Understanding DNA Matches
Better understanding genetic confidence intervals can boost YOUR confidence in assigning genealogical relationships to DNA matches.
One of the biggest problems we have as genealogists, and especially as genetic genealogists, is in confirmation bias. If we see an ancestor in the chart of a DNA match that we are pretty sure is also our ancestor, we might be tempted to say that your DNA connection with this match proves that this couple is your ancestral couple. You can use the shared DNA to support a connection, given other genealogical information, but DNA can’t ever really prove a connection to anyone outside of your immediate family.
Remember, DNA is just another record to use in your journey to document and tell the story of your ancestor, and it can be powerful, if you know how to use it.
Genetic v. genealogical relationships
One way we can better understand how to use our DNA matches to make connections in our family history is by understanding the difference between a genetic and a genealogical relationship, otherwise we end up making the wrong conclusion about our relationship to our DNA matches.
For example, let’s say that Meredith, a documented descendant of your 3x great grandparents Martha and Matthew, had her DNA tested. Your genealogical relationship to Meredith is 4th cousins. However, when the testing company completes the test, Meredith is not found on your match list! Does this mean she really isn’t related to Matthew and Martha? Not necessarily. Even though your genealogical relationship is 4th cousins, your genetic relationship can be non-existent as you will only share DNA with your 4th cousins about 50-60% of the time.
While you may not share DNA with a true cousin, you may also share DNA with someone who isn’t actually your cousin! You will share DNA with individuals who do not share a recent common ancestor with you, but who just share a similar heritage. For example, people from Italy may share DNA with each other not because of a single recent common ancestor, but because of a more distant connection to Italy.
Using DNA confidence intervals
There are a few ways you can distinguish the real matches (those who do share an identifiable common ancestor) from the population matches. The first is in the total amount of shared DNA. A good rule is that you need to share at least 20 cM (cM stands for centimorgan, a unit of measure for DNA) if you hope to find a recent common ancestor. You can see this total amount of shared DNA at every testing company. Additionally, at all companies except Ancestry, you can see the size of the biggest piece of shared DNA. The biggest piece should be 15 cM or longer to have a good chance at a recent shared ancestor.
AncestryDNA and MyHeritageDNA also provide their own internal analysis of how likely it is that you and another person can find their connecting ancestor within the last 6 generations. At Ancestry, each match is assigned a confidence interval of Very High, High, Good, or Moderate. They also provide a table to explain that a match in the Very High range has a 99% chance of being related in the last 6 generations, while a match at the Good level only has about a 50% chance of a connection in that time frame.
At MyHeritage they only label a match with a confidence interval when it is not High. So only Medium and Low Confidence matches will be indicated. If you don’t see a confidence interval, that means it is a High Confidence match.
Pay attention to the confidence intervals! This is putting the fancy math at our testing companies to work! While we can only see through our little window into the database, they have a panoramic view, and can spot trends in the data that we can’t see that affect our results. For example, they might see that there are 1000 people sharing that same piece of DNA with you that this particular cousin is sharing. They are going to downgrade their confidence in that match, because it looks like the DNA you are sharing is population-based (ie Italian) and not ancestor-based (ie from Martha and Matthew).
So by using the total amount of shared DNA (share at least 20 cM) and understanding the confidence intervals, we are one step closer to avoiding confirmation bias, and therefore become even better genetic genealogists.
Originally published December 2018 on genealogygems.com.