Jack wrote:Thanks David,
I can only presume the decision for that was to get the "best value for money".
(which can be counter-productive at times)
They don't do too badly !
The market is highly competitive, so that a contractor who doesn't perform knows very well that there won't be repeat business, and that QA procedures will soon reduce or eliminate their profit margin - not only that, but word quickly gets round GROS/GRO/GRONI/and many other similar institutions as well as commercial companies regarding which contractors can be relied on and which can't, - there are many, many other organisations subcontracting data input to such companies where the data sources are equally problematic.
I am assured that they are given comprehensive training by experienced Scots (fancy a fee weeks in Sri Lanka or China
), and have look up tables available for given names, surnames, and place names.
In addition, they work off the microfilms so don't have the advantage of "seeing past" the blue and red scores through the ages in the originals, but I'm told that they do brilliantly well !, certainly better than me
For a new contractor a particular area is first given to them and 100% checked by GROS. Thereafter, assuming phase one is satisfactory, only a certain proportion of the entries are checked, with a bias towards those aspects from phase one which came out worst.
I'm also assured that keying is genuinely independent inputting by two inputters, with any differences checked by a third, unlike some recent English databases where one inputter checks what was input by the first, which is not
true double-keying ......
I have been told that GROS aim to achieve an error rate of 4% or less, across all fields. In other words, assuming equal error rates for the four fields (which doesn't happen!, there always be some bias to one field or another), that's 1% each or less for given name, surname, age, and district.
I'd be completely discombobulated if the combined error rate of the Heads of Household, and the enumerators was anything approaching that.
I was recently given the details of a situation in the West of Scotland where part of a street was double-enumerated, - i.e. two different enumerators covered the same households, but have not yet had the time to analyze it. I'll be pleasantly surprised if the serious difference frequency is anything much less than found by Barney Tyrwhitt-Drake in a similar situation in London, - in the range 5 to 9% depending on how "serious" is defined, - a whole 15% (fifteen) different overall when trivial differences such as Elizabeth and Elisabeth, and other such "minor" spelling differences were taken into account.
This situation is obviously a compromise between cost and accuracy. In other words to achieve 2% could easily cost double the contract amount, and so on. To achieve less than that
, wha kens
Jack wrote:Just as well addresses didn't have to transcribed too...there are some odd named places in Scotland!
(that even folk in this country would have trouble with - i'm thinking me when i say this!)
Neatly put, in the sense that if we can't achieve 100% ourselves, how can we expect others to? I've been told that an advantage of using such sub-contactors is that they come to the situation with no pre-conceptions.
Regardless of how accurate any transcribers are, there will always be an irreducible minimum of entries where the entry is open to interpretation...... in which context 4% can't be bad
Nor is it possible to achieve total accuracy of what the Heads of Household themselves entered in the household returns, or told to the enumerator, due to the ear of the enumerator and his transcription from completed household returns .....
If this in any way sounds like an apologia for GROS it ain't!, - it's just a statement of the realities of the situation based on my experience and contacts.
Whatever might be our opinions about how much better the Scottish census indexes could be in terms of the information indexed, - the "thin" index now versus the "fatter" index later debate, - I still believe that we have to take a step back and congratulate GROS for their DIGROS initiative and the present situation regarding Scottish records and digitised images currently on line, and planned for the future, not only in terms of GROS sources, but also NAS sources, still to come ................