Tuesday, January 08, 2008

Will the Real Julia McDavid Please Stand Up?

Some Issues Concerning the Evaluation and Analysis of Evidence

We've been playing a genealogical version of To Tell The Truth in which Julia McDavid, born in the nineteenth century and with a daughter named Helen, has challenged us to find her on the 1880 and 1900 censuses. The problem is that there are several persons with the name "Julia McDavid" enumerated in these surveys. And a confounding array of shifting information has made the task very difficult. How do we proceed?

First, we need to accept that we may never know from census records alone, or for that matter, from any record, who the "real" Julia McDavid was. ["Real" meaning the one who was born in the nineteenth century and who had a daughter named Helen.] There might be more than one person who fits our criteria or there may be no such person. But given this knowledge, we can move forward if we develop a hypothesis that there was such a person.

My hypothesis is that the Julias of 1920 Louisiana, 1910 Arkansas, 1900 Arkansas, and 1880 South Carolina are all the same person. So how do we "prove" a hypothesis? We need to know something about evidence, sources, standards of proof, and evaluation or analysis.

Genealogical texts and courses spend a fair amount of time on "primary" and "secondary" sources. This is important, no doubt. The problem is that some researchers confuse "sources" with evidence. This is like confusing the bottle with the wine.

Evidence and Sources

Direct and Indirect (or Circumstantial) Evidence. Direct evidence tends to prove a fact directly, such as the statement of someone who saw a bear in the wilderness. Indirect or circumstantial evidence tends to prove a fact indirectly, such as the statement of a knowledgeable person who saw only the fresh bear tracks. Either way, there is evidence that a bear was recently in the area. In terms of evaluation or analysis, it makes no difference whether evidence is direct or indirect. The evaluator may choose to believe or disbelieve either kind. Whether direct or indirect, the evaluator should give every piece of evidence whatever weight he or she think it deserves. The evaluator can accept more persuasive indirect evidence even though contradicted by direct evidence. And the weight to be assigned any bit of evidence may depend on the nature of the source of that evidence.

Census records are sources which may or may not contain evidence. For the most part, census records contain direct evidence. Now some of this direct evidence is hearsay, but direct nonetheless. [Example: Census taker asks householder how many children he has. Householder, with personal knowledge of his procreative activities, says he has ten children. Direct evidence! Example: Census taker asks householder how long she’s been married. Householder with personal knowledge of her personal affairs, says 17 years. Direct evidence! Example: Enumerator asks householder when she was born. Householder says “June 1872.” Direct evidence! Also hearsay! Householder is merely repeating what she’s been told by others (presumably her parents). Example: Enumerator asks householder when her parents immigrated to America. Householder says, “They were here the year of the big snow, but they weren't here the year of the drought.” Census taker knows that “big snow” followed drought by two years. Indirect evidence!). It's important to remember that the information in census records is only as accurate as those who give it. With respect to age, for example, some people may not have known when they were born. I know some genealogists resolve discrepancies by taking as accurate the date on the earliest census. In the matter of places of birth, enumerators may have assumed that if one parent was born in a certain place, so was the other one!


There has been developed in genealogy the so-called Genealogical Proof Standard which consists of the following:

Was the underlying research reasonably exhaustive?

Are there clear and accurate source citations?

Has there been a skillful analysis and correlation of evidence?

How well have conflicts in the evidence been resolved?

Is there a clear conclusion?

These elements are extremely important. However, it seems to me that some genealogical texts and courses leave out some other important elements, such what is the analytical process involved in the evaluation of evidence? And to what level of confidence do we want or need to "prove" the hypothesis.

Proof is always subjective, even when it seems that there is but one answer. One must realize that “proof” is something that happens in the minds of the analyst and those who read his or her work.If one is convinced of a fact to one degree or another, it has been more or less “proven” in that person’s view. There is the question of what degree of confidence must the issue be proven. In my view, this is entirely up to the evaluator. However, if his or her work is to have any credibility with others, the evaluator must inform the reader of the degree of confidence assigned to each fact.I find it helpful to use the following descriptions:

Reasonably Certain: A reasonable person evaluating the evidence would come to the same conclusion; any other conclusion would be unreasonable.

Likely: A reasonable person evaluating the the same evidence would find the conclusion more likely true than not not true.

Probable or Probably: A reasonable person evaluating the same evidence would find such a degree of probable truth in the conclusion that a reasonable person would rely on the conclusion as if it concerned his or her own affairs.

Possibly: A reasonable person evaluating the same evidence would find that the evidence strongly suggests the conclusion. These are high standards. Even when the evaluator is “reasonably certain,” the true facts could be otherwise.

Now back to the Julias.

Let’s start with the 1910 Arkansas Julia. I'm reasonably certain that she was born in the nineteenth century. Her age is given as 37; nobody could mistake a nearly middled-aged woman for a child of ten or less. There is indirect evidence that she has a daughter named Helen. Living in her dwelling with a number of other adults are just three children--one being an eight year old girl named Helen who shares the McDavid surname. I am comfortable concluding that this is likely Julia's daughter.

Now to the 1900 Arkansas Julia: Her age is given as 27 years old; ten years younger than 1910 Julia. Her birth place is the same as 1910 Julia's, although her parents' places of birth are different. There are differences in her race and marital status. And there is the apparent absence of her two children, Sterling and Clayton (Helen wasn't born until 1902). But none of those items as reported in 1900 are surprising given she was apparently working as a prostitute. I'm comfortable saying that probably the 1900 Julia is the same person as the 1910 Julia.

Perhaps the biggest hurdle is correlating the 1880 South Carolina Julia with the other two. The 1880 Julia would be 3 or 4 years older than 1900/1910 Julia. It's hard in 2008 to make an eyeball mistake of 3-4 years as to a child of 11 years old; but it may not have been so in 1880. But then, this is why some genealogists give more credit the earliest reported age for a child. At an early point with parents available, it's easier to tell whether a child was born in 1869 or 1872, than it is to tell whether that same person as an adult woman is 37 or 40 in 1910. The other evidence connecting the Julias is the South Carolina birth place of the 1900 father and the 1880 father. This is admittedly somewhat thin. I can say that 1880 South Carolina Julia is possibly the same person as 1900/1910 Julia.

Why not connect the 1880 Missouri Julia to the 1900/1910 Julia? Consider that Missouri Julia is essentially the same age as South Carolina Julia and, perhaps more importantly, her race is the same as 1900 Julia and her mother's birthplace is the same. So isn't it more likely that that Missouri Julia is the same person as 1900/1910 Julia? Maybe. But just like a juror hearing the SODDI ["some other dude did it"] defense, this does not quite ring true to me. An evaluator of evidence is not required to set aside his or her experience in the world, judgment, or intuition. Indeed, these things often are imperative to the evaluation of evidence. Nonetheless, the analyst must be able to point to evidence substantial enough to support a determination and must be able to articulate clearly the reasons for that determination. In this case, I would say that the 1900/1910 Julia and the 1880 South Carolina Julia are probably the same person. We should understand t hat we would not use only census records in the “real” world to make these conclusions.

Finally, we have to understand that, to tell the truth genealogically speaking, the real Julia McDavid isn’t going to stand up from the grave or even from the all the evidence we might collect. It’s up to us to construct “the real Julia McDavid” from what we are now able to know.

1 comment:

Lee said...

If only our ancestors could stand up and say, I am the real _!

Great article, Craig! I struggle with how much weight to place on evidence sometimes, and your explanation, in terms of possible, probably, likely, reasonably certain, really helped me to see it more clearly. Thank you!