Skip to content Skip to footer

Our maps and data

Are the Agricultural Censuses for each territory 100% samples?

In Scotland the annual June census covers the 50,000 agricultural holdings with sampling of smaller holdings. The smallest spatial unit for which data are made available are parishes (subject to disclosure restrictions). Note however that the EDINA agcensus service does not distribute data at the parish level. Since the late 1990s the agricultural census in England and in Wales has been conducted as a sample survey (postal questionnaire or via internet), in which data is only sought from a proportion of holdings in each year. For England a completely random sample of the 200,000 agricultural holdings would be inefficient due to the large number of small holdings thus a stratified random sampling approach was been employed in which holdings are divided into groups on the basis of theoretical labour requirement with higher sampling rates being used in the larger start

Simply stratifying on labour requirements would however lead to low precision for some crops thus Defra employ a separate strata with higher sampling rates for horticultural holdings. The sample size has varied from between 25%-33% since the inception of stratified random sampling in the late 1990s. The imputation of values for the June survey for England is done by the separate ratio approach. In brief, this method ascertains the ratio between the current number of e.g. sheep and the previous year’s number of sheep for the holdings which responded to the current year’s survey. The ratio is then applied to the total number of e.g. sheep from last year to get a new estimate for the number of sheep in England. These calculations are applied on each stratum and on each census item. The estimated totals of which are summed to give a national figure. As of 2004 Defra are no longer distributing the small area data until the next full census in 2010 (as demanded by then current EU regulations). The Statistical Directorate of the Welsh Assembly Government commenced with an agricultural survey (to replace the agricultural census) in 2006. It should be noted that in the case of the Welsh Agricultural Survey whilst the accuracy of the sample survey data and its conversion to grid square estimated values remains consistent with those of earlier census years, the spatial granularity offered by the survey and subsequent data imputation could potentially diminish the geographical variation across the territories.

Some values for census items measured in hectares are greater than the total grid square area, why? There are anomalies in the collection of Agricultural Census data that distort the grid square estimates. For example, a farmer is only required to return one census form, even if s/he owns several holdings, some of which may be in other parishes or even quite remote from the main holding. Certain items, especially rough grazing etc., may be either grossly over- or under-represented in any particular area. For example, data on rough grazing and woodland are not available for many years for Scotland. The anomalies in the way these items are reported at parish level are too great to make conversion to grid square estimates sensible. A farmer can report all his holdings on one census form and these 2 items in particular may include land in other areas well away from the main holding. Other area items, e.g. total area, are similarly problematic. Grid square estimates are derived from parish summaries or Defra estimates for arbitrary geographies, taking account of potential landuse capability at 1km level. In the conversion algorithm the data are distributed to all suitable 1km squares for that item in the area of reporting, but no agricultural data are allocated to unsuitable squares, e.g. urban, inland water, non-agriculatural land. There may be too few suitable landuse squares in a reporting area and this can lead to an area greater than the size of the square.