The HGCA Planting Survey is the most comprehensive pre-harvest planting survey in the market arena, whose methodology closely resembles DEFRA's June Census. There are certain statistical features therefore that need to be clarified.
This survey is unlike any other surveys the HGCA routinely performs. It is called a 'ratio-raising' survey, because it uses a ratio (the difference between this and last year's plantings) to calculate the area planted. Put simply, if we ask producers what they grew both this year and last we have a percentage difference on the year. This figure (ratio) is multiplied with last year's final June Census results to give us an estimate of the cropping area. We effectively 'raise' this year's figures from last year's actual results. Each region, commodity, and holding size has an individual ratio, which can be used to raise the figures.
This technique (frequently employed by DEFRA) reduces sampling bias on a holding level basis. Where insufficient responses to the survey have been received, for any particular holding size range/commodity/region, then the results are combined with those of the next largest holding size/commodity/region. An assumption is made that cropping patterns will be similar across similar holding sizes for any one particular commodity and region. The original 'ratio' is thus reduced, more in line with other holding sizes. This technique is important for small surveys using 'ratio-raising' methods as extremes of responses can be minimised. This technique is used most frequently in areas with few respondents and with less popular crops, in practice usually with linseed, oats, peas and beans. This method can be employed because only the total area for each holding range is used, not individual data and the ratio (% difference between 2006 plantings and 2007) is the key factor not the absolute value.
DEFRA, SEERAD and the National Assembly for Wales provided the HGCA with June Census 2006 results stratified into the various holding sizes, and by commodity and region. For England we were supplied with both main and minor holding data from the 2006 June Census, whereas SEERAD and NAW provided main holding data only (>5ha). The survey results have thus been adjusted to take into account these smallholdings for Scotland and Wales.
For 2007 we have calculated confidence intervals for each regional estimate. The confidence interval, used alongside the raised estimate provides a range of values, either side of the estimate, within which the actual area sown lies within 95% of the time i.e. if the survey was carried out 100 times, the true area planted would lie within the published range at least 95 times out of 100. The confidence intervals for the 2007 HGCA Planting Survey are available on request. | 
