Why Your Ethnicity Estimates Change

Many users log into their AncestryDNA accounts after a year only to find their "30% Irish" has suddenly become "15% Irish and 15% Scottish." This is not an error; it is a reflection of how ethnicity estimates actually work.

The Reference Panel Problem

Your DNA is not compared to ancient historical populations. It is compared to modern people living in those regions today whose families have deep roots there. When 23andMe says you are 40% Italian, they mean 40% of your DNA sequence matches the statistical profile of their modern Italian reference group.

Worked Example: The Smoothing Algorithm

Imagine a chromosome sequence mapped across windows. The algorithm slides across the chromosome and assigns a probability to each segment. If it sees: German, German, French, German, German, the smoothing algorithm will likely correct the "French" segment to "German" because the populations are genetically highly similar (both falling under broadly Northwestern European).

As companies collect more samples from underrepresented regions (like central Africa or Southeast Asia), their statistical models become more confident, allowing them to break "Broadly West African" into specific regions like "Senegambian."

Common Mistakes

Do not treat ethnicity estimates as historical fact. They are probabilities with confidence intervals. If you look at the detailed data, a 5% Scandinavian estimate often has a confidence range of 0% to 15%. If the low end is 0%, it is entirely possible it is statistical noise.

Frequently Asked Questions

• Why do my siblings have different ethnicities? Because of genetic recombination. You each inherit a random 50% from each parent. Your sibling might have inherited the segments mapping to Irish populations, while you inherited segments mapping to German ones.

Next Step: Use the cM Calculator to confirm actual relationships rather than relying on ethnicity.