Predicted vs. Found Relations

[Peacock100]

Okay people, be nice here as the last thread was shut down:

Predicted: 2nd -4th
Paper trail: 4th cousin
Total cMs: 81.39
Largest block: 48.55 on chr 5

Predicted: 2nd - 4th
Paper trail: 3 x 5th C1R (3 cousinships at this level and all feeding in to same lineages and no closer relationships found)
Total cMs : 57.64 on 2 blocks; 23.03 on chr 15 and 15.81 on chr 7

Predicted: 3rd to 5th
Paper trail: 5th but with possible other relations interacting
Total cMs: 29.35
Largest block: 17.75 on chr 12

Predicted: 5th to remote
Paper trail: 3rd1R
Total Cms: 26.34
Largest block: 8.77 on chr 6

Predicted: 4th to remote
Paper trail: 7th cousin (this person also 7th couisn to above 3rd1R cousin)
Total cMs: 33.61
Largest block: 12.01 on chr 11


Predicted: 4th to remote
Paper trail: 4th1R
Total cMs: 37.76
Largest block: 11.04 on chr 2

Predicted: 5th to remote
Paper trail: 5th
Total cMs: 24.85
Largest block: 10.91 on chr 5


I think I have a few more I can add. Have to check.

Linda

[Peacock100]

Okay people, be nice here as the last thread was shut down:

Predicted: 2nd -4th
Paper trail: 4th cousin
Total cMs: 81.39
Largest block: 48.55 on chr 5

Predicted: 2nd - 4th
Paper trail: 3 x 5th C1R (3 cousinships at this level and all feeding in to same lineages and no closer relationships found)
Total cMs : 57.64 on 2 blocks; 23.03 on chr 15 and 15.81 on chr 7

Predicted: 3rd to 5th
Paper trail: 5th but with possible other relations interacting
Total cMs: 29.35
Largest block: 17.75 on chr 12

Predicted: 5th to remote
Paper trail: 3rd1R
Total Cms: 26.34
Largest block: 8.77 on chr 6

Predicted: 4th to remote
Paper trail: 7th cousin (this person also 7th couisn to above 3rd1R cousin)
Total cMs: 33.61
Largest block: 12.01 on chr 11


Predicted: 4th to remote
Paper trail: 4th1R
Total cMs: 37.76
Largest block: 11.04 on chr 2

Predicted: 5th to remote
Paper trail: 5th
Total cMs: 24.85
Largest block: 10.91 on chr 5


I think I have a few more I can add. Have to check.

Linda

Actually,

Looking at these results, posted this way, the FTDNA alogorithm seems to be pretty much on coarse as to my FF matches, at least those I have identified.

Linda

[thetick]

Let's hope everyone can stay on topic. Here is my re-post.

One confirmed with geneaology FF match:

Predicted 4th
Variance 3rd - 5th
Actual 5th cousin 1x removed
Total - 35.79 cM
Longest - 12.86 cM

[shandy4473]

2nd Cousin Matchings

Family Case 1 (Permission given by family members to post)
Family Member and Match: Lewis Ronald Lamar

Juliette Turner (Grandmother)
Shared cM: 345.91cM
Longest Segment: 104.40
Predicted: 2nd Cousin
Actual: 2nd Cousin

Stephen D. Handy Sr (Father)
Shared cM: 170.49cM
Longest Segment: 50.99
Predicted: 2nd Cousin
Actual: 2nd Cousin once removed

Stephen D. Handy Jr (Myself)
Shared cM 116.74cM
Longest Segment: 50.99
Predicted: 3rd Cousin
Actual: 2nd Cousin twice removed

Numbers weighed against concrete family knowledge and paper trail evidence.

[JSW]

DIY Relationship Prediction

Here is a chart you can use for Do It Yourself
prediction of a relationship using the sum of all segments.

[thetick]

Here is a chart you can use for Do It Yourself
prediction of a relationship using the sum of all segments.

Very cool. It took me a few minutes to understand the plots, so this might be a little too much for some newbies especially if they are scientifically challenged. Anyway I'm curious where you got the data for your plots? I just see adapted from ISOGG.

[JSW]

Data sources

Most of it is from test on my family but there are other data that
has been sent to me or posted in various places
I had about 60 data points to build the blue and green lines

[mkdexter]

It's a lot more complicated than that. I'd be careful on relying too heavily on a linear chart.

Look at two people who have three children, John, Jack and Jill.

John and Jack share 2950cM
John and Jill share 2850cM
Jack and Jill share 2900cM

Because each sibling has a different admixture of his or her past ancestry via recombination, their children, first cousins, when compared to each other will only share a portion of the amount the sibling parents share. In the case of John and Jack, 2950cM.

Those are moving reference points and they are relative.

A grandchild of John and a grandchild of Jack will be second cousins. They will have some DNA in common with each other as second cousins. If you could compare the second cousin test with the test of John and Jack you would find the second cousins only share a part of what John and Jack shared to begin with. A third person, a grandchild of Jill will share differing amounts with John's grandchild and Jack's grandchild. This is because the reference points change depending on which two siblings are in common to the two cousins.

The reason such a chart won't always work is because in the case I just outlined all descendants are sharing part maternal and part paternal alleles from the common ancestry and in each case different alleles are shared depending upon which two siblings are the origin of the reference pool. It's all relative as well. If another person is related by a different ancestor they may be phased when compared to the rest, it depends on where the common reference originales from.

My point being in the test environment, the reference point is relative to the ancestry in common. In some cases the reference is phased to one parent and in some cases it is not. In the case of people descending from two common ancestors it is actually the children of the ancestors that are the first reference point in the alleles being shared and as they have children then those children inherit a portion of alleles, caused the shared amount to be smaller each generation on each branch going down.

[Dumbldorf]

some fall out

Predicted: 2nd- 4th
Paper trail: 6th1R
Total cMs: 46.72
Largest block: 17.61

[Geneadict]

Here is a chart you can use for Do It Yourself
prediction of a relationship using the sum of all segments.

Would have better luck with segment count and sum of segments for segments 5 cM or higher, at least out to 2nd cousin. Beyond 2nd, recombination too random to predicted within 1 degree of relationship with high level of confidence. Beyond 2nd, best to focus on ranges and expand far end of ranges provided by FF.

[shandy4473]

Back on Track - 1st Half Cousin Relationships

Family Case 2 (Permission given by family members to post actual names)
Family Member and Match: Dr Martin Hardeman PHD

Juliette Turner (Grandmother)
Shared cM: 410.60cM
Longest Segment: 62.40
Predicted: 2nd Cousin
Actual Relationship: 1st Half Cousin

Stephen D. Handy Sr (Father)
Shared cM: 331.78cM
Longest Segment: 54.20
Predicted: 2nd Cousin
Actual Relationship: 1st Half Cousin Once Removed

Stephen D. Handy Jr (Myself)
Shared cM 203.68cM
Longest Segment: 50.91
Predicted: 2nd Cousin
Actual Relationship: 1st Half Cousin Twice Removed

Example of confirmed 1st cousins that are half relatives. Variance of FTDNA's matching algorithm to actual relationships is expected and acceptable given it's ignorance to the actual half relationships.

Numbers weighed against concrete family knowledge and paper trail evidence.

[Cats]

It's a lot more complicated than that. I'd be careful on relying too heavily on a linear chart.

Look at two people who have three children, John, Jack and Jill.

John and Jack share 2950cM
John and Jill share 2850cM
Jack and Jill share 2900cM

Because each sibling has a different admixture of his or her past ancestry via recombination, their children, first cousins, when compared to each other will only share a portion of the amount the sibling parents share. In the case of John and Jack, 2950cM.

Those are moving reference points and they are relative.

The reason such a chart won't always work is because in the case I just outlined all descendants are sharing part maternal and part paternal alleles from the common ancestry and in each case different alleles are shared depending upon which two siblings are the origin of the reference pool. It's all relative as well. If another person is related by a different ancestor they may be phased when compared to the rest, it depends on where the common reference originales from.
.

Great point. My brother and I certainly do not have each match listed the same way. Example, he has one match that FF says is a distant match and more than likely a 3rd to 5th cousin probable 4th and the same match for me is considered a remote 5th cousin.

[JSW]

DIY Chart

Matt is right when he wrote "The reason such a chart won't always work is because in the case I just outlined.........."

But that is not the only reason the chart won't always work
The processes we are random and thus there is always a "chance" that
the specific results we have are for out side of the normal range

The chart I posted should work "most of the time"
It also shows the wide range of possible relationships
when the result is less than 40 cM

Matt used an example of siblings in the 2900 cM range +/- a bit
But I have results that run from 2300 cM to 3000 cM for siblings
And as the generations go by these differences are multiplied

[JSW]

Siblings Matching others

RE "Great point. My brother and I certainly do not have each match listed the same way."

I would not expect siblings to match another person exactly - if at all.

Since each sibling got a different half of the DNA from the parent that
matched that third person. The results can be very different.

I match a 4th cousin and my brother has NO match at all with him
my match is 48 cM longest and 74 cM total.

As the relationship gets more distant and the number of segments drops
the odds that any segment makes it to the next generation is
always less than 50% - and if you are down to one segment
then it becomes all or nothing at that next generation.

[JSW]

DYI chart

RE "Would have better luck with segment count and sum of segments for segments 5 cM or higher, at least out to 2nd cousin. Beyond 2nd, recombination too random to predicted within 1 degree of relationship with high level of confidence. Beyond 2nd, best to focus on ranges and expand far end of ranges provided by FF. "

I did look at added complexity such as number of segments and length
of largest segment.
But none of those added enough discrimination to make it worth while
Others may be able to publish charts using more complicated data
and I welcome them.