.
I was not going to put these up for a while, but they really got me thinking, so thought I'd share them now. This is what happens if you apply the Heritage Action model to PAS data. I'm not going to put the numbers on, I want to check them, and that's several more hours work and they've taken up rather too much of Easter Saturday - so they are offered "as is" for the moment. The colours are a bit naff too.
This is a county-by-county breakdown of PAS-recording frequency among metal detectorists. What I did was take the data which is at the basis of the HA Erosion Counter model and apply them to the minimum number of detectorists in each county which I took from the PAS tables of detecting club membership rather than reporting detectorists.* These are minimum numbers, clubs sometimes limit membership (to ease pressure on 'club land') and not all detectorists belong to clubs (especially in the north of England). I then multiplied that by the average number of finds derived from the research behind the HA Artefact Erosion Counter model (see below). This gives for each county an idea of what number of finds are potentially being dug out of the archaeological record by that (minimum, remember) number of artefact hunters with metal detectors. Then I compared that with the real figures for reporting in the two periods noted in the posts earlier on this morning (here and the other one here). The first period studied was from 2003 when the Scheme went nationwide to the day before the download of a massive amount of extraneous data in March 2010. That gives a nice solid block of just over six years - arguably the heyday of the PAS - when reporting and recording were steadily progressing. One might expect therefore these figures to represent the great success of the Scheme in its outreach. Far from it, it reveals something totally different. Although the Scheme is the same right across the country, the application and response to it differ greatly. As I said, I'm not going to put numbers on this today. Suffice to say that dark red represents a level of recording which would perhaps satisfy the "better than nothing" brigade. These are regions which the figures we have suggest just a little more than half is being recorded, IF the actual number of active detectorists is not much higher than the figures used (and I think they are).** The light red areas have worryingly low recording. Anything below that (puce, dark green, dark brown) are deeply unsatisfactory - we are talking here about perhaps 70-80% of dugup finds NOT being recorded. The counties coloured black are - the figures suggest - being looted of artefacts on a tragically unmitigated scale. I'm talking here about it looking very much like 95% of dugup finds not getting reported or recorded. If that is true, it is nothing short of a scandal.
What is worrying is that every indication is that the real figures are worse in two regards, the first is that there are probably many more artefact hunters in each region than those who have joined clubs (and for the most part their finds are probably not getting to the FLOs) - secondly if you start dividing the number of finds reported for some regions (Norfolk for example) by the number of reporters, we find that the average number of objects ascribable to each is somewhat higher than the average taken as a basis for the HA counter algorithm. The latter is intended to be a conservative estimate, and it may well be that it is in fact too low.
In reply to those who suggest that the HA finds estimate is "wrong" (though on what basis?), I'd ask them to indicate how they wish to 'adjust' figures which are suggesting just 5% of dugup artefacts are getting reported in some regions. Postulating that each artefact hunter finds (for example) just three recordable items a year may well bring the results up to "50% reporting" in those regions, but then applying such a figure outside these regions would mean that artefact hunters in other regions are reporting many items that they are not finding! Obviously once again, the answer is that we need more reliable figures from the PAS in a form conducive to multivariate analysis. Until then, the HA figures are the best we have (though mainly because they are the only ones we have).
Here is the graphic representation of the situation these figures suggest in 2003-March 2010.
Unfortunately it gets worse. The period at the end of March 2010 (thus excluding the download of extraneous data which contaminates the picture) to the end of March 2013 is a short period and the numbers are correspondingly smaller - and therefore more sensitive to temporary glitches, like a FLO leaving the job and a period when there was nobody to fill the post (Essex for example). Nevertheless the picture seems clear. Here is the graphic representation, the same colour scheme as before.
The spread of the areas shown in black are noticeable. Almost everywhere, these figures suggest that there has actually been a decline in the percentage of dugup finds being recorded. In very few areas (Gloucestershire/Avon, Berkshire/ Oxfordshire, and the Isle of Wight) do the figures seem to remain the same. Why this is happening is unclear. It is also interesting to note a stronger differentiation emerging between the southeast/Home Counties and the rest of England. Nigel Swift and I have noted a similar divide in the character of metal detecting clubs between these two area (incidentally if you look at the map in the sidebar, the distribution of readers of this blog tends to divide into two clusters along a similar line - which actually, though I think coincidentally, reflects the underlying geology).
The extreme variation in proportions is difficult to explain, they go down to just one or two percent of all dugup finds of the HA model in some counties to 65 and 75% for two counties before March 2010. In fact, it is only certain counties (certain FLOs!) which are keeping the numbers up to their apparent high levels.
Also rather unclear is the interpretation of the anomalously high values of recorded finds from two northern counties, North Yorkshire and East Yorkshire (while South and West have an abysmal record). I am sure the FLOs there are very persuasive and approachable, but is that the only factor?
I stress that these are work-in-progress, but even so reveal a couple of things that really need to be looked at in more detail. Needless to say, any light the PAS itself can shed on the factors involved and the interpretation of these figures would be very helpful.
*since the latter tend to be more mobile than those regularly attending a club - I am of course aware that not all members of a club which is based in one county actually live and do all of their detecting in that county - but these are currently the best figures in the public domain which we have from the PAS.
** It should also be stressed that the PAS statistics being used lump metal detected finds together with accidental finds by other members of the public, so these values are in fact much higher than the actual ones.
I was not going to put these up for a while, but they really got me thinking, so thought I'd share them now. This is what happens if you apply the Heritage Action model to PAS data. I'm not going to put the numbers on, I want to check them, and that's several more hours work and they've taken up rather too much of Easter Saturday - so they are offered "as is" for the moment. The colours are a bit naff too.
This is a county-by-county breakdown of PAS-recording frequency among metal detectorists. What I did was take the data which is at the basis of the HA Erosion Counter model and apply them to the minimum number of detectorists in each county which I took from the PAS tables of detecting club membership rather than reporting detectorists.* These are minimum numbers, clubs sometimes limit membership (to ease pressure on 'club land') and not all detectorists belong to clubs (especially in the north of England). I then multiplied that by the average number of finds derived from the research behind the HA Artefact Erosion Counter model (see below). This gives for each county an idea of what number of finds are potentially being dug out of the archaeological record by that (minimum, remember) number of artefact hunters with metal detectors. Then I compared that with the real figures for reporting in the two periods noted in the posts earlier on this morning (here and the other one here). The first period studied was from 2003 when the Scheme went nationwide to the day before the download of a massive amount of extraneous data in March 2010. That gives a nice solid block of just over six years - arguably the heyday of the PAS - when reporting and recording were steadily progressing. One might expect therefore these figures to represent the great success of the Scheme in its outreach. Far from it, it reveals something totally different. Although the Scheme is the same right across the country, the application and response to it differ greatly. As I said, I'm not going to put numbers on this today. Suffice to say that dark red represents a level of recording which would perhaps satisfy the "better than nothing" brigade. These are regions which the figures we have suggest just a little more than half is being recorded, IF the actual number of active detectorists is not much higher than the figures used (and I think they are).** The light red areas have worryingly low recording. Anything below that (puce, dark green, dark brown) are deeply unsatisfactory - we are talking here about perhaps 70-80% of dugup finds NOT being recorded. The counties coloured black are - the figures suggest - being looted of artefacts on a tragically unmitigated scale. I'm talking here about it looking very much like 95% of dugup finds not getting reported or recorded. If that is true, it is nothing short of a scandal.
What is worrying is that every indication is that the real figures are worse in two regards, the first is that there are probably many more artefact hunters in each region than those who have joined clubs (and for the most part their finds are probably not getting to the FLOs) - secondly if you start dividing the number of finds reported for some regions (Norfolk for example) by the number of reporters, we find that the average number of objects ascribable to each is somewhat higher than the average taken as a basis for the HA counter algorithm. The latter is intended to be a conservative estimate, and it may well be that it is in fact too low.
In reply to those who suggest that the HA finds estimate is "wrong" (though on what basis?), I'd ask them to indicate how they wish to 'adjust' figures which are suggesting just 5% of dugup artefacts are getting reported in some regions. Postulating that each artefact hunter finds (for example) just three recordable items a year may well bring the results up to "50% reporting" in those regions, but then applying such a figure outside these regions would mean that artefact hunters in other regions are reporting many items that they are not finding! Obviously once again, the answer is that we need more reliable figures from the PAS in a form conducive to multivariate analysis. Until then, the HA figures are the best we have (though mainly because they are the only ones we have).
Here is the graphic representation of the situation these figures suggest in 2003-March 2010.
Unfortunately it gets worse. The period at the end of March 2010 (thus excluding the download of extraneous data which contaminates the picture) to the end of March 2013 is a short period and the numbers are correspondingly smaller - and therefore more sensitive to temporary glitches, like a FLO leaving the job and a period when there was nobody to fill the post (Essex for example). Nevertheless the picture seems clear. Here is the graphic representation, the same colour scheme as before.
The spread of the areas shown in black are noticeable. Almost everywhere, these figures suggest that there has actually been a decline in the percentage of dugup finds being recorded. In very few areas (Gloucestershire/Avon, Berkshire/ Oxfordshire, and the Isle of Wight) do the figures seem to remain the same. Why this is happening is unclear. It is also interesting to note a stronger differentiation emerging between the southeast/Home Counties and the rest of England. Nigel Swift and I have noted a similar divide in the character of metal detecting clubs between these two area (incidentally if you look at the map in the sidebar, the distribution of readers of this blog tends to divide into two clusters along a similar line - which actually, though I think coincidentally, reflects the underlying geology).
The extreme variation in proportions is difficult to explain, they go down to just one or two percent of all dugup finds of the HA model in some counties to 65 and 75% for two counties before March 2010. In fact, it is only certain counties (certain FLOs!) which are keeping the numbers up to their apparent high levels.
Also rather unclear is the interpretation of the anomalously high values of recorded finds from two northern counties, North Yorkshire and East Yorkshire (while South and West have an abysmal record). I am sure the FLOs there are very persuasive and approachable, but is that the only factor?
I stress that these are work-in-progress, but even so reveal a couple of things that really need to be looked at in more detail. Needless to say, any light the PAS itself can shed on the factors involved and the interpretation of these figures would be very helpful.
*since the latter tend to be more mobile than those regularly attending a club - I am of course aware that not all members of a club which is based in one county actually live and do all of their detecting in that county - but these are currently the best figures in the public domain which we have from the PAS.
** It should also be stressed that the PAS statistics being used lump metal detected finds together with accidental finds by other members of the public, so these values are in fact much higher than the actual ones.
No comments:
Post a Comment