Question Regarding Timespans
One of the data flags in the Data Quality Score is:
The date of … is in conflict with the standardized place of … which has only been called that since ….
In the Places database, I have seen that transitions between time periods have been handled in two different ways if the change was not made at midnight December 31st.
- The ending year for the earlier time period and the starting year for the later time period are the same.
- The ending year for the earlier time period has been rounded down or the starting year for the later time period has been rounded up.
Is there a provision in the quality checker to allow the year on the profile to be ±1 of the starting or ending year of the time period found in the Places database so that the flag is not triggered if, for example, a birth is entered with the correct place name as it was at the time but the Places database has a rounded up starting year? (Say a county was split 1 Nov 1890, a person was born in December and has the new county name in the place name, but the Places database has 1891 as the start for the new county.)
Or does an incorrect starting or ending date need to be reported to the Places team if this flag is triggered by a rounded year so they can correct the standard?
Kommentare
-
The thing is, the Places team cannot correct the endpoints for changes that didn't occur at New Years. Their database is only accurate down to the year. If a change went into effect on 1 July 1850, then whether the database has 1850 or 1851, the data rating will be wrong for half of the events that took place in 1850 — unless that particular message is set to accept either value for events at or near the endpoints.
0 -
Made up example to clarify my question:
My City was moved from County 1 to County 2 on 1 July 1890.
If the timespans are set up as:
- My City, County 1 - unknown to 1889
- My City, County 2 - 1890 to today
Then if someone puts an event date of 25 March 1890 in My City, County 1, it will be flagged even though it is correct.
If the timespans are set up as:
- My City, County 1 - unknown to 1890
- My City, County 2 - 1891 to today
Then if someone puts an event date of 25 Sep 1890 in My City, County 2, it will be flagged even though it is correct.
If the timespans are set up as:
- My City, County 1 - unknown to 1890
- My City, County 2 - 1890 to today
Then if someone puts an event date of 25 March 1890 in My City, County 1 or an event 25 Sep 1890 in My City, County 2 both will be just fine because the My City was called both of those in 1890.
As I've been jumping around looking in US counties and checked places within those counties that have changed jurisdications, all the ones I've checked are in the style of my third example. I didn't think that was the case. Either I've been confused about this for years or there has been some bulk editing of the Places database to have that overlap of years so that the quality checker doesn't cause problems or I've just looked in the wrong places for examples.
0 -
@Gordon Collett we pushed out an algorithm change today to give 1 year leeway on these kind of issues. Thanks for the feedback!
2 -
Wonderful to hear. I was getting a bit nervous.
1
