Defra - Department for Environment, Food and Rural Affairs.

Science Search

Science and Research Projects

Return to Science Search homepage   Return to Project List

Non- Response Study- Farm Business Survey (FBS) - DO0114

The non-response study would firstly involve matching the recruitment information, with the recruitment outcome, for the upcoming 2011/12 campaign to the 2010 June Business Survey (complete census). As the numbers agreeing to take part in each year is small, it would be beneficial if several years’ (2009/10 and 2010/11) information were available for this analysis. This matching exercise would be carried out by Rural Business Research (RBR)
By conducting a non-response study it would enable us to see if amongst those approached by RBR, if there is a difference between those agreeing to take part and those not agreeing (i.e. non-response bias in the current year). In addition, we would be able to see if there is a difference between those in the panel and those not in the panel, other than the difference created by the design and allowed for in the weighting. These differences may result from recent non-response bias, historic non-response bias, or bias in retention. This piece of work would help us to answer the key question ‘How do the farms in the survey differ from those that do not take part’. In addition to analysis of non-respondents, the matching exercise could very usefully meet a further objective the effectiveness of the calibration model can also be investigated as factors in the calibration model should reflect known non-response patterns.

Linking the recruitment information to the June Business Dataset.

As mentioned the matching of the recruitment information to the June Business Dataset would need to be carried out by RBR to overcome confidentiality issues.

For each recruitment round RBR receive a random list of potential recruits which will enable the sample quota’s to be maintained. This recruitment list contains name, address, telephone number, dominant CPH and other information relating to the characteristics of the farm. Before the recruitment list is transmitted to RBR the June business dataset ID is removed and replaced with a random number. For the purposes of the non-response study if the June business dataset ID was kept on the recruitment list, that is provided to RBR, this would provide a direct link to the June business dataset to enable the recruitment outcome for these farms to be recorded on the June dataset. In addition, a look up between the random numbers and business dataset ids could also be provided for the recruitment lists sent for the 09/10 and 10/11 FBS recruitment campaigns and the same process followed to link the recruitment outcome and June dataset (should recruitment information for those years still be available).

The use of the business Id’s instead of the CPH numbers should hopefully overcome the issues that arise when using CPH numbers. A subset of the June Business dataset will be provided to RBR, it will contain just those farms from the recruitment exercise and a set of key variables (e.g. indicative data such as farm type, the variables used in the calibration procedure). Figure 1 and table 1 show how the procedure would work and the codes to be used for recording the recruitment outcome.

Before transmitting the 2010 June business dataset back to Defra, RBR would remove the business ID from the dataset. This anonymised dataset can then be analysed by the statisticians in the FBS team (Katherine Merrett and Neville de Souza).
Time-Scale and Cost
From: 2012

To: 2013

Cost: £7,775
Contractor / Funded Organisations
University of Nottingham