Quality Score
Just wondering why person M56M-JCF has such a low quality score with 21 sources that all seem to match. How do I increase the quality score? Thanks
Best Answer
-
From previous discussions, in the PQS group, that residence after death is the deal-breaker. Presence of a residence fact after death lowers the score drastically.
3
Answers
-
@Larry R. White You'll want to post queries about the PQS in the PQS Group. https://community.familysearch.org/en/group/323-profile-quality-score-feedback
1 -
If you click on the orange box saying low, it opens a side banner that lists 11 issues it has found
1 -
Thanks. I already tried that but the deficiencies noted were really not accurate. All 21 sources are correctly attributed to my ancestor. 🤷♂️
0 -
A downside in this whole exercise does seem to be that it might encourage users to enter detail they do not have specific details for. For example, if you were to add an estimated date (beyond the year) and a city for the marriage it might help in raising that LOW rating. To some users, adding unverified detail might be an incentive (temptation!) for raising the Quality Score.
But, in using the example above, I notice the reason why your input here (year and state only) is relatively vague is that you have no source added as back-up - this also being a contributory factor in lowering the QS.
It could be argued that the criteria being used by the development team could be improved. For example, should the QS really be affected by the Residence data having no tagged sources? As suggested, the place for making points like these is really at the PQS group, rather than here, as a General Question.
There is one noticeable error in this case, however - you do have a residence place listed showing a date after the date of death (7 January 1954, when the death had taken place on the previous day).
I know that I, too, should be posting further comments at the PQS group, but I must add (since the topic has been raised here) that I am concerned about this whole concept of rating records - whether against individual profiles or (as recently found) among the collections found in the Genealogies section. Taking this case as an example, the sources attached do appear to relate to the individual concerned. However, there is no way of measuring the accuracy of source attachment when the items might appear to be an excellent match for the profile to whom they have been attached, but actually relate to a completely different person.
In the past I have found profiles with around 90 sources (ostensibly good matches) attached, but where a closer look reveals only a handful applied to that individual. As I have stated elsewhere, I'm afraid these ratings will lead many users to a false sense of security - in that they'll see a HIGH rating and think there is little investigation / work to be carried out on a profile that might be riddled with inaccurate data and attachments, which a computer program would be unable to spot.
1 -
Further to my previous comments, maybe I am misunderstanding the weighting being applied to different factors. I just looked at a profile (quite randomly) and found that, too, had 6 instances highlighted of "A residence has no tagged sources". (See KC36-NYS) Yet that profile has been assigned a HIGH rating.
Looking further at the comparison between this individual and M56M-JCF, I now see the LOW rating on the latter appears to be down to that one factor of a residence date falling after the date of death. "The Data is conflict free" section has "no stars", apparently merely on the basis of that. Yet maybe this input relates to the deceased being "resident" at the stated location immediately after his death!
I would be interested in the thoughts of users who have been working with this new feature to a far greater degree than me. Again, I realise I am raising the issue here, in line with where the topic has been raised, rather than, more suitably, within the PQS group, but I am a rather baffled as to how one of these profiles has been assigned a LOW rating and the other (also with some similar "flaws") in rated HIGH.
0 -
I note that it seems to be penalizing the score for a date of residence being one day after the date of death. You might try deleting that datum as it’s really not providing significant information anyway. I wonder if it got added as part of an obituary.
Edit: Sorry, it looks like someone beat me to this comment.1 -
This is a situation where the quality score is doing exactly what it is supposed to do and doing it quite well.
In this case, that residence event appears to have been moved over in the source linker and added to the record without actually evaluating exactly what it was or if it was correct. It came from an indexed obituary published Saturday January 9 which stated he died Thursday which would have been January 7. All other records say he died January 6.
So it is not a residence event. It is a death event. Also, the source is wrong. The way to improve the quality score in this case is to just delete that residence event. It would also be good to add a note to the source along the lines of "Although this obituary states he died the preceding Thursday without giving a date, according to all other sources he actually died Wednesday."
If you by chance can't view the image without going to a FamilySearch Center, it also states his funeral would be the following Monday. So if you otherwise trust the obituary, you could improve his burial date to be January 11.
If we take the term Profile Quality Score to mean the entire profile, not just the Details page but everything including the Sources tab, then it is giving us a chance to evaluate the accuracy of the sources, not just the person's entered details. Just because there is a data conflict, that does not mean the Details tab is wrong and a source is correct.
When we have a situation where there is a data conflict and the source is the problem, it should lead us to evaluate the source to first of all make sure it really is for the person it is attached to. If it is, then we have several things we can do:
- If there is an indexing error and we can edit the record, then fix the index (when that is finally working properly). If we can't edit the index, then we should still put a note on the soure as to what the indexing error is.
- If the indexing is accurate and the source is just wrong, then put an explanation in the source explaining why we know the source is wrong.
- Don't tag that particular source to its associated Details tab data if it is the tagging causing the notice.
- Don't put incorrect information from that source on the Details tab.
- Dismiss the notice in the Quality Checker.
4 -
Absolutely amazing comments! Thank you so much!
0 -
@Paul W said
" … 6 instances highlighted of "A residence has no tagged sources". (See KC36-NYS) Yet that profile has been assigned a HIGH rating. … "
(Yes, I know Quality Score concerns should be raised in that group but some background thoughts first may help…)
I would guess that there are 2 points to consider. Firstly a straightforward count of issues - some of these may not be the fault of the researcher, remember, they may be because the data simply doesn't exist.
Secondly, I would suggest that some assessment ought to be made of the count of issues against the amount of information on that profile. Four issues in four values is a bit different from four issues in twenty-four values, and ought to feed into the HML rating in my personal view. Whether a percentage consideration like this does exist, I don't know.
PS - yes, I have joined that group.
1 -
@Paul W Different issues are weighed differently, as you hypothesized. I'm honestly not sure if the "No tagged residence" decreases the score at all. Meanwhile, as Áine said already, having evidence of their life coming after their death will immediately drop the score to low, regardless of how accurate the profile is otherwise, because it indicates that something is definitely wrong.
(Usually, in my experience, it's just that something was mis-entered as a residence, as was the case here, but it could be used to catch far more egregious errors, so I don't mind that it weighs that issue so heavily.)
0 -
Thanks to everyone
0 -
As already noted by others, this discussion belongs in the Profile Quality Score Feedback Group. Please join the group and direct your questions and feedback there.
0
