New GRO pilot birth and death indexes - discover lost children - mortality rates

[BumbleB]

OK - but I don't understand why you are involving FreeBMD at all, they ONLY have access to the printed GRO indices, which were produced at the time.

The attached is the only information that is available to FreeBMD and their transcribers.  GRO is the ONLY organisation which has access to the full information, so they are the only ones who can make the alteration.

[Andrew Tarr]

It looks like melba is correct and that I'm too optimistic then. I just find it hard to understand how the GRO could have allowed the project to inaccurately transcribe the death ages, which is what has happened if they haven't recorded the time unit. A bit of clarification from the GRO would be nice.

I don't suppose any project is 'allowed' to transcribe inaccurately, but due to human error it happens, right from 1837 when official Registration started.  I wonder how you imagine all those reams of transcription might be supervised (and who by) to ensure total correctness?

[cuffie81]

OK - but I don't understand why you are involving FreeBMD at all, they ONLY have access to the printed GRO indices, which were produced at the time.

The attached is the only information that is available to FreeBMD and their transcribers.  GRO is the ONLY organisation which has access to the full information, so they are the only ones who can make the alteration.

I'm not suggesting FreeBMD make any alterations. I'm just suggesting the FreeBMD data could be used in comparison with the GRO data, thereby identifying any potential records at issue in the GRO data.

Quite who could or would do the data comparison is another matter but it'd be a fairly trivial task for someone who had access to both datasets. So if FreeBMD were to offer their dataset to the GRO then the GRO could do it. Comparing the datasets doesn't require access to original indices or the GRO scans.

Edit:
The reason I suggested using the FreeBMD dataset was because it is widely regarded as being both comprehensive and accurate, and is probably the best dataset around for the death indexes. And FreeBMD *may* be open to the idea, whereas other (commercial) organisations may not (at at least not for free).

[cuffie81]

I don't suppose any project is 'allowed' to transcribe inaccurately, but due to human error it happens, right from 1837 when official Registration started.  I wonder how you imagine all those reams of transcription might be supervised (and who by) to ensure total correctness?

It's not that I believe the transcriptions are inaccurate (at least not the numbers). But rather that I find it hard to believe those in charge of the project would have allowed any transcriptions to be done without ensuring the transcribers had the means to record the time unit. If those in charge did allow this then they were at fault, as its resulted in inaccurate records (albeit the numbers are correct), and not fault of the transcribers.

It would be such an oversight in the project that I'm hopeful that the data does actually exist but simply hasn't (yet?) been exposed to us via the website.

Maybe if the GRO were more open re. the process and data we wouldn't be left guessing...

[BumbleB]

cuffie81 - FreeBMD have access to NO OTHER dataset information than that published by GRO.  Never have had, and never will have.  Please get your head round this.

[Sloe Gin]

I think cuffie is saying that the ages on the old GRO death indexes post-1866 were correct, insofar as infant deaths were only shown in years, ie 0 or 1.  And that these are the indexes that FreeBMD have used.

Whereas the errors have crept in on the recently issued revised GRO indexes.

This does look to be the case.  I've just looked at a DC for a child who died in 1916 aged four months.
FreeBMD says he's 0 years old.
New GRO index says he's 4.  ("age at death in years").

[cuffie81]

cuffie81 - FreeBMD have access to NO OTHER dataset information than that published by GRO.  Never have had, and never will have.  Please get your head round this.

I haven't said they do.

I was merely floating the idea that, in theory, it would be possible to identify inaccurate records in the GRO dataset by comparing it to an accurate dataset (eg FreeBMD). This would be possible, despite the fact that the GRO dataset death ages are in days, weeks, months or years and the other (accurate) dataset death ages are in years. This would only be possible for post-1865 records.

I seem to be winding people up here, which was never my intention, so I'll leave it at that.

[JohninSussex]

Just to add I found Cuffie's suggestion perfectly easy to understand and Bumble has maybe caused a bit of winding up.
 
Thinking about it, data matching between the original quarterly indices and the new ones might throw up corrections of other types, not just these age at death issues. 

This is from the GRO site ( https://www.gro.gov.uk/gro/content/certificates/indexes_data_quality_error_reporting.asp ):

How is the age at death for infant deaths (i.e. died within 12 months of birth) recorded in the indexes?
Generally, age at death for 'infant deaths' will be shown as 0, in line with the microfiche index. However we are aware of a number of records where the age at death is showing a different value e.g. 9. If this is the case, it is likely that the information held in the online index relates to the age in minutes, hours, days, weeks, or months (recorded this way on the original death registration) rather than in years. It is not possible to identify which records maybe affected but we will continue to correct any errors that we do become aware of, or those reported to us.

[cuffie81]

I've just realised that it may be my references to comparing datasets that is causing the confusion. By this I mean comparing the datasets programmatically and not by a person visually comparing one scan to another, thus only access to the data would be required and not any scans.