Surname Tests - Realistic Expectations

[DevonCruwys]

Lindsay DNA testing and results is based on probabilities and therefore not 100% Therefore some supposition is indeed used.

Devon A-DNA at 5 generations has I believe only a 10% accuracy and therefore not worth anything.

Rob, I'm just replying to the above two points you raise. You are right that DNA testing can only ever give you a probability that two people are related within a given time frame. However, the more markers that are tested the more confident we can be in the predictions. For example if two men take a Y-chromosome DNA test and match on all 67 markers, the predictions are that in 90% of cases those men would share a common ancestor within the last five generations:

http://www.familytreedna.com/faq/answers.aspx?id=9#923

We do not use DNA testing on its own and instead we use it in combination with the documentary evidence in order to draw conclusions about our genealogical research.

The autosomal DNA tests look at 700,000 SNPs (markers) and offer highly accurate relationship predictions in recent generations. They are much more accurate than the legal paternity and relationship tests which use no more than 16 autosomal STR markers. An autosomal DNA test can conclusively prove a parent/child or sibling relationship and can identify first and second cousins. Relationship predictions become more difficult beyond the second cousin level but that is largely because of the random way in which autosomal DNA is inherited which makes it more difficult to predict the relationships. Cousin marriages can also confuse the picture and make relationships appear closer than they really are.  When two second cousins take an autosomal DNA test it is 99% certain that they will share enough DNA in common for them to show up as a match. If two fourth cousins take an autosomal DNA test there is only a 50% chance that they will share enough DNA in common to have a match. For fifth cousins the probability is down to 10%:

http://www.familytreedna.com/faq/answers.aspx?id=17#628

See also these autosomal DNA statistics:

http://www.isogg.org/wiki/Autosomal_DNA_statistics

Autosomal testing is therefore much better for confirming close relationships. However, you still get matches with lots of distant cousins simply because we all have thousands and thousands of distant cousins. Again the DNA is used in combination with the paper trail.  The test will give you a match with a predicted fourth cousin and you then have to go back to your family tree and try to work out how you are related.

[KGarrad]

. DNA testing does, however, have a very legitimate use for genealogy as described earlier.

Could somebody, in one or two simple sentances of plain Engilsh without jargon, please explain what that is?

Lizdb,   Doesn't look as if your going to get 1 or 2 sentences in plain English, does it?!

[lizdb]

I rest my (our) case.

[acorngen]

Devon,

It is interesting that you and the link you provide is at a direct conflict with people who work in the field on a daily basis.  You do however agree with what I stated about A-DNA.

With regards to the 67 marker test and the 90% accuracy.  There is still that 10% chance that the results will be wrong.  The question this raises though is how unique is DNA or rather the markers and the pattern these markers produce?  In criminal cases you are give a figure of 1/?? chance of this being another person.  In the US a few years ago a man was convicted of Rape based on DNA evidence.  After serving a number of years the police arrested another male for rape who admitted to the rape the first male was convicted of.  When his DNA was tested (I do not know to what degree and I cant now find the media piece I am paraphrasing from) it was an exact match of the first male.  Whilst this is rare it does lend to the problem of being able to say I am of that family does it not?

Rob

[DevonCruwys]

Could somebody, in one or two simple sentances of plain Engilsh without jargon, please explain what that is?

DNA testing can be used as a tool in combination with documentary records to verify relationships within a genealogical timeframe.

[lizdb]

How?

[DevonCruwys]

It is interesting that you and the link you provide is at a direct conflict with people who work in the field on a daily basis.  You do however agree with what I stated about A-DNA.

Could you possibly clarify and provide links to the information that is in conflict with the data I provided? Who are the people you are referring to?

With regards to the 67 marker test and the 90% accuracy.  There is still that 10% chance that the results will be wrong.  The question this raises though is how unique is DNA or rather the markers and the pattern these markers produce?  In criminal cases you are give a figure of 1/?? chance of this being another person.  In the US a few years ago a man was convicted of Rape based on DNA evidence.  After serving a number of years the police arrested another male for rape who admitted to the rape the first male was convicted of.  When his DNA was tested (I do not know to what degree and I cant now find the media piece I am paraphrasing from) it was an exact match of the first male.  Whilst this is rare it does lend to the problem of being able to say I am of that family does it not?

Probabilities are just that. They give an indication of the possible range of dates in which the common ancestor might have lived. If there is a 90% probability that two men will share a common ancestor within five generations this simply means that in 9 out of 10 cases the common ancestor will be found to have lived within five generations and in one case the common ancestor will be found to live outside of those five generations. This does not in any way imply that 10% of the results are wrong. That is what we would expect on the evidence available. There is a 50% chance that if you toss a coin you will get a heads. That doesn't imply that there is anything wrong with the the result if you happen to get a tails instead.

Forensic DNA is a different matter altogether. Forensic DNA tests use 10 (in the UK) or 13 (in the US) autosomal STR markers. The autosomal tests used for genealogy use 700,000 autosomal SNP markers and are much more accurate than the forensic DNA tests. I seem to remember that there was a case in Arizona where two random men were shown to have identical DNA signatures. The forensic DNA databases have also never been opened up for inspection by independent experts. You might like to read some of the links here:

http://www.isogg.org/wiki/Forensic_DNA

[DevonCruwys]

How?

You might like to read this article I was commissioned to write for Sense About Science:

http://www.senseaboutscience.org/blog.php/41/sense-about-genealogical-dna-testing

[lizdb]

Its more than 2 or 3 sentences!

I am not trying to be awkward, but everything I have seen about DNA testing just reads like gobblygook to a lay person (and I am not totally thick, do have a basic knowledge of genetics etc), and those who are fans of it can write at great length, but to the run-of-the mill family historian, it just goes over the head.
I cannot see any worthwhile use in it, but seeing as people were saying it has a very legitimate use in genealogy, I was prepared to be openminded and invited someone to tell us exactly how. How does it have a use, how can it be used as a tool?