Tuesday, January 3, 2012

The Black Swan Fallacy

Modification of 2009 post

The Black Swan Fallacy is taught in college logic courses.

"I've never seen a black swan," the logic goes, "so black swans don't exist."

The logic is obviously absurd. No one has the opportunity to personally examine the color of every swan on the planet. That doesn't stop us from accidentally falling into this logical trap.

We may be very familiar with American swans, for example, and think our experience is sufficient. But black swans actually exist in Australia, if not elsewhere.

 [image source]

 There are also people who are very adamant that no two snowflakes look alike, or that there's no such thing as an honest politician. Science tells us the truth about snowflakes. Though I fear the existence of an honest politician is an element of faith one must either have, or not have.

Genealogical Black Swans

In genealogy we run across this fallacy when we assume something didn't happen, because we haven't seen the evidence for it.

I'll use an example from my family history.

Fact 1: Melvin Van Every, his wife, and children appear in the 1900 census in Caldwell County, Texas

Fact 2: Dates of birth for the children from a Family Record (believed to be copied from a family Bible) indicate births in Caldwell County in 1884, 1886, 1888, 1890, 1898, and 1900.  (With one birth in Oklahoma in 1892)

Fact 3: in 1900, in front of the Dawes Commission, Melvin Van Every testifies that his mother-in-law, Sarah (Hartley) (Denyer) Foster, was living with them in 1898 when she died.

Fact 4: The eldest child above has written notes indicating they only remained in Oklahoma briefly - between December of 1891 and Spring of 1892.

Most genealogists would look for records of the death of a Sarah Foster in Caldwell County, Texas.  Due to the scarcity of death records in 1898, there would probably be little effort made looking elsewhere when they weren't found.  Maybe you'd look in neighboring counties. However, if we catch ourselves stating with certainty that she died in Caldwell County, we have fallen for the Black Swan fallacy. We may have found no evidence to the contrary, but we don't have proof.

The Genealogical Proof Standard includes a "reasonably exhaustive search." Though we must remember reasonably exhaustive will still miss records. And truly exhaustive is next to impossible.

I was lucky, and a young Minnie Van Every wrote letters to The Houston Post between 1897 and 1899 (age 13-15).  She signed her name with the town she was writing from, so I know in November 1897 they traveled to Ganado, Jackson County, Texas, and if the Family Record is correct, they returned to Caldwell County prior to May 1898.

I now believe that Sarah Foster lived in Jackson County, and the Van Everys spent a year living with her. (Perhaps she was ill, so when she died, they returned to Caldwell. However, this is an unproven hypothesis.)  In the above map of counties in Southeastern Texas, I've colored in Caldwell and Jackson.  They are near each other, but probably not near enough that I would ever consider searching in Jackson if it weren't for those letters Minnie Van Every wrote.

There were enough missing years in the example above to raise some suspicion.  How about this for an example?

Entries from the St. Louis City Directory for my grandmother, Myrtle Van Every
  • 1921 - Astor Hotel
  • 1921 - 4528a Enright
  • 1922 - Westgate Hotel
  • 1922 - 4123 Westminster
  • 1923 - 4515 Washington
  • 1924 - 5630 Delmar
  • 1925 - 5540 Pershing
  • 1926 - 4506 Forest Park
  • 1927 - 4545 Washington
  • 1928 - 5707 McPherson apt 111 
  • 1928 - 5656 Kingsbury apt 203
  • 1930 - Georgiana Court Apartments, 5660 Kingsbury, apt 203, St. Louis, MO (census - ED 169 - Sheet 1B) 
She moved around a lot, but she remained in St. Louis.  She doesn't appear in the 1929 directory, but not appearing in a particular directory isn't uncommon.  I can imagine many genealogists, of varying experience levels, stating as fact that she lived in St. Louis continually from 1921 to 1930, citing the St. Louis City Directories and the 1930 census as evidence.

Unfortunately, it's not true.  In April of 1927 she was married in Oakland, California.  In October of 1927 they divorced, and Myrtle returned to St. Louis. She didn't miss appearing in the 1927 directory.  She appeared in the 1928 directory under her maiden name, and her married name.  And she did appear in the 1929 directory, but only under her married name. She returned to her maiden name by the census in 1930.

Since she worked for the US Postal Service, I was able to obtain her personnel records, which indicated her name change, but said nothing about geographical relocation.  Her brief husband was born in Illinois. not too far from St. Louis.  I'd have had absolutely no reason to conduct any research in California.

Once again, I was lucky, and my grandmother saved her divorce papers, which I found in a box of her effects after obtaining her personnel records.

Each document we uncover is but a snapshot in time.  Adding more snapshots increases our knowledge, but we need to beware of Black Swans nesting in our genealogical assumptions and conclusions.

1 comment:

Becky said...

Very good points, John. Making assumptions can get us into trouble - whether researching our families or even in our own daily lives!