Many 1950 US Census records indexed/grouped poorly?
I have run into many messed up 1950 US Census records, including: (1) people in the main area being improperly linked to the additional information (sample) lines at the bottom, (2) the additional information at the bottom frequently being mangled text, and (3) groupings of households are frequently messed up.
The following link to "Carma Lee Lancaster" illustrates all of these:
https://www.familysearch.org/ark:/61903/1:1:6FSX-Y4DM
First, the highlight is too high, and it thinks her information is at the bottom. However, it is her sister "Patsy Suzanne Lancaster" (the entry above) that should be linked to the additional information below.
Second, the actual 1949 location for Patsy is "Camp Hood, Texas," but the text that comes through in Family Tree is "camphord, tunknown, tenas."
Third, the full household is Harold Duff (Head), Lenora Lee (Wife), and Patsy (Daughter), and Carma (Daughter). However, in addition to Harold showing up as a "Principal," both Lenora and Carma are also listed as Principals, and Carma is listed twice--both as a non-principal with the rest of her family and in a separate area as a Principal. I see these presumed grouping irregularities frequently, particularly when families are split across two pages.
Also note that Patsy does not show up as a daughter in Family Tree.
I also have tried to make some changes when ages or other information are incorrect, but they don't ever seem to translate to changes in the data in Family Tree. I still haven't really figured out how to fix any of this--I gather I largely cannot.
Answers
-
Carma probably doesn't show up in Family Tree because it's entirely possible that she's still alive. (You miswrote which daughter isn't in the tree.)
I think most of the errors you saw were the result of people trying to fix things but not managing to figure out how. I don't blame them; the interface is confusing as heck.
I think I've managed to mostly fix this family: I've made Harold the only principal, and entered his wife and daughters as his wife and daughters, and I've tried to move most of the highlights to where they belong. (That part's especially tedious, so there's room for improvement.) The only thing I can't figure out how to change is the order in which they appear on the Image Index sidebar and on individual index detail pages (such as https://www.familysearch.org/ark:/61903/1:1:6FSV-7KRC).
The part that's especially confusing is how to set up families: you can't enter the relationship that's actually on the census. That'll try to make the daughter into someone's parent. The "Relationship" field is in terms of the other person in the relationship, not the one you're actually editing. This could be made clearer. (Or be revised.)
As for your last paragraph: I'm not sure I understand. Are you expecting changes in a record index to somehow automatically translate into changes in the Family Tree? That's like expecting an annotation in the card catalog to also magically show up in the book on the shelf, so it can't be what you meant. Could you please clarify?
0 -
Hi Julia, thanks for your reply and for making some corrections to Carma's entry on the 1950 Census. My main point, though, was that a sizable proportion of the 1950 Census records seem to have one or more of the types of errors I reported in the Carma Lancaster example.
If you click on "SHOW CHANGE HISTORY" for her, it appears to me that all of the information was automatically generated, and the only user changes are the ones you made, so this is not a case of other users messing things up. I certainly agree with you, though, that the interface is very confusing. It also just seems that many of the errors cannot be corrected, and it seems like a lot more work needs to be done to generally clean up the systematic errors that appear in a sizable proportion of the records.
The household groupings are particularly confusing. The names do not appear in the same order as on the Census (the Principal is often in the middle of the list), and I have no idea why they are so out of order (it's not alphabetical). And I have not been able to find any good explanation of how the whole interface is supposed to work.
Specifically related to Carma's record, the link I provided to her 1950 Census record seems to be correct. I had entered her into Family Tree, as well, but she was not visible to anybody but me, because I had not marked her as deceased. I have done that now: https://www.familysearch.org/tree/person/details/GKB3-DHH
In my last paragraph, I was referring to what appears under "SOURCES" after adding it. In some cases, the highlight for a person's age is in the wrong row. For example, a mother may be 50 and a son below her 20, but the highlighting creates an indexed age of the son as 50, so the calculated year of birth is 30 years too early. I attach the source anyway, knowing that it is an indexing error. I then go in and edit the age in the source document to correct it to 20. However, when I go back and look under the person's sources, it still shows up as 30 years too early. Now, maybe I have to wait for a while for that update? Or, it just stays incorrect there?
0 -
SteveLinke and Julia Szent-Györgyi,
Unfortunately, when the highlights are off for a whole page and the question rows got indexed to the wrong people, it is more time-consuming than the initial indexing of the page. If only part of the page is done, it can throw all the other entries off if those who come later to link other entries don't realize this. For example, Line 30 for Alvin Adkins has data entered, but has no highlights.
As a 1950 Census FamilySearch Community Project volunteer when reviewing we were told to send pages like these back to the engineers with the note "highlights too high/too low." Requests sent through the Feedback button from the individual's 1950 Census page do go directly to the FamilySearch engineers (see instructions below). They have a list of corrections to be made. However, the turnaround time on making any corrections is unknown.
Report indexing errors by doing the following:
- Click EDIT in the indexed record to open the editing page.
- Click on the Feedback tab, usually shown vertically on the left-hand side of the screen.
- Choose an emoji.
- Enter the error or issue with the record in the box below the emojis.
- Click Submit.
I have fixed a whole page that has highlights off-kilter, and it can take over an hour (depending on the amount of entries and data) relinking highlights and reinputting text plus the question entries from the bottom (and I know now which ones to add). I wish I knew of a way to select all highlights and move them up all together to make this a quicker and more accurate process for those who know what they are doing. Then you get those cases where the paper was slanted, or the highlights were over two lines, and the data comes from both lines. So part is correct and part isn't.
Artificial intelligence is hardly perfect, but neither is human intelligence.
0 -
Thanks so much for the explanation. I will report future systematic problems through the Feedback tab.
0 -
I did fix the Camp Hood, Texas information. The "Unknown" in County is there because when indexing we are told to enter what is on the form. The end user (e.g., you) may be able edit and make corrections for their tree.
I know it can be frustrating. The 1950 U.S. Census was a new artificial intelligence collaboration with Ancestry and a new learning process for all involved in a short time span. While the indexing is done, FamilySearch engineers are still working on fixing some of the more complex problems like you brought up that aren't easily edited by a user.
0 -
From all of us doing our family trees, we appreciate all of the work you and so many others have done to make it easier!
0 -
Steve wrote: "I then go in and edit the age in the source document to correct it to 20. However, when I go back and look under the person's sources, it still shows up as 30 years too early."
Well, source titles don't get updated by later index corrections, but I believe the "Indexed Information" section does...? It does with the old index-correction system: I just checked it with a 1789 baptism, and the correction immediately showed in the attached source.
Oh: I just looked at Harold Duff Lancaster's 1950 census record, which uses the new correction system, and compared it to Haraszthy Károly's baptism, which uses the old system, and the two differ in what they show by default, and what shows only if you click the 'v': in the old system, the correction shows by default, while in the new, the old value is shown. Perhaps this is what you're referring to?
Another possibly-relevant consideration is that a birth year in a U.S. Census index is a calculated field, not an indexed one, and I don't know when/how such fields are recalculated.
0 -
Yes, if ages get corrected in a source, then the calculated dates should also get updated. That is exactly what I was referring to. Let's say the mother's calculated birth is 1850 based on her age of 50 in a 1900 US Census, and the son shows up as 50 (1850), as well. Then, the son's calculated birth should get corrected to 1880 after correcting his age on the Census from 50 to 20 years old. I will have to pay closer attention in the future if that actually happens--or maybe that takes a while? Thanks for your feedback.
0 -
If the age/year was calculated in a U.S. census, you can look at the date the enumerator's were supposed to be using (this was usually in the name field, see example below). The date did vary to different months. It does not appear in the heading of the 1950 U.S. Census, but is April 1, 1950.
Calculated ages/years are also why a census should never be used to verify a birth year (or month). Some are often off +/- by several years.
In older census records the birth country is another tricky one, if the country's name had changed or was occupied by another country during that census year (e.g., Poland).
0