GENEALOGY-DNA-L ArchivesArchiver > GENEALOGY-DNA > 2004-12 > 1102428402
Subject: Non-paternity rate [was Re: [DNA] Father-son Study by Univ of AZ]
Date: Tue, 7 Dec 2004 09:06:42 EST
In a message dated 12/07/04 4:26:58 AM Pacific Standard Time,
> There is a paper from earlier this year by Anderson of University of
> Oklahoma on paternity confidence levels and the % of non-paternal events in
> father-son tests. To summarize briefly, he used data from 66 published
> studies to establish how frequently non-paternal events happened in
> father-son test subjects. He divided his finding into 3 groupings based on
> the test subjects confidence level of their paternity. The High Paternity
> Confidence group included those participating in 22 genetic projects which
> he assumed would have a higher confidence level in their paternity since
> many were lineage studies . Their average rate of non-paternity was 1.9%.
> The second group was of Unknown Paternity Confidence level from 14 studies
> from which confidence levels could not be determined. The rate of
> non-paternity for this group was 3.9%. The last group of 30 studies was of
> Low Paternity Confidence level. It included people who had been tested
> because of paternity dispute issues they were involved in. The average rate
> of non-paternity in this last group dramatically jumped to 30.2%.
THANK YOU, Eagle Eyes! I've been looking for a study of that caliber for a
long time. So many of the more sensational claims (e.g. 30% non-paternity rate)
can be traced back to studies done in paternity testing laboratories,
obviously not a representative sample. From the results of various surname studies, I
had been more comfortable with a rate of 2% or so.
I found the poster presentation on Google, then the full-length paper at
Kermyt Anderson's home page. It has been submitted for publication in Current
Anthropology, so that explains why I haven't seen an abstract at PubMed yet.
As we've mentioned before on this list, Sykes' study on his own surname,
dating back to the 13th century, came up with a non-paternity rate of 1.3% to
account for the distinct cluster of haplotypes still remaining today.
> No matter which figure you prefer to use, if you extrapolate the odds out
> over several generations, it becomes quickly apparent that non-paternity
> events could easily be discovered in many surname projects. Consider a
> hypothetical example of 3 descendent who all share a common ancestor exactly
> 10 generations back. Based on Anderson's study there would be an average (at
> best) of 1.9% chance of a non-paternal event for each of the 30 births (10
> generations x 3 lines) that would seperate these 3 lines from their common
> ancestor. Its no sure thing, but if you were betting on the test results in
> Vegas, you should put your money on them not all 3 matching each other.
The way to calculate this is to consider how many times true paternity must
hold up. The probability of that occurring is .981^30 (.981 to the 30th power),
or a 56% chance of all three being true descendants of their putative common
Ann Turner - GENEALOGY-DNA List Administrator
Search or Browse the archives, Subscribe or Unsubscribe at