entered on the original image. I've seen that instruction so many times while indexing records: "enter only what you see on the image". For some reason the 1900 Census index stands alone.
0 -
Here are two theories to mull over.
1) Maybe when the census was first indexed, the age was not included for whatever reason someone writing the instructions thought was a good idea at the time. Then maybe later on someone else realized this was a mistake and the ages were calculated from the birth year.
2) Maybe when the census was first indexed, the age was indexed but in post processing of the data set was modified, again for some reason that seemed reasonable at the time, to change the age from age at last birthday to age as of Dec 31, 1900.
In any event, other than going crazy trying to figure out what some now retired data manager could possibly have been thinking, I don't think this is a major issue since I never trust a birth date in the census anyway without confirmatory evidence.
0 -
true - that we dont rarely trust a census age.
but why make things worse . . . pretty weird.
0 -
Well, for what it's worth, the 1900 US Census *IS* unique in the fact that in addition to the age at last birthday, it is the only one that also includes the month and year of birth. When you include the full date of the census page for a given record (i.e., day month and year), the year is redundant, meaning that incorrect entries can disagree with other entries in the census.
0 -
You can only do automatic processing of indexed data. For some reason in ALL of the US Census pages that I have ever looked at, even though the full "Event Date" for that page of the census is almost always shown at the top of the page, only the YEAR has been indexed. THAT's what forces any calculations to make assumptions. And as Gordon points out, what were those assumptions?
So if the "Age at Last Birthday" was 10 on a 1900 Census page, the theoretical birthday could be calculated by computer to "from 1 January 1889 to 31 December 1890" (a span of TWO years). But if the FULL Event Date had actually been indexed from the Census page, the span would only be ONE year. E.g, if the Census for that person occurred on 1 July 1900, then the theoretical birthday would be "from 2 July 1889 to 1 July 1890".
So from a normal age given in an indexed US Census, the best you can do in estimating birth is a two year span, but if the full event date had been indexed, that accuracy would be reduced to a one year span. The fact that the full event date was never indexed typically adds a two year inaccuracy.
Now the month/year format uniquely given for the birthday in the 1900 census can be more accurate but since the event date is only indexed to the year, it is possible for the age information to "disagree" with the month/date given in the 1900 Census.
And NONE of this takes into account all the errors from people losing track of their age as they get older, Neighbors answering census questions for the absent family, or ages for young couples who recently got married and lied about their ages on their marriage licenses (a VERY common practice int the areas where I've researched).
For indexing purposes, it doesn't matter so much because only approximates are necessary to allow indexed data to be digitally searched for. But for record conclusions (which should NOT be included in indexed data), you have to go to the source images to get a better handle on what was really happening.
Unfortunately, it seems that in order to automate more things, companies are putting "preliminary" type conclusions in their indexed data (such as standardized places that don't match the actual indexed location at all).
0
