16 December 2013

Thoughts on browsing "Mastering Genealogical Proof"

Beware: this article contains unvarnished opinion. It is not a review of Thomas Jones's widely acclaimed book Mastering Genealogical Proof but thoughts prompted by browsing it.

The process called the Genealogical Proof Standard, is explained in the book as providing "a way to differentiate correct from incorrect information, to determine unspecified relationships, and to demonstrate that (genealogical) research results are credible."

With the GPS its axiomatic that if you follow the process you "minimize the risk of polluting sound research with dubious conclusions."

I have two concerns regarding this "textbook on genealogical methods and reasoning in the 21st century."

First, if anything says 21st century genealogy it's DNA evidence. Yet there is only a single paragraph on DNA evidence in the whole book.  As renowned genealogist Helen Leary wrote in 1998 “Science and the law are in agreement: there is only one way to prove kinships beyond reasonable doubt — DNA.”  At that time she had to add the caveat that such testing was not practicable. The 21st century has changed that. Spectacular successes in tracing kinships of adoptees, where documentary evidence was unavailable or unproductive, is just one area that bears witness that DNA deserves more than this passing mention.

Second, the text, and the GPS, skirt the question of expressing confidence. The emphasis is on resolving conflicting evidence. A section on page 75 deals with unresolved conflicts.
"Stopping short of proof, we may state that the point is unresolved, summarize the related evidence, and explain why the conflict is not resolved. We can express a belief that one side of an unresolved conflict is more likely correct. The discussion should make it clear to readers that we are presenting an opinion, not a conclusion from evidence. Recognizing that in such cases we have no conclusion or proof – all we have is a possibility – we qualify the discussion with words like perhaps and possibly."
Chapter 1, Genealogy's Standard of Proof , makes it clear that "standards stop short of absolute certainty." A proof is still liable to be overturned by new evidence. So the difference between a "proof" which is not absolutely certain and unresolved conflicting conclusion possibilities is the degree of confidence. The solid line drawn between proof and absence of proof is an invention of the genealogical establishment.

As Jones writes in his preface clients look to a professional genealogist for a determination of the reliability with which genealogical findings reflect the past. Is a finding nearly certain, highly likely, probable, somewhat likely, etc? Unfortunately this book, and the GPS, fail when it comes to providing any such standard. A client cannot be confident that a professional genealogist's expression of reliability is based on a reproducible standard.

9 comments:

Paul and Rona said...

The word "proof" is deeply problematic as a description of the kinds of inferences drawn by family historians from oral and documentary records.

Many of us understand the word to connote mathematical rigor. For example, most scientists, trained in mathematical thinking, shrink from ever asserting that a scientific theory has been proven (although disproof is entirely possible).

Genealogy owes its use of the term to another intellectual tradition, possibly derived from philosophy or the law, in which a "proof" is nothing more than a persuasive argument.

At best the term is confusing.

I find it absurd that some genealogical fundamentalists have denounced Bayesian methods as implying a false level of accuracy or certainty--which is not the case, by the way--at the same time as they're brandishing the wildly misleading word "proof" with respect to word arguments.

How about "Genealogical Inference Standards"?

Unknown said...

John,

I couldn't agree more about your point re: DNA. DNA is far too prevalent - and powerful - to leave out of any discussion on GPS.

But, I'm predicting that 2014 will be the year that the "DNA and the Genealogical Proof Standard" revolution begins. For example, see Debbie Parker Wayne's classes on the subject at our 2014 GRIP course. And I'll be lecturing on the subject in 2014 at a time & location I can't share yet, but it will cover exactly the issue you raised.

Unknown said...

An excellent critique of a glaring problem in our industry. I've said it before in other forums: Words Matter.

The use of the word "proof" in genealogy has always given me pause. My degree is in Biology. In science, medicine and law, the term “proof” has been so thoroughly bastardized in the layman's mind as to be useless as a stamp of accuracy/reliability.

In genealogy, we operate on the knowledge that any newly discovered piece of evidence could alter our conclusions. That is the purpose of a reasonably exhaustive search, to minimize (but never eliminate) that chance as much as possible. We do ourselves and our industry a disservice - and give clients a false sense of accuracy - by using a word like "proof". Which, I might add, we proceed to contradict in our reports with the heavy reliance on qualifier words such as "likely" and "probably". Consider these statement side by side: “we’ve proved our hypothesis” vs. “we’ve probably proved our hypothesis.” Doesn’t instill the same level of confidence does it? Nor should it.

In our personal genealogy to which we dedicate decades, we might come to something approaching the reliability of proof. But in the client relationship, given the information obtainable in the amount of time they have afforded us to search we don’t always have that luxury. Instead, we use our years of experience, our subject matter expertise and our analytical skills to reach a conclusion that is still, in truth, just our “Professional Opinion.”

Tony Proctor said...

I agree with the terminology surrounding "proof", but not with the suggestion that DNA evidence can prove kinship. Just as with other forms of evidence, it can support a claim of kinship, and even allow a level of confidence to be attached to it, but it does not constitute a proof. On the other hand, it can certainly disprove kinship when there is none. This is the nature of scientific evidence: it can support or disprove but it can never prove. In other words, there is no such thing as a 'scientific proof' in reality.

JDR said...

Tony: If two people match exactly in mitochondrial DNA and have identical DNA across one of each 22 autosomal pair what other kinship possibility would you suggest except mother and child?

Edward Black said...

"what other kinship possibility would you suggest except mother and child" ... well, you have not considered the whole concept of adoption. Marriage is a legal contract and we say it makes a family. Adoption is also a legal contact, and believe me it makes a family. There are plenty of women who have never laid eyes on their DNA, never held their DNA in their arms, never taught their DNA right from wrong, etc.

I'm not making a pejorative comment about a woman who has given up a child, I'm saying they ceased to be a part of the same family. They may reestablish a relationship. Evidence would demonstrate a relationship.

Dan Stone said...

John: While extremely remote, couldn't the relationship in the case you described possibly be aunt and niece/nephew? The supposed mother could have a twin sister. While I agree with the point you make about incorporating DNA into any type of "Genealogy Proof Standard," I think Tony's point is that even with DNA the result is presented as a degree of confidence level rather than as an absolute proof.

JDR said...

Dan: As I understand it, it would depend on whether the twins were dizygotic (fraternal)or monozygotic (identical). With fraternal twins you'd expect on average to find 25% sharing with an aunt. With identical twins a traditional autosomal test might not be able to distinguish which twin was the mother. I recently saw an article that found a few SNPs were different even between identical twins.

Your point is well taken that even so no "proof" is absolute, just very much more likely than any other alternative, perhaps to a degree that could/can be scientifically quantified.

JDR said...

Edward: I recognize your point and could better have phrased it as "what other kinship possibility would you suggest except biological mother and child"