Correlation and Validation

Eureka! You’ve found the record you’ve been looking for. It’s right where you expected it to be. The names, dates, places, and situations are consistent with the people you’re researching. You’re tempted to celebrate, thinking you have the information you need. But you’re a responsible researcher. And responsible researchers don’t celebrate based upon something found in just one source. Celebrating must wait. There’s more work to do.

Because any source can be wrong, conscientious researchers check multiple records that should or may provide the same piece of information. They compare and contrast their findings from those sources. This process of comparing and contrasting is called correlation. It is an important element of research, for there is no way to tell if a piece of information is likely correct or incorrect unless it is compared to other data found in other sources.

Looking at the number of times a piece of evidence is found in different sources is only part of the task of correlation. It is important to remember that the sources being compared and contrasted should be unrelated. That means that they should not have been created one from the other and different people should have provided the information in those sources. For example, if the same person provided information for a man’s death record and obituary, those two sources are not independent. Similarly, if a clerk who completes a marriage certificate copies information from the couple’s license, the license and return are dependent. In these examples, even though the same information is found in multiple sources, that fact cannot help a researcher evaluate accuracy because the information came from the same person. Examining independent sources and analyzing independent pieces of information is essential.

Most researchers understand the need to test answers found in one source by comparing them to answers found in other places. When the answers are directly stated, the comparisons are straightforward. Parents’ names listed on a death record can be compared to those listed on the person’s marriage and birth records, for example. But correlation is also needed when the answers are hazy—for example, when names aren’t directly stated, but clues point to an answer. In any case, researchers must work to find evidence that either supports or disproves the proposed answer.

Researchers who are interested in learning more about correlation will enjoy Brenda Dougall Merriman’s 2006 article in the National Genealogical Society Quarterly entitled “Validating Inferences from Records: Jane Baker and Thomas Burnett of Kingston, Ontario.”[1] In her effort to establish a maiden name, Merriman searched for a woman’s marriage record around the year 1842 (the year before the woman’s first-known child was born). She found a possible match in a December 1842 marriage. The man’s name was correct, as was the woman’s given name. All elements seemed to fit the basic outline of the family—names, date, and location. The record listed the bride’s surname, which seemed to answer the question about her maiden name. Or did it? It wasn’t specifically stated to be her maiden name and the record did not list the names of her parents.

Merriman knew that when looking at any source it is important to keep any open mind and consider all possibilities. Although the 1842 marriage record was intriguing, she couldn’t be certain of its relevance. And she recognized that even if it was relevant, there may have been more to the story than the single record revealed. Merriman explored all members of the family as well as their associates. She traced the couple’s children from cradle to grave, and along the way she discovered evidence that helped assemble the puzzle pieces.

The target couple’s son died in 1895. His obituary mentioned an uncle and three surviving sisters. By exploring those identities and relationships, Merriman discovered that the dead man’s mother had been married twice and had children with both husbands. The woman’s surname listed on the 1842 marriage record was that of her first husband. Armed with additional details, Merriman looked further and eventually uncovered the woman’s name at birth.

In some situations, direct evidence can be found in multiple sources. In others, such as the research problem faced by Brenda Merriman, direct evidence is lacking. Absence of direct evidence doesn’t mean a research question can’t be answered, as Merriman’s work illustrates. Based on information from the marriage record, she inferred a possible answer about the woman’s maiden name. She sought evidence in independent sources, compared and contrasted her findings at each step, adjusted her hypothesis, and eventually arrived at a sound conclusion.

All researchers rejoice over exciting finds, but careful researchers remain on task. They work to correlate their evidence and try to validate their theories. Had Merriman halted her research when she found that 1842 marriage record, she would have misidentified the bride’s family name and would have overlooked additional offspring from an unnoticed marriage. She understood the importance of thorough research and painstaking correlation. She continued to work. Through persistence she uncovered additional clues from independent sources that helped her see the entire picture—and that is certainly worth celebrating.

[1] Brenda Dougall Merriman, “Validating Inferences from Records: Jane Baker and Thomas Burnett of Kingston, Ontario,” National Genealogical Society Quarterly 94 (December 2006): 259–66; PDF, NGSQ Archives (http://www.ngsgenealogy.org/cs/ngsq_archives : 30 January 2017).

Leave a Reply

Your email address will not be published. Required fields are marked *