What do you think about adding certainty status for relationships?

Question

Hi WikiTreers,

 

There's an idea that's been floating around for some time: certainty status for parent-child relationships.

 

Just like you can mark a date as certain, uncertain, etc., you'd be able to mark a person's mother and a person's father as certain or uncertain.

 

There might be three certainty options:

1. Uncertain
2. Genealogically certain
3. DNA confirmed

What do people think of this idea? I don't think it's really been discussed.

 

I'm sure it would cause conflicts. People will certainly disagree about what "genealogically certain" means. We can try to develop a standard, but it will be hard. It will be a sensitive subject.

 

Genetic genealogists will also argue about what "DNA confirmed" means. Again, we can try to develop a standard, but it will be a new area for mistakes and misunderstandings and hurt feelings.

 

Even after the community develops standards for their meaning, the utility will always be limited by inconsistent application. People could never expect that all the relationships marked as certain really are certain. We would have to be careful in how we present "certain" relationships, especially to non-genealogists who wouldn't understand that all of WikiTree is always a work-in-progress.

 

That said, having this could be very useful. For a long time WikiTreers have wanted a way to make tentative parent-child relationships but show they are uncertain. Genetic genealogists want a way to see which lines on a tree are confirmed through DNA and which still need to be.

 

I'm sure many other possibilities would open up. Today someone asked about how to easily see which lines in their family tree need more research. This would help enable tools for that sort of thing.

 

Thoughts?

 

Chris

Comments

I could just reach over these inter-webbed-thingies and hug you!!

Will pull some specifics on this an post a bit more.

---

Do others see a need to have two levels of uncertain (e.g. something like speculitve and unsure)?

I know of several instances where Y-DNA has confirmed two males share the same direct paternal line ancestor with their surname, but traditional genealogical evidence does not exist to connect each of their direct paternal lines together "on paper".

---

I love the "uncertain," and I love the "DNA confirmed."  Not sure about "genealogically certain" -- people have been know to have incorrect info on birth & death certs (and we don't even want to talk about census accuracy!).

Question -- would we be required to use these, or would this be something we could leave blank?

---

Thanks, GeneJ and Peter.

Nan, I don't imagine you'd need to select a status when adding a parent.  It's always possible someone else might add it later.

---

Perhaps for a separate discussion, but I'm not sure why we'd describe this as a way to introduce speculative relationships to the tree. Speculation is something that directs your research, as opposed to a conclusion in and of itself. So many families (like my own) were gobbled up by speculators working on other familes--enough to all but drive the ancestral line to extinction. 

Okay, back to what I hope is helpful information. For your viewing pleasure ...

In the latest New England .... Register, I find child list entries are prepended with "(possible)" and "(probable)." I also find names pre-pended with "?."

In Evidence Explained (2007), Mills writes about "Levels of Confidence." She says, "Within sound historical studies, statements about dates, events, identities, places, relationship, and similar matters are frequently prefaced by qualifiers ..." She then recognizes four as "apparently, likely, possibly, or probably." Saying there is no universal scheme, she then sets out to devise one, offering up definions for six. These six are certainly, probably, possibly, likely, apparently and perhaps.

For the what it's worth gallery, based on my own recollection, the terms "probably" and "possibly" are the qualifiers I have more often observed. 

Isolating Mills definition for these two terms, below

Probably: The author feels the assertion is more likely than not, based upon sound research and good evidence. 

Possibly: The author feels some evidence supports the assertion, but the assertion is far from proved.

Hope this helps. 

---

Chris,

What's the problem you're trying to solve?

It's difficult for me to judge a proposed solution when it's not clear to me what it's a solution TO. Thanks.

---

Jillaine, the DNA feature is what's most interesting to me right now, i.e. showing DNA confirmed lines on a tree.

The DNA Ancestor Confirmation Aid is designed to help you confirm relationships. But right now the only way we can show a relationship is DNA-confirmed is inside the profile using a template. That's difficult for users and its of limited value. If we did it the way I'm talking about it would enable visual trees. It would be easier to use and much more powerful.

A long time ago the certain/uncertain issue came up in relation to the question of what level of certainty you should have before adding a parent on WikiTree. Right now it's a black and white choice. You have to enter a parent (or connect the relationship) or leave them out. There is nothing in between. An "uncertain" status would enable that middle ground. Lianne especially has been excited about the possibiity of showing uncertain relationships on trees.

Again, like with DNA confirmations, right now we can only explain that a relationship is questionable in the profile. This wouldn't be a substitute for that. We still want all the details -- all the research -- in the text of the profile but having it in a database field enables a lot more functionality.

---

GeneJ, I really appreciate your adding the above. I'm surprised all this has never come up in regards to the certainty status indicators we already have. In retrospect, I'm surprised there hasn't been more controversy about them. Or maybe I've missed it.

---

This would be a multi-select option right?

As I mentioned in a recent discussion, if identical twin brothers married identical twin sisters the children of the two couples would (within most currently available tests) all likely share the same yDNA (if male), the same mtDNA, and a percentage of auDNA expected of siblings. The only thing that would identify who each childs parents were would be the documentation of the births.

While this would be an exceedingly rare circumstance I bring it up to illustrate that even if every individual involved is DNA tested, "DNA confirmed" and "Genealogically Certain" (i.e. certain by paper-trail) are complementary, not either/or.

---

Adding to Rob Ton's thoughts on say YDNA

Kinship is a specific finding--this child, this parent; this child, this sibling, etc., whereas I think of say YDNA alone as a finding of ancestral lineage--being a match between a descendant of someone and another descendant of someone. (P.S. What Cynthia Billups just said, and said much better.)

---

Wow!  Some great ideas.  Really three distinct questions, as I see it, (and my old eyes are not so good nowadays).

GeneJ, I love your post on the levels of certainty for relationships per Mills.  We could do it like the dates, (which have 5 choices, and can choose any or none of them).  Perfect!  Let's do it!

DNA tracking, Good idea, Chris, but I see it as a separate issue from certainty levels, with different parameters.  Not knowing much about it, doesn't DNA give more of a family line result than an exact parent?  Does it actually distinguish between say, brothers 200 yrs ago?  Very different than paper trails, no?

Roland and Robin, quality score for profiles, needs discussion.  Don't we already have this in maintenance categories?  I am probably misunderstanding the idea.  Maybe it would cover a different area?

---

Rob's point about it being "multi-valued" is an important one to consider up front.

In my original presentation (certain, genealogically certain, or DNA confirmed), you'd be choosing one status. It would be like a scale or range of options on one spectrum. You couldn't choose both "genealogically certain" and "DNA confirmed."

We could do this in a multi-valued way. In addition to "genealogically certain" and "DNA confirmed" both being possible, we might consider "YDNA confirmed", "MtDNA confirmed, "Autosomal DNA confirmed", etc.  That gets to Cynthia and GeneJ's last comments.

But ... multi-valued would add a lot of complication. Not just the technical back-end. It would complicate the user-interface for adding and editing, and for presenting what it means.

---

If a male is a Y-DNA (Y-STR) match with his "3rd" cousin with the same surname then this confirms their direct paternal line back to their great-great grandfather with that surname.  It confirms (but does not prove) that each father along those lines is the father.  The real father could be a brother, uncle, etc. with that surname.  These exceptions happen but are relatively rare.

The DNA can confirm or disprove other genealogical evidence.

Roughly 2% of all births are non-paternity events.  http://en.wikipedia.org/wiki/Non-paternity_event

---

To build on what Peter wrote and connect it back to the question raised by Rob: There is a sense in which "DNA confirmed "is a level above "genealogically certain" on the same scale. The relationships are discovered through genealogy. The genealogy is confirmed through DNA.

---

I like the idea. We should tie 'certainty' back to the quality of the source information supporting that fact, i.e. if supported by primary/secondary sources then it should be marked as certain. If estimated, or supported by low quality derivative sources, or not at all, then marked 'uncertain'. Personally, if I find an exact unsupported date/location which does not conflict with other sources, I use the exact date/location and mark as uncertain. Same thing with name variations. All supported by notes in the biography, of course.

One additional use case we see a lot in PGM and EuroAristo: LNAB (last name at birth) uncertain. We see may cases with circumstantial evidence pointing to a (usually female's) parents, when only her married name is certain. This usually comes up when merging a LNAB Unknown into a LNAB known profile.

I can definitely see the usefulness of tying the relationship certainty back to DNA evidence among multiple distant descendants.

We should also resolve the question (current state) of 'how certain/uncertain must a relationship be before it should be added/removed'.

---

OK, I am starting to understand the DNA certainty, and I think that someday it will be integral to the paper evidence, but not so much until at least one male from each "line" is tested.  Am I understanding this?

Right now, we have more paper than DNA (I think, maybe I'm wrong). If we were to set up two different systems, with the idea that they could be combined somehow, would that work?  Say we had the Mills levels for each relationship and the DNA evidence for the parentage separately ... with the goal in mind to possibly add the "genealogically certain" designation later.  Wouldn't it accomplish the same thing?

For instance, if a profile were designated Mills "certainly" and it also was a DNA match, then this would award it the "genealogically certain" designation.

This way we could use one without the other.  I just think it's too early for this to be one standard, and we can accomplish your goal, Chris.  Am I seeing this correctly?

---

Cynthia, I don't think it's too early for almost any genealogist to be confirming some of their family relationships through DNA.

The autosomal DNA tests, the ones that cost about $99 and look at your entire chromosome, are changing everything.

I don't want to open this can of worms here (with Peter, especially) but I think "confirmed with DNA" would appropriate simply based on matches from autosomal DNA tests with distant cousins you would expect to match with based on paper genealogy.

---

auDNA is more complicated than Y-DNA and mtDNA partly because with much of the current auDNA reporting (AncestryDNA especially) it is too easy to mistakenly attribute a match to a known shared ancestor (or ancestral couple) when actually that shared segment could be from an unknown shared ancestor.  Phasing or trangulated groups and chromosome mapping are what are needed.

---

Speculative

Documented

DNA-confirmed

---

1. I really like Jillaine's suggested wording of "Documented" rather than genealogically certain.

2. It seems we are mostly in agreement that DNA can confirm the documented genealogy or disprove a relationship (documented or speculative). I think this is what is intended by saying DNA Confirmed or Confirmed with DNA. The problem I still envision with having that as the only alternative to "documented" is that some users will use a DNA test as a substitute for documentation rather than as corroboration.

Since a multi-valued selection is technically problematic, how about simply giving 4 options: A) Speculative, B) Documented, C) "Genetically Linked", D) Confirmed.

I'm sure someone can come up with something better/clearer than "genetically linked", but hopefully it is clear enough for this discussion. "Genetically Linked" would be when a speculated relationship is supported by DNA test results. "Confirmed" would be reserved for when a documented genealogy is supported by DNA testing or, if some 'higher threshold' of analyzing DNA results (one of the ones Peter has mentioned?) can distinguish a parent-child relationship from all other types of relationships without documentation.

3. I'm not clear on the value of, in addition to the DNA template, specifying what type of DNA test confirms parent and child are related; If a parent-child relationship meets the threshold for 'confirmed' using one type of test, any other type of DNA test logically must also support that finding. I suggest that the addition of the 4th option above establishes reasonable confidence levels for both genetic and non-genetic genealogists looking at trees.

---

Please please please consider the impact of any choice on the end user and on the community overall.

You know I'm a fan (most of the time) for us using the work of Elizabeth Shown Mills.  But five levels of certainty? The amount of disagreement and confusion about what each of those mean is so likely to result that I'm just not sure it's worth it.  

I'm also concerned about the lack of multi-choice and people coming to believe that if they've done DNA testing they don't have to do any documentation research. That could lead to purely GENEalogy and discount/leave behind family history. That could lead to long and very boring lines of begats. Blech.

---

Overall I am in support of the idea.  I agree with Ron regarding the need to have a multi select option.  Ideally the long term goal should be to have (1) documented and (2) DNA confirmed.

---

I like the focus on documentation or evidence.  Something like: Undocumented, Documented (or Some evidence), Well-documented (or Primary documentation), DNA evidence.  Words to that effect.

This may be a can o' worms, but is it true that DNA (depending on the test) usually gives supporting evidence, not proof?  If that's true, I think I like DNA supported or DNA evidence.

As long as it doesn't have to be either/or, paper/DNA. (multi-select), yes!

---

Maybe I'm being dense, but I'm confused about what undocumented/speculative would mean.

When is there no documentation, but a reason to create a WikiTree profile? Is it for the newbies who are entering an unsourced tree that was given to them or that they found elsewhere? If so, I'm not sure "Documented" would mean anything to them. They wouldn't know how to use it.

---

I guess you're right about that, Chris.  We don't really want any undocumented entries (brain glitch).  The real point was to focus on documentation, maybe "Not well documented."  This is when we find a list of children's names in a book, secondary or tertiary.  (Yeah, that's what I meant to say)

OK, here's an example.  Re. Abner Mead - Doherty in Beekman said "He is supposed to have been born 1747, son of Amos Mead."  No documentation or Not well documented.  I had him attached to Amos on that basis, until someone found a conflicting father (now he's unattached). BTW, both profiles already existed.

---

1. Speculative - based on shaky evidence (someone else's family tree, family legend without any documentation to back it up)

2. Documented - existence of documents that support the relationship -- a marriage record in a church register, a marriage or birth certificate

3. DNA-supported - DNA tests demonstrate a genetic relationship between these individuals.

The problem is, however, that #2 and #3 are not mutually exclusive. If we're stuck with choose only one, we have another problem.

---

I don't understand why #2 and #3 aren't mutually exclusive.

Number 3 is for relationships that were established through genealogy documentation and then confirmed through DNA testing.

As I understand things, we would not be entering new relationships that were only discovered through DNA testing, with no other sources. I'm not even sure how that would work.

As for the terms we use, "confirmed" is one that's already used in the genetic genealogy community. http://www.wikitree.com/wiki/DNA_Confirmations Therefore, I think it would be the most intuitive for them.

My thinking on using the terms "uncertain" and "certain" for the other classes is that it's what we use everywhere else. If we switch around the terminology for this certainty status, should we also change all the other certainty status indicators?

---

Sorry I did not understand before that option #3 required both documentation AND DNA.

(I could certainly envision a scenario in which documentation indicated that John Smith was my brother's father but where DNA might prove otherwise.  Well, not in MY family... :-b

If you treated DNA exclusive of documentation you'd exclude a huge chunk of family history. )

---

We want to move forward with this.

The obstacle is that we need simple instructions for "citing your source" when you say a relationship is confirmed with DNA. We basically have this for yDNA and mtDNA but not for the far more popular autosomal tests.

See http://www.wikitree.com/g2g/130600/draft-instructions-confirming-relationships-with-autosomal

I know there will be a lot of controversy and different opinions about the details. We could debate them endlessly. I think we will be debating them for years to come. I just want to keep this moving, one step at a time.

Answer 1

I think it would be a big help & reduce problems when using the relationship finder.

Answer 2

I like the idea. The first thing that pops into my mind is using that certainty info while displaying trees or relationship calculator by coloring the links based on certainty level. For example, uncertain relationships may show up in red. That could help focus where extra work is needed.

This idea is a bit similar to an idea that's been floating around in my head, where profiles could get some sort of quality score. The goal would be again to help us find the profiles most in need of help.

Comments

I really like the idea of going the step further and giving profiles a "certainty, or validity score", I'm not sure quality is the right term...but, yes, adding a indicator for the relationships, the same way we do for the dates and places would help a lot with understanding which profiles need further investigation.

Answer 3

I like the idea of certin/uncertain or no check mark. Let's staart with those. Parentage would be certain if documented in a source, uncertain if doubt is expressed in a source, and unmarked if unknown or unsourced. Wait to set higher standards for certain until more DNA results are available.

Answer 4

While the overall wording and content will no doubt need to be tweaked, particularly the very last paragraph where I am talking outside my knowledge, here is a suggestion that people can vote for or against and can make specific suggestions about:

 

Suggested guideline on Confidence in parent-child relationships:

Speculative

Speculative is the lowest and default confidence level and should be used when the parent-child relationship does not meet the guidelines for a higher confidence level. In general this is the appropriate confidence level whenever there is no known primary source origin for the relationship. For example, a family tree posted on the internet without sources should, in most cases, be treated as speculative.

Documented

Documented, as a level of confidence, means simply that the information originates in a primary source. This is not an evaluation of the quality or acuracy of the specific primary source(s). A parent-child relationship should be set as documented when:

1. The information comes from a relative who is a 1st or 2nd degree relative to either the asserted parent or child.

2. The information comes from a person who personally met and knows the individuals as parent and child. (for example, a god-parent, teacher, or the next door neighbour etc.);

3. The information comes from a primary source document. Examples: birth or baptismal certificate, records in a family bible, wedding license, death registration, will, obituary, personal correspondence/memoires of the individuals listed in 1 and 2 above, a passport application, a census record that specifically gives relationships of the inhabitants of a household, etc.;

4. A source citing/derived from one of the above, such as a published genealogy that gives sources (i.e. the fact does or is claimed to originate in a primary source). Ideally the cited sources should also be independently examined.

Genetically Linked

Genetically linked should be used when DNA evidence supports a speculated parent-child relationship, but the relationship does not satisfy the requirements of being documented as given above. While there are cicumstances where DNA tests can produce false positives, or support more than one relationship, this is still considered a higher confidence than being documented only..

Confirmed

Confirmed should only be used when:

1. DNA evidence corroborates documentary evidence of a parent-child relationship. While this could still theoretically contain errors, this represents the highest possible confidence; or,

2. Where advanced DNA tests or analysis methods, such as Whole Gene Sequencing (WGS) or (triangulation?) have 'proven' a speculated parent-child relationship.

Comments

Hi Rob,

 

To play Devil's Advocate, quoting Jillaine: "What's the problem you're trying to solve? It's difficult for me to judge a proposed solution when it's not clear to me what it's a solution TO."

 

:-)

 

In proposing three levels, I had two problems in mind:

 

1.) We would like a way to indicate that a father or mother is questionable, without removing (or not adding) the relationship. Right now it's a black and white choice. The only way to add a tentative link is in the profile. It would be nice if we could have a way to highlight that a relationship is uncertain without losing the ability to show the relationship altogether.

 

2.) Genetic genealogists would like a way to indicate that a father or mother has been confirmed through DNA testing. Right now it can only be done in the text of profiles with a complex template. It would be great to see it on a tree.

 

We could solve these two problems with one certainty status indicator that has two or three levels. Strictly speaking it would only need two levels, #1 (uncertain) and #2 (confirmed with DNA). But practically speaking, having a middle level makes it more usable. It would also be more usable if it parallels the uncertain/certain indicators on other fields.

 

Chris

---

1) In your original post you stated "Genetic genealogists want a way to see which lines on a tree are confirmed through DNA and which still need to be." The reverse is also true: non-genetic genealogists want a way to see which DNA confirmed lines on a tree are 'documented' and which still need to be.

2) In a response above you identified that "DNA-Confirmed" is "for relationships that were established through genealogy documentation and then confirmed through DNA testing."... Speculative/uncertain relationships are(usually) established through genealogy documentation, it just uses a 'lower standard' of sources - secondary, indirect, and circumstantial evidence. These uncertain relationships can be supported by DNA evidence, but if the test is less accurate (for example uses less markers) we may not be 'certain' a specific relationship is correct where that test would be perfectly adequete to be 'certain' of a better documented relationship.

If I have no primary sources for a relationship AND the DNA test cannot conclusively say two individuals are parent-child (rather than, say siblings, or grandparent-grandchild), only that they are related, then that is obviously not "Certain". However the genetic genealogists would still want to see that a DNA test had already been done.

---

---

Food for thought ... (anyone like pot roast?)

What is certainty, unless defined as standards with a cut point? The words, "certain" and "uncertain" are subjective, denoting personal psychological belief.  Some folks feel certain that they are descended from royalty, but ... can we all agree, if there's no documentation?

Choices here, keep the current words and define them on a separate page (hope everyone reads it).  Or use the defining words right there on the profile, (much more foolproof, and much clearer) a reason to prefer "documented" over "certain."

You've heard the one about ... how to cook a pot roast?  Always cut the ends off first.  Why?  Because that's how mother always did it.  (as it turns out, her pan was too small)

---

Chris I like this, especially in those lines that are being added to on a regular basis. And I think it clarifies alot of information in a simple way and makes collaboration easier. Instead of trying to track what the profile creator was thinking, having to discuss questionable info with multiple managers. I think it would also help understand why changes are taking place in a profile. As its being discussed now it's simple enough that even I could understand and implement it.  I think this would be helpful for newbies because when I started I assumed that the profiles of my relatives that were already in existence were correct and my info was wrong. In some cases the dates were way off and no sources, and it was hard to work out. This system would be a good way to guide newbies.

Answer 5

I have a fundamental problem with the ranking of DNA results as superior to "mere" paper documentation that seems to permeate this discussion. While this may be true when used for exclusion, it is far from certain when using it in an inclusionary way. Has anybody ever wondered why in forensic genetics nobody is ever 100% identified? And that applies to paternity testing when samples from both immediately involved persons are compared.

In my opinion genetic testing is a supportive method, primarily to disprove family relations, and it may be used to strengthen associations established by some documentary evidence. To look to it as the crowning achievement of genealogy is a little premature.

Comments

I completely agree with Helmut on this.  I am of the opinion that DNA testing is receiving far too much hype - which seems to lead many people to ignore its flaws.

One of my pet peeves is the throwaway conclusion that people make suggesting that if the results of two test-takers from the same family don't match, this is evidence of a "non-paternal event".  Meanwhile, if you read about the science behind the DNA tests, they document that there is a high likelihood of mutations along various markers that are used in the tests.

I need to see more on how these tests account for the possibility of mutations (especially when you start going back several generations) before I start accepting DNA tests as proof of a non-paternal event.

LOL - you should see the vast range of surnames among my matches in the 12 marker and 25 marker Y-DNA test results (I just took a quick look - there are 23 different surnames on my first 3 pages of results at 12 markers).  If we accept the test results as the highest possible standard of evidence, then I guess my paternal ancestors must have been part of some free love cult.

Either that - or the test results just aren't very reliable going back in time due to the common frequency of mutations between generations (which is an unpopular view when you are talking to a fan of DNA in genealogy).

And - I won't even get started on the ethnicity percentages that some people fixate on - i.e.,  "I never knew that my family was 3% Viking..." or the astounding "My test results show that I have 1% Neanderthal DNA."

---

Hello Ray and Helmut, 

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

 

---

As my original comment stated my problem is with ranking genetic testing as somewhat superior to paper documentation. I have no problem with it as long as it is used to corroborate or exclude relationships otherwise established or as pointers in the direction one needs to research.

---

Helmut, I agree with you. Using DNA for exclusion is good. Using DNA to help find connections is also good, but it doesn't prove or confirm a specific connection. Maybe there are cases where DNA is stronger evidence, but in general, it should be treated as another "source" of information. I am not aware of any source that is considered infallible.