Thoughts on Correcting Errors in the Records

LegacyUser

Robert Kehrer said: Situation:
FamilySearch offerings do not today allow a user to annotate or correct an error in the record transcription, nor is there currently a means for anyone internally to do this for the user.

Status:
We clearly understand the need, it is on our roadmaps and drawing boards and a significant amount of time is being invested to design, test and build such a capability. No expected date has been announced.

A few thoughts:
Transcription (indexing) errors happen for a number of reasons and in some cases are pretty much unavoidable no matter how good the software or our eyes are today. These errors are found in every record collection and are introduced by every indexing system out there. Here's a few reasons:

1) There may be a mismatch between the language of the record and the indexer's knowledge causing diacritics to be missed or combined characters to be misinterpreted
2) The handwriting may be nearly incomprehensible, nearly being the operative word, and the transcription is attempted but transcribed wrong
3) The census taker may have written down what he heard rather than clarify the name's spelling
4) The office census worker who transcribed into the official record what the door to door census taker wrote may have introduced spelling errors (they had to read the guy's crazy handwriting too).
5) The variant spelling may actually have been used by the new immigrant for a while after they arrived in the country because that is what some employer or immigration worker wrote down.
6) any number of others

To minimize the risk of additional errors being introduced into the records during indexing a redundant indexing with arbitration system has been devised. Each record is key twice by 2 different indexers. Entered data that mismatches between the two is sent to a third indexer that has been designated as an arbiter based on skill to make the final call. Statistical analysis has shown that this process results in very low novel error rates. Unfortunately nothing can programmatically address errors written into the record around the time of creation.

When a patron finds their ancestor and knows that a name is incorrect they logically want to correct it so that it can be properly searched, found and recorded by those who come after.

The FamilySearch system does not currently allow a user to make corrections or annotations on a record. Nor does it provide a way for a user to submit these errors so that someone internally can make the correction. This user need, however, is well understood and a significant amount of time is being invested in mapping out a system that would allow user corrections/annotation to a record. We cannot yet announce a date when such a feature will be available, but it is on the roadmaps and drawing boards.

-Robert

LegacyUser

Pat Mann said: I would like a way to send in corrections to data. Changing Lawrason to Lawrison or Tilley to Lilley can make someone miss a family member. I hope a way to correct mistakes can be up and running soon.

LegacyUser

Beverly Taylor said: Yes, transcription errors happen. Hope you will eventually be able to work out a way to allow changes, maybe with a double or triple check on the suggested spelling of the original document. It might be somewhat biased to the new interpretation but it might also act as an "aha" moment.

LegacyUser

Jade said: This is a great idea, and the means for correcting mistaken extract / index items has been needed for a long time.

This should, however, clearly be differentiated from "corrections" to the actual record. For example, some living descendants of an individual may not be aware of the many sound-alike variants for a surname that were used during the 18th and 19th centuries in original records. Some of the items on this message board complain about wrongness of an actual entry in a given record.

There should be some way for users to make a comment concerning contemporary usage, or their opinion as to "actual" spelling of grandmother's name.

There should similarly be a way to add an annotation, that would be added to the index, where the original record is wrong in some crucial way. I am thinking of some instances where the copy sent to the Census Bureau reversed first name and surname of head of household, and the firstname was given as the other household members' surname.

LegacyUser

C Haddock said: I ditto what Jade wrote and more. I have uncovered an instance where the indexer "changed the surname" of person, from that shown on the Census Record. Why? Probably because the entry appeared confusing to them.
Example: 1880 US Census, Philadelphia.
Mary Richardson (a widow) had her younger brother, Robert Hill, his wife, Margaret, their two children, and Mary & Robert's very elderly father, Robert Hill living with her. So, what did the indexer do? Changed Robert Hill senior's name to read; Robert Richardson (thinking he was Mary Richardson's husband), but he was in fact her father. If this has happened once, it has happened on other occasions, and there needs to be a facility for righting such errors.

LegacyUser

Jim Bowden said: I applaud this effort.

LegacyUser

C Haddock said: Thank you, Jim. But that is only half of the story. Omitted from the entry on Mary Richardson (418 Gaskill Street) was her other brother, William Hill, 35yrs, Porter. William's details are found on the following page of the 1880 Census. (Wm appears to be visiting the local Shoemaker, Tandell.) The indexer totally overlooked the detail that; William Hill also belonged to the Richardson & Hill family group.

LegacyUser

Corlis Hicks said: When it becomes possible for users to submit corrections, an explanation and or evidence needs to be provided to show the the correction is...correct. If the evidence is insufficient and stronger evidence or an indirect proof is required, perhaps the error can be submitted to FamilySearch users as a whole. I would welcome the opportunity to improve my research skills by helping someone prove/disprove a pending correction.

LegacyUser

Wayne Morris Wright said: Most of the above comments were directed toward correcting errors in the original records, commendable, but almost entirely separate from the corrections of errors in the process of transcribing the imperfect records into the index.

'Correcting' errors in the original records can be taken care of by leaving the records as is, [that is what they say] but adding a comments section where users could enter a 'but should have been... click to see documentation' out where it can be seen, and by making the record searchable both by the original entry and by the 'but should have been' name, date, or place.

The transcribing process could he raised to a much higher level by feedback to the indexers as to their error rates and error types, and by having the arbitrator look at all entries not just those that differ. Duplicate errors are frequent. We hope they are of minor importance, like recording items the instructions say to skip, or expanding abbreviations the instructions say to record as is. Still, if indexers are encouraged to view all entries, there will be less of these error duplicates..

LegacyUser

C Haddock said: Wayne, the suggestion of a 'wiki' like feature sounds a great idea. National and State Archive are beginning to introduce such a feature. Here's just one example.http://wiki.prov.vic.gov.au/index.php...
Joy to those who can discover 'errors on originals' - spelling mistakes are common. But more power to those who discover errors made by the transcribers. Yes, Wayne, more quality control and more feedback to errant transcriber.

LegacyUser

Wayne Morris Wright said: It is not giving me an edit option,but I should have said If arbitrators are encouraged to view all entries there will be less of these error duplicates surviving.

LegacyUser

R Galloway said: If a head of household name is incorrect for a census and they are difficult to find in that census year...check for one of the known children or the spouse. This can help locate a head of house whose name may have been difficult to read and transcribe and is indexed incorrectly. This has been helpful in putting all the records together to make for a complete family with annotations as to the variations! Thus we are correcting them as we find them in the family groups. (Yea for Beta that helps with the variant spellings on the surnames!)

LegacyUser

Gene Kuechmann said: It is urgent to provide a way for users to have transcription (index) entries corrected. Re-examining the record would often be all that is necessary to verify the correction.
It might also be useful, but not as important, to provide a way for users to point other researchers to other (more correct?) sources.

LegacyUser

anzenketh said: This is great news.

Perhaps a very easy way to do this is add a Edit Field up by print that sets the fields to editable.

Then to view other opinions you could either go the way of having a icon to notify their is another opinion about this or Have a view other opinions link under the view image link.

LegacyUser

C Haddock said: Urgent - is the operative word here. Sadly, I continue to uncover transcription errors (when compared with the original document) and I do not mean spelling mistakes. The last find was people on the US Census being described as - Chinese - when they were in fact - Colored.

LegacyUser

Mary Susan (Carlson) Scott said: If we cannot make corrects to the indexing, we should (at the very least) be able to add a comment to the document for various reasons.

(1) correct transcription errors
(2) clarify information
(3) add relationship information

There are probably other reasons as well.

This morning I spent hours looking for the death record in Chicago of my great-grandfather's brother.

I finally was able to browse the death certificates and figure out a way to find John J. Osterhaus.

The surname was indexed as Osleitevus which is a terrible misreading of the name.

I admit that the handwriting on the death certificate was horrible but this was a bad reading at best. (The crossing of the "t" was through the "h".)

Another problem is that we cannot sort the list in the findings of a search. There doesn't seem to be any rhyme or reason to the lists -- they are NOT by date, they are NOT by surname.

On the GenLias website (Netherlands archives), we can sort by dates, record types, surnames, given names, places, etc.

FamilySearch software engineers should be finding ways to help researchers find exactly the records we need.

The hours I spent finding this one death record could have been used more productively if FamilySearch was more friendly to the research process.

Thank you for listening.

Mary S Scott

LegacyUser

Maggie Mcdade said: World Connect, as well as other databases, allow for post-ems. Originals are not changed but additional information, corrections can be made. Works pretty well, I think.

LegacyUser

Jade said: Margaret, WorldConnect is a web display of individual trees, not a records database. The post-ems can be helpful where based on evidence that is stated, but any information in them is not incorporated in the owner's tree except at their whim. They are text messages, not linked or encoded to relate to any particular datum on a given page. So they can not be incorporated in any index.

In general, trees are not a vehicle for doing evidentiary research, although there are a very few sterling examples of WorldConnect trees that supply good documentary evidence concerning the assertions presented.

LegacyUser

Paula said: When I work in Ancestry.com and get to a census record for example, it does allow me to indicate any errors and it keeps the original record the way it is and then all future searches show the corrections/additions from people. This has been extremely important as the names get butchered all the time since they are French.

LegacyUser

Diane Elaine Barnes said: I have problems in that the whole family was based on a lie. The man I thought was my grandfather took my true grandfathers name. It is a mess, I have the documentation to clear it up and want to get the right people sealed to each other.

LegacyUser

Peggy Stevens said: Hope this will be up and running soon. It's available on ancestry.com so it's not rocket science. In 1930 census, my father's surname STEVENS was mistranscribed on both ancestry.com as Stmary and on familysearch.org as Stomero. Since I clearly know the family members and the last name I can see Stevens, but understand a transcribers error. But you can imagine how long it took me to find the census data. Forturnately my father had an unusual first name, as did his father, and a first name, no last name search limited to the state where I knew they lived proved successful.

LegacyUser

Jim Carter said: Please provide a way for a user to submit proposed corrections to errors so that someone internally can make the correction.

LegacyUser

sue rowe said: Hi, judging by the date of the census (1880, whereas the British census was taken in 1881) I presume you are talking about an American census. I cannot speak for the way the American censuses were conducted but in the UK the census taker only recorded who was actually in the premises at that time...not who's usual abode it was, so if the brother was indeed visiting the shoemaker when the census taker came round he would indeed be listed as being at that address. e.i. a 'visitor'...so it may not have been a mistake at all and not for the indexer to place him with the Hill family... (;o)

LegacyUser

Deanna Denis said: hAHAHA I found my great-grandmother listed as Bennie..and my great grandpa's name mispelled too...and 1 of their daughters...if i hadn't done alot of cross-referencing i would have never found them. BUT--I did find out what the 'S" stood for in my Uncle's name! THANKS

LegacyUser

Edward Thomas Watts said: The gender of John M. Tools is a female in the ORDINANCE records of the church.
John M. Tools is a male not a female. Please change the ordinance records from
female to a male.

Edward T. Watts

LegacyUser

Mary Susan (Carlson) Scott said: I have some of these situations too. This is probably not the best way to have the person's gender changed. There is a Help Center document (Document #108337) which provides the steps to have the gender changed in New FamilySearch. If the person is male and the ordinances were done for a female, I am pretty sure that the ordinances will need to be redone. See the help document and/or call FamilySearch support (1-866-406-1830).

LegacyUser

Lynne Farr said: Having a way to correct errors is very important and I hope someone will devise a way to do it soon. My gr-grandfather's birthdate is wrong due to a transcription error in the 1850 Census. I have two family Bibles and a court case to prove the correct birthdate. A search will yield incorrect results of 1786 rather than 1807. That's a big mistake. I'd be happy to correct this if I could, but can't. I can't stress the importance of this enough.

LegacyUser

Kris Kahler said: On your search I finally found my grandmother whose last name was Mrugalski. The 1920's census has it as Mujlski. Ancestry.com matches yours but has a way to submit a correction.Heritage Quest has her last name as Muglski and no way to submit a correction. So there is a difference in the websites. Hopefully you will find a way to submit corrections. It took me hours and I mean hours to find her. It is important that we get corrections made. I sent a correction to Ellis Island and they corrected it within two days.

LegacyUser

Leonard McCown said: One thing that would certainly help folks doing indexing is an option that they can enlarge the handwriting to read it. I find most of the transcription errors I find are clear once it is enlarged. Jackson Co., Arkansas Marriage, 25 June 1876, Dascrida C. Dowden and F.M. Blackwood, is clearly Mrs. Dorinda C. Dowden and F.M. Blackford. Indexing is serious business and should be taken that way! Leonard J. McCown (leonard@mccown.org).

LegacyUser

Paula said: Great idea! It's so easy to be able to enlarge on computers now that if they make that option available, it's much better.

LegacyUser

Mary Susan (Carlson) Scott said: I believe it is already possible for the indexers to enlarge the handwriting of the original document.