Saturday, May 16, 2009

Ancestry Quirks - the missing Janes

A reader named Judy sent me a fascinating, and disturbing, Ancestry quirk. Usually, I can analyze them and explain how things happen, but this one is a real mystery to me. Here's the problem:

Judy knew that her Slone Janes resided in Sonoma County CA in the 1910 to 1930 time period, because she has census records and a World War 1 Draft Registration card for him. So it made sense to look for him in the California Voter Registrations 1900-1968 database on Ancestry.com. When she did, there were no matches for any surname "Janes" in the records. She tried the first name (since it is rather rare) and got no results, and eventually found her Slone Janes by going page-by-page through the Sonoma County listings - so she knows that he's there!

I said to myself "hey, that can't be right..." Wrong. I started my search in the home page search box, and input the exact last name = "Janes" and pressed Search (see screen 1 below):




The California Voter Registration database does not have a Janes indexed, so it is not listed in the Matches. Am I sure? No, so I went to the Card Catalog, got the California Voter Registration database on my screen, and input "Janes" in the surname field. The result was:


Nothing. Nada! Zero!! Were there no Janes who signed up to vote? At this point, I asked Judy to tell me what page he was on so I don't duplicate her tedious page-by-page search through hundreds of pages. Rather than answer my question, her reply said that she used the Keyword field on the Search form last night to find him again. The Keyword field? It takes names? Who knew? I didn't!

Ever the glutton for new search techniques, I input the surname "Janes" in the Keyword field on the Search form. Here's what I got:



There are 4,131 matches to the keyword "Janes." Are they all the surname of someone, or maybe a street name? I don't know. I'll put in "janes slone" (with quotes) in the Keyword field and see (I chose last name first because that was how all of the names were listed - alphabetical, last name first):


Hey -- four matches, all in Sonoma County between 1914 and 1922. But are these the right person? I clicked on one of them, and here's the page with the name of Slone Janes on it:




There he is, the 5th one down in the left column, a barber, resides at 135 Howard St. and a Republican.

Being the out-of-the-box thinker that I am, I decided to check some of the other persons on this page - maybe this page was skipped by the indexers for some reason. I picked Sine Jacobsen, Achille Kahn, Sidney Lehman and Leune Johnstone to see if they are indexed (becuase they have unusual names - surely the indexers can't mess up TYPED names!). The search for Sine Jacobsen, Achille Kahn and Sidney Lehman, using the first name and last name fields in the Search form, found them easily - they were indexed correctly and the pages are linked correctly. However, anyone searching for Leune Johnstone in Sonoma County using the name fields would be fooled into thinking he didn't register to vote. Like Slone Janes, he is easily found using the Keyword field.

I did not make this a senior project today - I didn't check out the 4,131 Janes entries from the Keyword field matches to see if they were first names, last names or street names (or something else - ah, I got it! James spelled wrong!!!). I'm guessing that they were mostly last names.

To verify that, I input "Janes" in the given name field and got 4,131 matches! That's the same number I got when I put "Janes" in the Keyword field. Wrong guess... I did a random sampling from the first 100 matches and only one out of ten had the last name Janes. The rest were James entries that were misspelled. By the way, there are 394,970 matches for first name = "James" and none for last name = "James."

Is this typical of the indexing for the California Voter Registration lists? Sadly, it is. It appears that the name boxes are just there to provide two names to search for. If I put "John" and "Seaver" in the two name fields, I get 58 matches (only some of which are a John Seaver), but if I put "seaver john" in the Keyword field, I get 17 matches of actual "John Seaver" persons. If I put "seaver" in the first name field, I get 0 matrches; I get 1,056 if I put "seaver" in the last name field; I get 1,056 matches if I put "seaver" in the Keyword field. Now that is different from the Janes entries, isn't it? Somethings not right in Provo!

Many researchers have noticed the quirk in the name indexing for this database - it finds pages containing the first name and the last name, but not necessary persons named "First Last."

My reader found one way around this for this database - use the Last name first in the Keyword field and put the name in quotes. More good news - wild cards work in the Keyword field with quotes - e.g. "Seav* Joh*" found 52 matches, including John Seavers and other matches.

My advice is to search using only the Keyword field, and if you are searching for a known person use last name first then first name and put quotes around them, and use wild cards if necessary. At least until Ancestry.com fixes the indexing.
Judy's last line in her email to me said "It really makes me wonder what I'm missing on my quest to find other family members." Amen! Many of us wonder the same thing about the online genealogy databases.

Thank you, Judy, for an interesting exercise. I know that quirks like this are frustrating for users.

At the NGS Conference yesterday, Ancestry.com said that they were spending money to fix databases with missing content, erroneous indexing and/or wrong links. I hope that Ancestry will move this database to the top of the list to be indexed accurately and completely. All of us California researchers will appreciate it!

4 comments:

Becky Wiseman said...

Randy, if you put "janes" in the first name field and "slone" in the last name field and select "sonoma county" and check "exact search" using "old search" you get 4 results. All are for Slone J. Janes, barber.

Mel said...

I blogged about this awhile back. I ran into the same problem with name searches in the CA Voter Reg indexes at ancestry. When I couldn't find anyone, I tried the keyword search. Bingo!

If you have the street name, try a search like this...put in the county, then put in the keyword box the street name and a surname:
"east 25th" souza
I got all the pages with souza and East 25th Street (which is where all my people were). Much more helpful!

Unknown said...

A similar, but different, problem with the U.S. Civil War Soldier Records and Profiles. Searching "Quattlebaum" (soundex or exact) - if you fill in the "Home State" field no results are returned. Leave it blank and there are 59 results.

Lots of folks have complained since this database was announced in Feb - but no fix yet!

Craig Manson said...

A similar problem exists with the US City Directory database! Very frustrating.