"Sunday and Monday were our best indexing days this year," according to a statement issued by FamilySearch.org. "We indexed a combined total of two million records [and] arbitrated almost one million." Over 14,000 volunteers contributed to the record. Over 450 were first timers. About 180 people indexed over 1,000 records.
"Thank you," said the statement. "It's amazing what we can accomplish together in behalf of researchers around the world."
You probably know that FamilySearch uses double key indexing. Each record is indexed twice, called an A Key and a B Key. The two keyings are compared and corrected when necessary. For its 1910, 1920, and 1930 census probjects, FamilySearch is using indexes from Ancestry.com as the A Key. FamilySearch volunteers key the information a second time, the B Key. Anytime the two differ, the record is sent to an arbitrator to look more carefully and resolve the difference.
For the 1930 census, Ancestry.com did not index several key fields, including gender, race, and marital status. FamilySearch is adding these fields. Consequently, indexers will sometimes get 1930 census batches that ask only for the additional information. These batches supplement the existing A Key. Batches for the B Key will ask for all the fields.
"We indexed a combined total of two million records [and] arbitrated almost one million."ReplyDelete
Wow! A 50% disagreement rate.
AI, you say "For its 1910, 1920, and 1930 census probjects, FamilySearch is using indexes from Ancestry.com as the A Key."ReplyDelete
Is FamilySearch discarding the many instances where Ancestry.com has made up relationships that are not given in the enumerations? Such as where a grandchild of head of household is listed, Ancestry.com ~always~ picks someone else in the household to be parent of the granchild, if someone of suitable age is present. If there are only an additional male and female, they pick the male. Oh, so often wrong, wrong, wrong, not least because a parent of the grandchild may not be in the household at all.
Anonymous, arbitration and indexing are two separate activities that take place at different times for the same batches. The names being arbitrated are not necessarily the same as those being indexed on a given day. For many collections (especially international collections) the indexing tends to go much faster than arbitration. Arbitration then catches up near the end of the project, resulting in many more names being arbitrated than indexed at that point of the project.ReplyDelete
Geolover, I think the answer to your question is that these errors will be cleaned up. The FamilySearch indexing process "requires" indexers to enter the information, including relationships, as recorded. I believe that "relationship" is one of the fields where the indexers will be doing both the A Key and the B Key on these records. The same is true on the line number, family number, and place of birth fields.ReplyDelete
Jim, thank you for your reply.ReplyDelete
You say "The FamilySearch indexing process "requires" indexers to enter the information, including relationships, as recorded. I believe that "relationship" is one of the fields where the indexers will be doing both the A Key and the B Key on these records."
In the situation I gave, Ancestry.com has invented fields for their extracts (which they call "the record") that do not exist in the enumerations, concerning parent-child relationships between members of the household in addition to relationship to head of household.
In uncommon instances the enumerator did write in some details on this point, but not in the examples I know of where Ancestry.com specified a person (erroneously) as parent of the grandchild.
At least Ancestry.com has deleted the invented "ethnicity" field for the 1900 US Census, which was another invention (but by whom I do not know).
Thank you again.