Understanding ethnicity estimates - an analogy

[Mart 'n' Al]

Recently I've been reading a lot about DNA tests and ethnicity estimates. I am a statistician and try to explain things by using simple analogies. Hopefully the following explanation will help, but if anyone has any enhancements to suggest I would be most grateful.

I have a jar holding 100 blue marbles representing a father. I have another jar holding 100 red marbles representing a mother. The father is 100% blue and the mother is 100% red. When they have a child they each give 50% of their DNA marbles to the child, so the child will have 50% blue DNA and 50% red DNA. Any further siblings will also have the same 50-50 blue and red DNA.

Let's look at another example. I have a jar holding 100 blue marbles representing a father. I have another jar holding 50 green marbles and 50 red marbles representing a mother.  Any children they have will have DNA consisting of 50 blue marbles inherited from the father and a random selection of green and red marbles from the mother. That random selection could be 50 green marbles or 50 red marbles but is statistically more likely to be a random combination of both. The children would all have 50 blue marbles and a varying proportion of green and red marbles.

Now let's look at a more complex scenario. I have a jar containing 50 blue marbles and 50 yellow marbles representing a father.  I have another jar containing 50 green marbles and 50 red marbles representing the mother. Any children they have will have a random selection of 50 blue or yellow marbles and 50 red or green marbles in varying proportions. Statistically it is likely that the children will have roughly 25 of each colour, but extremes in either direction could and would occasionally happen.

Now let's take it back a few generations to either 64 or 128 ancestors. For simplicity we'll go for 100 ancestors!  Let's say that each of those 100 has 99 marbles of one colour and one black marble. Statistically there is a 50% chance that anyone person will pass the Black Marble on to their child. Then, statistically it is likely that many of the black marbles will not be passed on over several generations. However if we have 1 million sets of 100 ancestors then there is a chance that very occasionally the Black Marble will be successively passed down so that occasionally out of the 1 million sets somebody could end up with 50 or even 100 black marbles. Although it is statistically unlikely that if your parent has 50 black marbles and 50 black marbles, it is possible occasionally that you would inherit 50 black marbles, possibly even from both parents.

I hope this long but hopefully clear analogy explains how there can be such diversity in ethnicity estimates. It is therefore important to not just look at your current ethnicity but at the ethnicity of your ancestors, for many generations. As only a very small percentage of our DNA differs from person to person the extreme examples I have given are not quite so unlikely.

I would recommend the following 8-minute video on YouTube, which explains how we really only inherit our DNA from about 120 ancestors.  Try and avoid the numerous other videos from people failing to correctly interpret their ethnicity ESTIMATES!!

https://www.youtube.com/watch?v=9g-VwtRd8ks

Martin

[Gadget]

Lots of threads and links on this board about it:

http://www.rootschat.com/forum/ancestral-family-tree-dna-testing/


 

[Mart 'n' Al]

Thanks, Gadget, I just thought my explanation simplified it.

Martin

[mike175]

It surely depends on how you define ethnicity; the OED and many other sources define it as, "... belonging to a social group that has a common national or cultural tradition". I'm not sure that DNA can determine social and cultural origins 


And what if one of your ancestors lost their marbles? 

[Mart 'n' Al]

It doesn't really determine ethnicity and social backgrounds, but when you think how little people travelled 200 years ago you almost certainly ended up marrying somebody from a surrounding Village. This would be even more common in closed religious communities such as a mesh or certain Jewish groups. Also people living on Islands.

Martin

[Gadget]

I've had fun using Gedmatch modern admixture origins, using British Iron Age. I could have used a range of processes in the series but they all come out more or less the same

Diagram below shows my ancient past - pinch of salt time but all good fun!

Many years ago, I studied Anthropology

[smudwhisk]

It doesn't really determine ethnicity and social backgrounds, but when you think how little people travelled 200 years ago you almost certainly ended up marrying somebody from a surrounding Village. This would be even more common in closed religious communities such as a mesh or certain Jewish groups. Also people living on Islands.

Martin

Not necessarily if you have ancestors born in cities like London.  Definitely doesn't apply in my case with four grandparents born in London of whom many of their ancestors went into London in the early 1800s, and also a number of others who moved into London before 1800 (some before 1700).

Our ancestors were far more mobile than people give them credit for.  I've an ancestor whose family moved from north Hertfordshire to north Lincolnshire in the late 1500s.  Another who moved from Thrapston, Northamptonshire, to Lavenham, Suffolk in the mid 1600s about the time of the civil war.  Another born in Fife in 1793 who married in Gateshead in the early 1800s and then went down to London.  I can quote numerous other examples too.

While anyone with ancestors in the same area over generations may have married someone from the next village, many did not.

[sugarfizzle]

As I understand it, ethnicity results are supposed to give us an idea of where our long distant ancestors were from.

People saying they have gone back so many years and they are British through and through, wondering why they have Eastern European, or Scandinavian, or even XYZ do not appear to understand that they are supposed to represent 500 - 1000 years ago, or perhaps even further.

Look at all your American matches, some may go back to the Mayflower in 1620, perhaps.  My American matches are mostly estimated to have Great British or European ancestry. And that's going back 400 years in some cases. Paper records may or may not have survived, but their ancient ancestry is from Great Britain, which in itself is/was made up of various different ethnicities.

However, at present, as has been said many times, they are generally to be taken with a pinch of salt.

Regards Margaret

[Redroger]

It doesn't really determine ethnicity and social backgrounds, but when you think how little people travelled 200 years ago you almost certainly ended up marrying somebody from a surrounding Village. This would be even more common in closed religious communities such as a mesh or certain Jewish groups. Also people living on Islands.

Martin

Whilst this is generally true, we must never forget the effects of Wars, and military movements made as a result which certainly stirred up the gene pool. My 2x great grandfather joined the Somerset Fencible Militia in Bristol in 1793, and as it was a volunteer non overseas unit was able to leave it during 1796 in Alford, Lincs to marry a local woman. Whilst the unit was stationed there several other men also left to marry. On the other side, my mother's ancestors seem to have lived in a village (Bottisham) near Cambridge over several centuries and have extensively intermarried and interbred, though intriguingly my maternal grandfather has a gypsy surname. His ancestors had also lived locally for a considerable period. I have already had a Y chromsome DNA test a few years back, and now await the result of a test on my mother's side.