Why do my results differ? Are you sure these are my results? My results are wrong?
I'm glad to help! In general there are lots of differences in the algorithms and the sample data sets used by genealogy analysis between services like 23andMe, Ancestry.com, MyHeritage, and FamilyTreeDNA. There are even differences between the various genealogy analysis offered in our marketplace.
There’s no single genealogy analysis that is the gold standard—instead, everyone has their own unique take on genealogy analysis.
For our marketplace, the "Genetic Ancestry with Haplogroups" app has many more sample populations (compared to other genealogy analysis) for European and Middle Eastern populations, while the "Ancestry and Genealogy" app has more representation for populations throughout Asia and the Pacific. A user who initially used the former app may discover some Southeast Asian ancestry if they use the latter app which didn't show up in the first.
While this doesn’t mean the Ancestry and Genealogy app will always provide most accurate analysis for people with Asian ancestry, it may provide more accurate analysis some of the time. These tools are presently only able to analyze based on sample reference populations available and get refined as additional information is made available.
When testing for cystic fibrosis, for example, we're looking at one gene (CFTR) and analyzing if a person is likely a carrier of, or affected by, cystic fibrosis. This is pretty straightforward. Genealogy analysis, however, uses tens of thousands of data points throughout the genome and each analysis uses a different set of data points (when comparing two different genealogy analysis algorithms, some data points will overlap but others will not). There is no defined algorithm for genealogy analysis and no defined population dataset for genealogical analysis that is the one standard way to analyze genealogy. Because of this, genealogy analysis is very much a combination of art and science, more so than with other genetic applications.