Comparison of three Surname Projects

Image:Long Boone Cumberland--thin.jpg
St Anne's Reel. See Tapestry Explanation

Go to Old Chester

Go to Old Augusta

Go to Southwest Virginia Project


The following presents similarity analysis matricies for STR data from three FTDNA surname projects:Campbell, MacDonald, and Stuart. All data was taken from the public FTDNA surname project web site, or (in the case of the MacDonald Project, from the independent but publicly available web site maintained by the project managers. The analysis is restricted to kits with 111 markers, and was performed using two separate mutation rate weighting schemes. In the first scheme (Flat Rate) mutation rates were assumed to be invariant. In the second scheme (HI LO), mutation rates were based on the average results for individual markers, from four studies, excluding the highest and lowest rates for each marker. Data was provided courtesy of Tim Janzen. The four studies used were:

1. Chandler revised
2. Heinila, 2012
3. Ballantyne et al. 2010
4. Burgarella et al. 2011


The following similarity matrices were generated for kits with 111 markers in each of the three surname projects being explored. By focusing on 111 markers some of the issues related to the precision in STR YDNA comparisons are substantially reduced. The three projects were chosen because they are large enough to have numerous 111 marker kits. (each project has more than 100 such kits).

Mutation Rate Scheme Campbell MacDonald Stuart
Flat Rate Image:Similarity Diagram, Campbell, Flat Rate.jpg Image:Similarity Diagram, MacDonald, Flat Rate.jpg Image:Similarity Diagram, Stuart, Flat Rate.jpg
HI LO Rate Image:Similarity Diagram, Campbell, HILO Rate.jpg Image:Similarity Diagram, MacDonald, HILO Rate.jpg Image:Similarity Diagram, Stuart, Flat HILO.jpg


The above results show a consistent pattern. Roughly put, when HI-LO weighting is used the resulting display is substantially more coherent. The increase in coherence is visually dramatic, especially in the Stuart project, but clearly occurs in all three. Since it occurs in all three projects, the increase in coherence is not likely to be related to project specify differences, but rather is related to the difference in mutation rate weighting scheme employed, with HI-LO mutation rate weighting giving the most coherent results.