Tuesday, April 10, 2012

Accuracy in FamilySearch Indexing - My Results To Date

FamilySearch Indexing is sort of fun - I enjoy both the challenge and the feeling of accomplishment and helping the genealogy community.  However - there is JUDGMENT after every batch is submitted.

I thought that I would review the three batches of 1940 U.S. Census enumerations that I have submitted to date.  All three have been arbitrated already, which means that I, another indexer, and an arbitrator have reviewed them, and the arbitrator has selected the "final" indexed information where there were differences between the two indexers.

Here is the page of my Indexed batches to date (from the Indexing program, click on "Arbitration Results" and then on "Review Batches" which opens your browser to the FamilySearch Indexing page, you log in, and then see your results summary):


The three 1940 U.S. Census batches are at the top of the list.  My arbitration results are 99%, 97% and 98% for these three batches.

When you click on the batch of interest, the screen opens with the page image in the upper half and the indexing information in the bottom half.  On the left side is the Record List section, where the line number on the image is the second part of each number.  If there is an error in the line, the line number has an asterisk.

The middle portion of the bottom half is the "Field List" which highlights the error in the Record in brown.  The right side has "Field Help" for the specific field of interest.

In the screen below, you can see that I had an error on Line 31 in the Field for 1935 Residence State or Country:


For the batch that I indexed on 4 April from Colorado, I had these errors:

*  Line 31, Field = 1935 State/Country:  Randy = blank, Arbitrated = Colorado.  Correct!  I missed it.
*  Line 32, Field = 1935 State/Country:  Randy = blank, Arbitrated = Colorado.  Correct!  I missed it.
*  Line 34, Field = Relationship:  Randy = Wife, Arbitrated = Son.  Correct.
*  Line 36, Field = 1935 State/country:  Randy = blank, Arbitrated = Colorado  Correct.  I missed it.
*  Line 37, Field = Age:  Randy = 36, Arbitrated = 26.  Correct.

For the batch that I indexed on 5 April from California (Alameda County), I had these errors:

*  Line 1, Field = Age:  Randy = 44, Arbitrated = 66.  Difficult to decipher, probably correct.
*  Line 2, Field = Age:  Randy = 42, Arbitrated = 63.  Difficult to decipher, probably correct.
*  Line 4, Field = Relationship:  Randy = Grand daughter, Arbitrated = Granddaughter.  I typed what I saw, two separate words.
*  Line 5, Field = Line No.: Randy = 5, Arbitrated = blank.  Line 5 had no data entered, number crossed out.
*  Line 6, Field = Line No.:  Randy = 6, Arbitrated = blank.  Line 6 had no data entered, number crossed out.
*  Line 9, Field = Relationship:  Randy = Mother in Law, Arbitrated = Mother-in-Law.  I typed what I saw. *  Line 10, Field = Relationship:  Randy = Brother in Law, Arbitrated = Brother-in-Law.  I typed what I saw.
*  Line 11, Field = Line No.:  Randy = 11, Arbitrated = blank.  Line 11 had no data entered, number crossed out.
*  Line 12, Field = Line No.:  Randy = 12, Arbitrated = blank.  Line 12 had no data entered, number crossed out.
*  Line 21, Field = Birthplace:  Randy = Azore Island, Arbitrated = Azores.  I typed what I saw.
*  Line 29, Field = Line No.  Randy = 29, Arbitrated = blank.  Line 29 had no data entered, number crossed out.
*  Line 37, Field = Line No.:  Randy = 37, Arbitrated = blank.  Line 37 had no data entered, number crossed out.
*  Line 38, Field = Line No.:  Randy = 38, Arbitrated = blank.  Line 38 had no data entered, number crossed out.

Here is a screen showing Lines 5 and 6 from this batch.  There are no data entries on these lines, only notes.  The Line No. on the far left is crossed out:


I took note of the above before I did the third batch of California (Fresno County) on 8 April.   The very first line of my batch has no data on it, and the line number is crossed out:


The errors I made in this batch are (mostly Japanese, Chines and Yugoslavian names):

*  Line 1, Field = Line No.:  Randy = blank, Arbitrated = 1.  go figure!
*  Line 5, Field = Given Names:  Randy = Edna, Arbitrated = Edne.  Disputed.
*  Line 9, Field = Given Name:  Randy = Norman, Arbitrated = Normen.  Disputed.
*  Line 13, Field = Given Name:  Randy = Shie, Arbitrated = Shee.  Could be either.
*  Line 17, Field = Line No.:  Randy = blank, Arbitrated = 17.
*  Line 18, Field = Given Name:  Randy = Justine, Arbitrated = Justina.  Correct.
*  Line 22, Field = Given Name:  Randy = Teruishi, Arbitrated = Teruichi.  Disputed.  Looks like "s" to me.
*  Line 24, Field =  1935 State/Country:  Randy = California, arbitrated = blank.  Correct.
*  Line 27, Field = Age:  Randy = 57, Arbitrated = 51.  probably correct.  difficult to decipher.
*  Line 31, Field = Given Name:  Randy = Dioncia, Arbitrated = Dianda. Disputed.  Letters ran together.
*  Line 36, Field = Given Name:  Randy = Kikuo, Arbitrated = Kikus.  Disputed.  No doubt in my mind!

The screen below is for the given name Teruishi in Line 22.  I think it looks like an "s": and doesn't look like a "c":


For many of the ones that I disagree with the arbitrated value, I clicked on the "Feedback" link and saw:


The text in the popup box says:

We welcome your feedback. The most common causes for arbitration changes are:
  • Different interpretations of the original handwriting.
  • Not following the guidelines in the project helps.
If you believe you followed the guidelines in the field help, project instructions, and basic indexing guidelines and are confident your original answer was correct, check Please Review below to help us in the future.


  Please Review

If you have any questions or need clarification about how certain fields should be indexed, please click here to contact us. 

I clicked on the "Please Review" button for the items I disputed.  I don't know what kind of information will come back to me, but we'll see.

In summary, I see that I made some outright errors of omission and commission (ages, relationships, states), but I also had differences of opinion on the names and the line number problems.  For each of my problems, the Field Help was no help at all.

As shown above, the Line Number for a line with no information was required by one arbitrator, but was not required by another arbitrator.  I disagreed with the arbitrator both times.  I don't know if there is a standard for this.  I couldn't find it in the Resource Guide.  

I was dinged for three Relationships and one Country name.  Are there standard lists for these?  If so, will someone please tell me!!


UPDATED:  I found the "standard lists" for relationships and country names.  They are on the dropdown lists for those fields (thank you, Candace in comments).

There are 560 separate fields on each 1940 Census page (14 columns, 40 lines), and I had 5, 13 and 11 errors on the three batches arbitrated.  I disputed about half of them!  

My purpose in writing this was to highlight some of the Indexing problems that I've faced, to show the review process, and to raise the issue of the Line Numbers when there is no data in the other fields..

The URL for this post is:  http://www.geneamusings.com/2012/04/accuracy-in-familysearch-indexing-my.html

Copyright (c) 2012, Randall J. Seaver

14 comments:

Candace said...

There is a standard list for relationships. If you click on the drop down box in the relationship field and then select the bottom choice (I don't have a batch active now and don't remember the exact wording) you can select from a list.
My main complaint with the arbitrations is that the instructions say to correct the spelling of geographic locations, but whenever I have done that, the arbitrator has changed it back to what appeared in the record.

theKiwi said...

The list is called the "Lookup" list. As Candace says if you're using Windows, it's the last item in the menu that drops down when in say the Relationship field.

This doesn't work on the Macintosh, but in the Edit menu is an item "Lookup" - you can get the list open in a window there.

I've been dinged in the past with putting for example

Mother in Law

when it should be

Mother-in-law

Roger

Unknown said...

Hi Randy,

I've also been indexing the line # for a blank record and then marking all required fields "blank".

The Field Help says that we are to index the stamped line number for each entry line unless it was unrecorded or written as a variation of the word "unknown".

So I think you indexed correctly. And if it wasn't correct, then I've been indexing incorrectly also. =)

Unknown said...

So with the relationships always use the lookup (drop down lists) first and then if you can't find it in there type what you see.

Blank lines should not have the line numbers included so the arbitrator in the one batch who put them in was wrong.

As far as the review nothing will come back to you at this time. FamilySearch is working on a new indexing program that will have more features that the indexers have been asking for. Currently the review goes to a special team that looks them over and new updates are created from the reviews. It doesn't go back to the arbitrator and doesn't get redone unless the project has too many errors. Which has happened.

Otherwise the problems are with the field helps and the spread out places where the updates and other information are located.

If you're on Facebook check out our share batch group if you ever want more help.

https://www.facebook.com/groups/unofficialsharebatchfsi/

Cousin Russ said...

Randy,

Thank you for posting this. I was about to stop indexing. Too frustrating. 76% on a BLANK page. Guess I should try to find instructions rather then recording what is see.

The Family Numbers were a problem. I recorded the first column of numbers, when I should have recorded the second column on at least one batch.

In looking at the rest besides some typo's, was the alignment of the image on the screen. I could not get the Hi-lighter to line up with where I was working. I watched the folk at Family Search demonstrate how do to it, would not work, on my Mac, which is what they were using. I was taking too much time to line stuff up, especially if you scroll back or around to check something.

Lesson Learned Know the county structure in the state that you are working on. The 1935 "census" / where did you live in 1935, drove me crazy.

Oh well.

Russ

Oh, it's some times like reading the "please prove you're not a robot" for some of the census recorders.

Kimberly Powell said...

RE: Line Numbers The Project Instructions say "If an entry line on the census form is blank or has no data to be indexed, mark the record as blank in the data entry area." The important part of this instruction is "Record" as this refers to the entire record, including the line number (you can view each record individually in the Form Entry view). The easiest way to mark an entire record (line) as blank is to enter Ctrl+Shift+B while your cursor is in the line number tab location. The arbitrator that said the line number is required was incorrect to my knowledge.

Sharon said...

Randy:

As for what will happen when you ask for feedback -- nothing. You will not hear from anyone, no one will re-check the batch, and the entry will not be changed. This has been said repeatedly by FamilySearch personnel on the forums.

Gee, I thought they might take another look. My batches usually average 99%. All of sudden I had a batch that was 71%! Almost all of the difference was arbitrator error (not following instructions). I clicked on "feedback." Silly me.

The line number/blank record issue has to be fixed. The general instructions say type the line number. The 1940 specific instructions in the 29-page (Read the Fancy) Manual say leave line number blank. Field instructions should be specific on this.


Sharon

Elizabeth Handler said...

Basic Indexing Guidelines can be found at https://help.familysearch.org/publishing/347/102817_f.SAL_Public.html and a relatively newly published FamilySearch Indexer User Guide (a PDF) can be found at http://broadcast.lds.org/elearning/FHD/Local_Support/FamilySearchIndexing/EN/fsi_user_guide.pdf.

I agree, however, that there are times when I don't agree with what is arbitrated and I do click on "Feedback." I certainly hope someone is looking at those.

Unknown said...

Randy, thanks for posting about your indexing experience and how one can look at the feedback once your work has been arbitrated. I've indexed about 1500 records; my accuracy ranges from 95% to 100%. I've looked at the arbitration results. There is some inconsistency in the arbitration, particularly on the line number when there is no further information and on the residence items for children under 5. I know that everyone is doing their best to get it right! Sometimes the enumerator's handwriting is very difficult to read and it's just a judgment call.

Focus Grandma! FOCUS!! said...

Randy, Thank you for posting about your indexing adventures! ~laughs~ I also have had a terrible time with the lines numbers on blank lines. Seems like every arbitrator is editing different. I checked the instructions for the 1940 census and if the line is blank then you do NOT put a line number. I also click the review button.

Another problem I have had is with misspelling of the town, county or country. The arbitrators are not using the dropdown or "Lookup", some are using 'type what you see' even if it is incorrect.

Also the same with the surname and given names. Sometimes a letter or two or hard to read, the pull down 'Lookup' has been very helpful for me, but if the arbitrator is not taking advantage of that help and only using their judgement, I get it 'wrong'. I am glad for the review button, but not sure anyone is benefiting with that yet. Hopefully soon.

Hate it when most of my results are 98-100% and all of a sudden, I have an 82% and it is all line numbers! Line numbers are not an item at will be searched on, seems a shame that we loose accuracy because of it.

Anonymous said...

Thank you so much for this extensive--"what happened to me" with examples, etc. I also copied the links posted by others.Very informative.

Rorey Cathcart said...

Hi Randy,

In terms of the census, the format of 'relationship' has been particularly vexing for me. I've used the drop down/look up menu to fill in things like Mother-In-Law only to then have it arbitrated Mother In Law.

I haven't run into any conflict regarding how I identify blank lines but then again I've only come across that twice.

Mostly my arbitration differnces have been largley interpretive such as I say there was only one r in Doris but the arbitrator (and presumably the other indexer) saw it as Dorris.

Generally speaking I'm willing to shrug these minor differences off in any given record. However, I always push back on records - such as marriage licenses - where the 'filled in' name disagrees with the signature of the person named. I will always fight for the index to reflect the signed name. I understand the 'type it as you see it' convention but seriously, when the primary-direct is right there in front of you, use it.

Sorry - rant off ;-)

I've really enjoyed the census indexing though. I hope to fit in more batches as the week progresses. My last 6 were 4 99%, 1 97% and 1 100%. I'm striving for more 100% certainly.

Jeanette Drake said...

Thanks for sharing your experiences Randy. I have had similar ones. I called Family Search Indexing with a list of questions and after she patiently answered them all she referred me to the project instructions tab. There is click here link in the message and that will take you to project updates which are made on a daily basis. I noticed that many of the questions I had asked were answered there. There is also presentation just about the 1940 U.S. Census and I am going to use this when I sign up new indexers. It is helpful to open the project instructions and have it minimized on your screen so it is handy to refer to. I usually also have an Internet site to the localities minimized there also to refer to when I have a question about spelling the county or city, etc. And I have also had the experience of looking up the correct spelling and then having the arbitrator "correct" my efforts. I am a registered arbitrator and eventually I plan on arbitrating this project when I feel like I will be proficient.

Unknown said...

Thanks for sharing your experiences. I'm new to indexing. One trick that I've found valuable is to go to Google Images, and type in a name, and see if anything comes up. For example, Teruichi does indeed seem to be a common male Japanese name, whereas Teruishi does not. I agree with you that it looks like an "s", but it could also be a c, and Teruichi would seem to be a more logical choice based on the Google image results.