Postcode areas will not necessarily have agricultural census
data for every year (particularly urban areas).
The data is redistributed from the ward-level dataset. Because
the Isle of Wight and the
Isles of Scilly are not included
in the ward98 geographical dataset that is currently used as standard,
there are no results for 2002.
Note: that because of the nature of repopulation (see below)
there may be some postcode areas where the difference between
the actual data and the redistributed data is larger than expected.
'Double Counting'
For subtotals (e.g. total cattle), the values are given (i.e.
area of land, number of livestock) but not the number of holdings,
this is to avoid 'double counting', e.g:
Example 1:
A postcode district may have 1 farm with 10 dairy cows
and 15 beef cows.
Therefore the number of holdings with dairy cattle in
that district will be 1,
and the number of holdings with beef cattle in that
district will also be 1.
- but the number of holdings in that district with cattle,
either dairy or beef (i.e. 'total cattle'), is also 1
- not 2 - (with a value of 25 cattle).
Example 2:
However if this postcode district has 2 farms, one with
10 dairy cows and the other with 15 beef cows,
then the number of holdings with dairy cattle in that
district will still be 1 and the number of holdings with
beef cattle will also be 1 as before.
- but the number of holdings in that district with cattle,
either dairy or beef (i.e. 'total cattle'), is now 2 - (with
a value of 25 cattle as before).
Because it is not possible to say which is the true situation
for the repopulated datasets, it has been decided to just show
the value for subtotals
Methodology
The hidden data in the 'Nuts' ward-level dataset was repopulated
as follows:
(Data is 'hidden' when it is disclosive or dominant. It
may also be 'hidden' in order to protect other disclosive
or dominant data from being derived, this is known as suppression.)
For each census item, the number of hidden wards counted
(x) and the sum of the hidden data was calculated (y).
The hidden data was then redistributed to all the hidden
wards by averaging (x/y).
[Note: if there is a dominant ward (i.e. with a very
large value) and a disclosive ward (i.e. with a very small
value) in a set, then the disclosive ward may be repopulated
with a value that is much higher than the actual situation.]
The repopulated ward-level dataset is then redistributed to
all the postcode areas as follows:
Each postcode area will consist of a certain percentage
area of one or more wards:
- e.g. Postcode area 'P' will cover 30% of ward 'q', 25%
of ward 'r' and 15% of ward 's'.
- if the number of sheep (note - the repopulated value)
in each ward is: 1200 in ward 'q'; 800 in ward 'r'; 2400 in
ward 's', then the number of sheep assigned to Postcode area
'P' will be (360+200+360), i.e. 920 sheep.
[Note: it may be that there are no sheep in
that postcode area (i.e. the sheep belong to holdings elsewhere
in the ward) or it may be that all the sheep from the
wards belong to holdings in that postcode area. The value
given is simply an average of the repopulated ward-level dataset,
redistributed proportionately over the postcode areas and
as such gives an approximation of the actual situation.]
