Confirmation Bias

I am my best example of bad practices! In this case, confirmation bias. It created a false brick wall of my own making. Only recently did I see how it delayed the outcome of the right answer.

But, first, what is “confirmation bias”?

Confirmation bias is where you have some assumption of an answer to your question and you set out to prove it….you set out to prove it without thinking about what other answers there are to this problem.

Let’s say your research question is “Was John F. Smith the father of Mary Smith, b. 1895 in Avon County, MA? In this research question you are assuming the John F. Smith was the father and structured your question around John. Your research focus will be on the unknown father, John F. Smith. Having crafted your research question in that manner, you blithely set out to find all the John Smiths in Avon County, only to discover that no one had a child named Mary!

“I have a brick wall!” you wail.

NO, YOU DON’T!

You instead have committed confirmation bias.

Research questions should start with known information. Reword your research question to “Who was the father of Mary Smith b. 1895 in Avon County, MA.” Now, you have shifted the focus from what you didn’t know (the father) to what you do know (the year and place of birth of Mary). You also shifted your research focus from John to Mary.

Of course all the above is fiction, but serves to illustrate my point.

We can also commit confirmation bias any time we are looking for a DNA match for a bio-father. We see the same surname and decide that is the one. We can discover later, after making the “big reveal” to unsuspecting and proposed relatives, that the person changed their name when they, too, were adopted. And the real DNA link was the next person down on the list.

We can commit confirmation bias any time we assume an answer and set out to prove it without checking the other candidates.

I am my best example of bad practices. I have worked on and off for years hoping to identify the mother of Margareta Andersdotter, b. 8 February 1750 and her place of birth. I knew she was born in 1750 because the 1817 Swedish census said so. There were other sources that gave me a range of 1749 to 1753.

I committed confirmation bias by trying to prove she was born on 8 February 1750. I focused on that date and made the range of years I would look at extremely narrow. While I was working though all the possible parishes, I failed to notice the importance that one location had a year missing, 1752. It wasn’t what I was looking for.

I decided to build a time line and things started falling into place–every activity of the family was revolving around 1752. Oddities in the records came into focus like I had just changed the prescription on my glasses! Those oddities could now be interpreted with more meaning, rather than just someone’s mistake.

The article is now submitted to the Swedish American Genealogical Quarterly for their possible publication.

Here are some things I do to help me avoid confirmation bias:

1. I write up my problem ancestors in research reports. This consolidates my research in one place, including my research plan, my research log and my notes in one place. It focuses my mind on one problem, avoiding “bright shiny objects.”
2. I work slowly. Some of that is because I am a mediocre analyst and it takes sometime for me to internalize the evidence I do have. I cite everything, which slows me down but reaps great benefits later.
3. I work with hypotheses. My language reflects that the options are open. I work each hypothesis to prove it viable or not. It’s a “last man (or woman) standing” approach.
4. I build timelines to help me see what I have and what I don’t have. I can see where there are gaps and overlaps.
5. I assess every record set I use for completeness at a macro level (does it have that year?) and the micro level (is it missing pages)?
6. I work hard at trying not to come to a conclusion too quickly. It closes doors to other options.

I hope this has helped you. I suspect that I will commit this “genealogical sin” in the future, but I will (hopefully) catch it sooner.

Take care; stay safe,

Jill

What I have done since the last post. I am in Boston awaiting the birth of our next (and probably last) grandchild, Charlotte Emma Shannan. ❤